This PDF file contains the front matter associated with SPIE Proceedings Volume 12615, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

The traditional color transfer method does not consider the influence of the light source of the reference image well, which is easy to cause problems such as poor visual effect and fuzzy texture details in the resulting image. To solve these problems, a color transfer method combining cartoon decomposition and illuminant estimation is proposed in this paper. Firstly, considering that the resulting image is easily affected by the brightness of the reference image illuminant, an efficient illuminant estimation for color constancy using grey pixels is introduced to alleviate the influence of the reference image on the resulting image. Secondly, to better preserve the texture details of the source image, the cartoon decomposition is used to decompose the source and reference images into a texture part containing the main details and a cartoon part containing the main contour structure. Finally, a full color gamut mapping technique is executed on the cartoon portion of the source and reference images to complete the color transfer. The experimental results show that the proposed method can not only effectively maintain the texture details of the source image but also achieve good results in terms of objective evaluation indexes, compared with the traditional color migration method, while achieving better visual effects.

In the indoor distributed antenna system, the uplink and downlink of the digital remote unit combine the multi-carrier and output, and the signal power increases after the combination. When the output power is greater than the maximum limit of the system, it may cause damage to the analog devices in the system and output wrong signals. Since the uplink and downlink are respectively oriented to the digital expansion unit and the RF analog unit, power control needs to be performed for different links respectively. Therefore, an FPGA-based uplink and downlink automatic power control technical scheme is proposed. By counting the digital signal power at different times, a short-time automatic power control module suitable for uplink and a short-time automatic power control module suitable for downlink are designed respectively. Long-time automatic power control module, which effectively solves the problem that the signal power exceeds the system threshold after combining, and ensures that the power of the output signal is basically stable. At the same time, the DSP resources are time-division multiplexed according to the multiple relationship between the sampling rate and the clock frequency, which optimizes the consumption of DSP resources. After simulation and actual test on the board, the results show that the control accuracy of the uplink output power is ≤±0.5dB, and the maximum attenuation of 8dB can be achieved. The downlink output power control accuracy is ≤±1dB, the maximum attenuation is 22dB, and the DSP resource consumption is reduced by 50% after time division multiplexing.

In this paper, we propose an image encryption algorithm via deep compressed sensing. First, a convolutional network is used to sample the image and produce the results of the sampling. Second, the matrix STP calculation is conducted on the sampled results utilizing a logistic chaos system, SHA-256, and a chaotic sequence matrix produced from a normal image in order to produce the final encrypted image. During the decryption procedure, the decryption measurements are obtained by performing the STP inverse operation on the encrypted image and the chaotic sequence matrix, and the measurements are then transmitted to the reconstructed image network to get the final decrypted image. Experiments demonstrate that the method provides a high level of image reconstruction quality, security, and runtime simplicity.

Ground penetrating radar (GPR) signals are non-stationary signals that are easily affected by interference and noise, resulting in reduced exploration capability. This paper proposes a multi-task encoder network model (ADAE) based on deep neural networks, which integrates channel and spatial attention mechanisms. According to ground penetrating radar noise characteristics, the channel and spatial attention mechanism are added to the encoding and decoding stage and the intermediate separation mapping stage, respectively. The composite attention mechanism of channel and space is used to optimize the feature extraction method and improve the characterization ability of features to suppress the influence of noise and irrelevant components from different angles. Experimental results show that the average boost signal to noise ratio of GPR data processed by multi-task encoder is 10.449dB. Furthermore, the results show that the ADAE multi-task encoder has apparent advantages in suppressing the noise of ground-penetrating radar and maintaining the signal amplitude effectively.

The multi-classification of finger movements is a difficult challenge at present. In this paper, a multi-task Siamese network is proposed. Based on the basic Siamese network architecture, this study designs the position relationships between sample pairs and constructs single-sample subtasks for the left and right input samples. The position relationships between samples not only additionally supplement the spatial factors between electroencephalogram (EEG) signals that facilitate the generalization performance of position network learning, but also turn the EEG signals set structured so that the multi-task Siamese network is used to effectively learn multiple relationships between samples. During the model testing phase, a balanced matching algorithm of single test sample VS multi-training sample is used to improve the robustness of the multi-task Siamese network model. The experimental results on the 5F motor imagery data set from The Largest SCP of Motor-Imagery, an EEG data signal repository, and data set4 of the 4^th brain-computer interface (BCI) competition show that compared with the single-subtask Siamese network, EEGNet, ResNet-50 and the public experimental results of the original dataset, the multi-task Siamese network in this paper can obtain better multiclass motor imagery (MI) classification accuracy and can better realize the multi-classification of MI-EEG signals. This study could be helpful to promote the development of finger-based MI-BCI system.

In this paper, a stable optical frequency comb generation scheme based on Mach-Zehnder modulator (MZM) is designed. The key to this scheme is to determine the type of modulator, and then through the analysis of modulation characteristics and theoretical calculation, the driving condition of the flattest optical frequency comb can be achieved. This scheme finally generates 35 stable optical frequency lines with flatness less than 1.1 dB and tone to noise ratio (TNR) more than 37 dB. Compared with other research on the modulation of optical cascade number of frequency lines and the flatness have an excellent improvement in performance. Moreover, the experimental scheme has the advantages of simple structure, low hardware and driving voltage requirements, and can be referenced by more researches and applications. At the same time, the designed 12.5 GHz frequency interval provides a stable frequency source for wavelength division multiplexing. At the same time, the frequency lines generated by the same light source have good correlation, which provides a basis for coherent communication research for generating stable microwave and terahertz carrier. Therefore, the stable multi-light source generated by the optical frequency comb provides more possibilities for future terahertz generation and communication.

Radar signal sorting is a key technology in electronic warfare. The quality of signal sorting directly determines the outcome of electronic warfare. The accuracy and real-time performance of signal sorting are two important indicators to evaluate the quality of signal sorting. With the increasingly complex electronic countermeasure environment, there are many radar radiation sources in the battlefield, and the signal aliasing is serious. At the same time, there may be pulse loss and interference pulse in the electromagnetic environment. In such a complex electromagnetic environment, in order to improve the accuracy and real-time of radar signal sorting and master the battlefield initiative, this paper proposes a method of radar signal batch processing sorting based on sliding window. Firstly, the radar signal sorting transient library and the sorted radar signal library are set up, and the transient library is used to store radar signals in the sliding window. Move the sliding window continuously from the starting point. When the number of signals in the transient library is higher than the specified threshold, sort the radar signals in the transient library, and then store the sorted radar signals in the sorted radar signal library and delete them from the transient library. When the sliding window moves to the end point, the signals in the sorted radar signal library are merged. Simulation results show that compared with traditional methods, the proposed method has higher sorting accuracy and better real-time performance when there are more radar pulses and pulse loss.

In recent years, many decolorization methods have been proposed in various fields. In the traditional decolorization methods, people often adopt strict color order to reduce the dimension of the original color image, but the gray image resulting from this method often has low contrast. However, there are still many obvious problems such as low contrast, lack of detail and lack of layering in the current proposed de-color methods. In this paper, by studying the projection amplification of global mapping, a bimodal mapping method is designed by using gradient correlation as the standard. Under the premise of enhancing the gray contrast of the color image, the detail features of the original color image can be better preserved. At the same time, the new method takes into account the computational load of the algorithm, reduces the complexity and time cost of the algorithm, and improves the real-time performance.

In order to solve the problems of identifying the remote ground vibration signals with complex signal components and low signal-to-noise ratio, this paper proposes a denoising method combining wavelet transform and empirical mode decomposition: Wavelet EMD Denoising. First, wavelet transform is applied to the noisy signal for initial noise reduction, then empirical mode decomposition is applied to the signal after initial noise reduction, then wavelet decomposition is applied to the sub signal for noise reduction again, and then the sub signal is filtered and synthesized to obtain a signal with high signal-to-noise ratio for subsequent processing and analysis. In this paper, EMD synthesis and wavelet EMD noise reduction processing are respectively carried out for the collected long-distance personnel vibration signal and vehicle vibration signal, and the time-frequency domain of the results of both are analyzed. It is proved that the wavelet EMD noise reduction proposed in this paper can effectively reduce the noise of this type of signal compared with EMD synthesis, and its frequency domain energy is more concentrated: the energy proportion of the frequency band where the effective signal of vehicle signal is located is increased by 29%, The proportion of energy in the frequency band where the effective signal of personnel signal is located is increased by 51%. The noise reduction effect is better than the traditional EMD synthesis algorithm.

In this paper, a transform domain information hiding algorithm with visual security based on compressed sensing (CS) is proposed. To increase the security, an improved coupled map lattices is used for keystream and measurement matrix generation. To save the storage space, the semi-tensor product is applied to the measurement matrix. Using the image transform domain property, the compressed and encrypted image to be hidden is embedded into the high-frequency part of the carrier image through mathematical operations, and then the carrier image is inverted to finally form a visually secure image. Simulation results and performance analysis show that the designed information hiding scheme has better hiding effect and high security.

In view of the fact that the current radar signal modulation recognition method based on depth learning needs enough samples to achieve good recognition effect, this paper designs a depth metric learning recognition method based on Cosine Softmax (CS Softmax), which has strong intra class aggregation and inter class separation, and can recognize unknown signals. First, the Wigner Ville distribution (WVD), Pseudo Wigner Ville distribution (PWVD) and Smooth Pseudo Wigner Ville distribution (SPWVD) of known signals are taken as the input, the vectors obtained from the network output are learned into the embedded space to identify radar signals, and several unknown signals are combined with the initial training set in the verification phase. The simulation results show that under the signal to noise ratio of 10dB, the recognition accuracy of the known signal of this method can reach about 94%, can distinguish the unknown signal from the known signal, and can distinguish the unknown signal separately, with an accuracy of about 85%.

In the complex remote sensing image detection, there are still many challenges in maritime ship detection. This paper combines the latest swin-transformer model with satellite remote sensing big data, uses the SSDD dataset, researches for maritime ship target detection, realizes the target detection support for SSDD dataset, at the same time, this paper proposes an improved version of augmix method Multiple-augmix, based on the principle of data enhancement for SAR image dataset, is proposed to achieve target detection and data enhancement for SSDD dataset and make comparison experiments. The detection accuracy of swin-transformer can reach 86.5% for the original SSDD dataset, and 96.2% for the SSDD dataset enhanced with Multiple-augmix method, which is 10 percentage points higher than that before enhancement, and at the same time, by conducting experiments with Fasterrcnn, RetinaNet and Focos algorithms and making comparisons, the experiments prove that This framework can achieve high detection accuracy in remote sensing image detection, and the Multiple-augmix method proposed in this paper is very effective in improving the detection accuracy of SSDD dataset. In addition, we verify the effectiveness, stability and accuracy of the above algorithm for SAR image target detection in complex remote sensing scenarios.

This paper mainly analyzes the detection algorithm of UAV signal, compares and analyzes matched filter detection, energy detection and static period. Advantages, disadvantages and realizability of feature detection methods. Unmanned aerial vehicle (UAV) brings us convenience in life and enhanced national strength, but it also brings many potential safety hazards. In order to ensure that the territorial sovereignty of the country is not affected, the residents of cities are not disturbed, and the orderly development of the society is ensured, whether it is a military stronghold, some specific residential areas or public places, it is necessary to set up a UAV defense system to effectively monitor and intercept some UAVs that pose security threats, and effectively strike at serious threats. To safeguard the security of the country and society, this paper analyzes and studies the UAV defense system, and discusses the components and functions of the defense system as a whole, for reference only.

To address the problem of insufficient detail retention in traditional color transfer, a color migration combining intrinsic decomposition and histogram matching is proposed in the paper. Firstly, the source image and the reference image are intrinsically decomposed to obtain their respective reflectance and irradiance maps, and the new reflectance maps after color transfer are obtained by histogram matching using the decomposed reflectance maps of the source and reference images. Secondly, in order to better preserve the texture details of the resultant image, the new reflectance map is intrinsically synthesized with the irradiance map of the source image enhanced by the global change regulariser to obtain the resultant map. The experimental results show that the new method proposed in the paper better preserves the texture structure of the result image, and the resulting result image has better visual effects and performs better in both subjective and objective evaluations.

This paper introduces an optimized de-noising method based on the wavelet transform hard and soft threshold de-noising method, which is used to remove noise from electrocardiogram signals to achieve clinical high-precision diagnosis. An appropriate threshold processing method is required in the process of denoising and restoring the electrocardiogram signals carrying a certain degree of noise. However, the traditional hard and soft thresholds have serious problems such as distortion and oscillation. By using the optimized new threshold algorithm, the problem of hard and soft threshold processing can be improved to a certain extent under the condition of higher degree of signal restoration and better denoising effect. The experimental results show that with the help of the optimized algorithm in Python, the denoising and restoration of noisy electrocardiogram signals have indeed achieved better results than the traditional methods.

Signal de-noising process is the process of signal characteristic components extraction. In view of the singular value decomposition de-noising method is difficult to separate under low SNR, through observing the singular value distribution under different decomposition order, select the appropriate number of decomposition, and combining the wavelet packet de-noising ability and good frequency resolution, put forward a kind of effective signal de-noising method. Firstly, based on different SNR to choose the appropriate number of singular value decomposition, select bigger singular value reconstructing the signal, preliminary to remove noise, and then make three layer wavelet packet decomposition, compare the energy distribution of all nodes, and make selection of energy concentration of several nodes, reconstruct the signal. Simulation results verify the validity of this method.

Aiming at the problem of accurate estimation of direction-of-arrival(DOA) and polarization parameters of coherent sources by anti-radiation seeker under polarization sensitive circular array, this paper proposes a DOA and polarization parameter estimation method based on peak comparison(PC) and oblique projection(OP) transformation. At first, the polarization forward smoothing(PFS) method is used to realize the decoherence of array received signal under uniform circular array(UCA). Then, based on the PC strategy, the spectrum peak of the two-dimensional spatial spectrum is searched to achieve the direction finding of the target. Finally, using the estimated DOA information of the signal source, multiple coherent signals are decomposed into single signal units based on the OP transformation, and the polarization parameters of each signal unit are estimated. On the one hand, the proposed method speeds up the two-dimensional spectral peak search, on the other hand, it solves the problem that the PFS algorithm cannot estimate the polarization parameters, realizes the automatic pairing of DOA and polarization parameters, and has high estimation accuracy. The effectiveness of the proposed method is verified by simulation.

In this study, the concept, main technology, and application scenarios of cluster analysis are briefly introduced for the Gong-che Notation sheet music, which was a widely used in ancient China. Various classification methods using different clustering algorithms are analyzed; accordingly, some application scenarios are discussed. This study focuses on extracting the columns from Chinese Gong-che Notation sheet music score images using systematic tree clustering. The experimental results show that different column’s number of Gong-che Notation sheet music score correspond to different extraction accuracy.

There are inevitable problems in the imaging mechanism of visible and infrared images, which bring more difficulties to the detection task. Firstly, this paper introduces the principle of SSD one-stage detection model. Aiming at the characteristics and problems of the model, residual network is used to improve it, and the prediction feature layer is reselected and sorted. Based on the photoelectric equipment, the video data of marine ship targets are collected and the data set is constructed. Through the comparison of the training results of various models, it is found that the detection accuracy of small-sized targets is low. In order to solve this problem, First, reduce the size of the anchor , and the feature extraction network is modified. The experimental results show that the average detection accuracy of the improved model increases by 7%, and the single category accuracy of small targets increases to 87%.

In many complex wireless communication scenarios, frequency hopping technology has attracted widespread attention due to its strong anti-fading and anti-interference characteristics. In this paper, we propose an improved support vector machine (SVM) classifier based on the decision-level fusion (DLF) model, referred to as DLF-SVM, for automatic modulation classification (AMC) of frequency hopping signals in multipath channels. To estimate parameters and get the sliced samples, the frequency hopping pattern is first obtained by performing the time-frequency analysis on the received signal. Then, for samples after being preprocessed, we extract the correlation coefficients of the cyclic spectral section as categorical features. Finally, focusing on the idea of the DLF model, an improved SVM classifier based on the prior probability weight matrix is constructed to realize the AMC of frequency hopping signals. Simulation results show that our proposed algorithm can significantly improve the classification accuracy of binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), and minimum shift keying (MSK) signals under low signal-to-noise ratio (SNR) and multipath channels.

The calibration and compensation of systematic biases of air-to-air 7radars based on ADS-B messages is an easy-toimplement radar calibration technique. However, the calibration accuracy is limited by the lack of accurate timestamp information in ADS-B messages. To address this problem, a technique to perform accurate time-delay search based on the maximum likelihood objective function is proposed. Simulations demonstrate that this time-delay search and estimation algorithm can achieve the function of time-delay estimation with high accuracy and stable performance and better than sample selection calibration method, thereby meeting the requirements of search and surveillance radar calibration.

Traditional compressed sensing image reconstruction algorithms have poor image reconstruction quality at low sampling rates and is time-consuming due to the iterative optimization process. Deep learning-based compressed sensing reconstruction algorithms can significantly reduce time complexity and improve reconstruction quality compared to traditional iterative reconstruction algorithms. In this paper, we propose a new residual dense reconstruction network (RDRNet) for compressed sensing image reconstruction. RDRNet consists of a sampling network and a reconstruction network. The reconstruction network uses multiple residual dense blocks to extract image information. The residual dense blocks can be used to extract rich detail features by fusing local features to obtain the final reconstructed image. Through experimental validation on different datasets, the network has better image reconstruction quality and comparable runtime complexity compared to the existing DCS model.

In the study of heterogeneous cellular networks (HetNets) using stochastic geometry theory, the location of base stations is usually regarded as the point of homogeneous Poisson point process (PPP) distribution. The PPP model, however, is unable to adequately depict how base stations are distributed in hot regions. To solve this problem, the Matern cluster process is used for modeling, and the downlink coverage probability is used. The simulation proves that the base station deployment based on the Poisson cluster has a higher coverage probability.

Magnetic Resonance Imaging (MRI) acquisition is a long time process that leads to patient discomfort and motion artifacts. Convolution network has been implemented in MRI image reconstruction to speed up MRI. Convolution network utilizes spacial local convolution kernel successfully, but non-local information with self-similarity is ignored. In order to effectively explore non-local information in MRI reconstruction, we propose graph convolutional Unet (GCN-Unet) for MRI image reconstruction. In the GCN-Unet, non-local information is represented by a patch graph extracted from MRI images, in which nodes are consist of image patches, and edges are self-similarities between nodes, reflecting non-local structure similarity intra an MRI image. We utilize graph convolution with a Unet framework to effectively represent features of non-local information with self-similarities and aggregate the extracted non-local features to reconstruct MRI image. Experiments demonstrate the effectiveness of the GCN-Unet, the Peak Signal-to-Noise Ratio (PSNR) and Structure Similarty Index Measure (SSIM) of reconstructions are improved.

The high cycle fatigue of aero-engine blades is mainly vibration bending fatigue. To study the high-frequency vibration characteristics of the blade under actual working conditions, the high-frequency vibration characteristics test of the blade under ultrasonic excitation was carried out by using the horn-blade ultrasonic excitation laser vibrometry system and analyzing the blade high-frequency vibration characteristics. The results show that the ultrasonic generator can produce stable 20k Hz ultrasonic vibration, and the laser vibration measurement system can accurately measure the high-frequency vibration. Due to the structure shape of the complex blade, there is energy dissipation, which results in the blade amplitude being much smaller than the excitation source amplitude. Under the same excitation position and ultrasonic excitation conditions, the smaller the blade mass can cause the larger the amplitude, the higher the number of high harmonics, the lower the fundamental frequency, and the more complex the vibration waveform. The research provides a test method for the analysis of the blade high-frequency vibration characteristics and has a certain guiding significance for the research on the blade ultra-high cycle fatigue.

Vacuum optical trap is a technology that captures and manipulates particles in a near-vacuum environment by means of laser mechanical effects. It is an ideal platform for precision measurement with high quality factor. Due to its low environmental damping, vacuum optical traps require feedback cooling to reduce the equivalent temperature of particle centroid motion to prevent particle escape. Parametric feedback cooling is a mainstream solution to achieve stable trapping of particles in optical traps. Monte Carlo method and finite difference method were used to simulate the displacement fluctuation of the microsphere without feedback, and the FPGA circuit was used to output control signals with different waveforms to test the feedback cooling effect. Under the premise of reaching mK cooling level, square wave control signals have advantages in fast feedback cooling, and sine wave control signals have advantages in maintaining phase matching.

Aiming at the problem that it is difficult to identify the complex radar signal under the condition of low signal-to-noise ratio using the traditional pulse-to-pulse parameters, a method using deep learning model to assist training and identify radar radiation sources is proposed. Firstly, the time-frequency image of radar signal is generated as training set 1 by the method of time-frequency analysis. Then, using the sample learning ability of the deep convolutional generative adversarial network, the time-frequency images are generated as training set 2 on the basis of training set 1. Compared with 1, training set 2 has the effect of denoising and data enhancement. Finally, the training set 2 is used to assist the training of the visual geometry group on the training set 1 to identify the radar radiation source. Simulation experiments are carried out on five common radar signals, and the experimental results verify the effectiveness of the method.

With the vigorous promotion of railway infrastructure construction in China, higher requirements have been put forward for railway running safety. Railway signal monitoring equipment, as the key equipment to ensure driving safety, find and solve the fault, has a broad development prospect in the process of railway system modernization[1]. By analyzing the practical significance of railway signal training system and referring to the technical conditions of railway signal centralized monitoring system, the DC switch acquisition module of railway signal training system is developed and designed based on the microcontroller of Cortex-M4 kernel. The acquisition board has unique advantages in power consumption, acquisition speed and scalability, high integration, good reliability, its accuracy and stability meet the requirements of railway signal training, has a broad space for development, and is of great significance for the pre-job training of railway related majors and the improvement of teaching quality.

To address the problems of difficult data synchronization and insufficient protection capability of transmission line longitudinal protection, this paper proposes a digital twin solution for transmission line longitudinal differential protection scenarios. A 5G communication system is also introduced to collect and interact with the electrical quantities at both ends of the transmission line. Through the digital twin modeling of the entire transmission line, various electrical quantities of the simulated distribution network transmission line are obtained. And the obtained modeled values are used for current prediction by BP neural network. Experiments demonstrate that the method provides multiple insurances for longitudinal differential protection by analyzing the data generated by the digital twin for prediction.

This paper briefly introduces the Internet of Vehicles, 5G communication and digital twins. Aiming at the driverless tracked vehicle group developed in the earlier stage, combined with vehicle network and 5G communication, a 5G communication based vehicle networking framework for driverless tracked vehicle group is proposed. On the basis of 5G communication based unmanned tracked vehicle group Internet of Vehicles framework, combined with digital twins, the digital twin framework of unmanned tracked vehicle group is proposed at two levels, and comparative analysis experiments are carried out.

A signal is a physical quantity that represents information and is also a carrier of information. As signals are indispensable in every aspect of life, signal processing has always been a hot topic of research. With the continuous improvement of signal processing technology, it is used in a wide range of fields, such as digital filters, speech signal processing, image signal processing, biomedical aspects, etc. This paper first introduces the types of signals and basic concepts and some modern signal processing techniques which are commonly appeared. Afterwards, the principles and characteristics of the Fourier Transform, the Wavelet Transform and the Hilbert-Huang Transform are described. Furthermore, the advantages and disadvantages of the three signal processing methods and their respective fields of application are derived by means of comparative analysis. At the same time, the connections and unique features that exist between them are explained. Finally, the paper gives a conclusion and an outlook.

Because the time of traffic signal lights when vehicles are released is the same, it is easy to cause vehicle accumulation, traffic congestion and other problems. Therefore, a coordinated control method of intelligent traffic lights based on YOLOv3 and DQN is proposed. Obtain the basic parameters of intelligent traffic, set the green signal ratio of each phase in a balanced manner, define the reward of coordinated control as the change in cumulative waiting time between two adjacent cycles, set the intersection control domain decision sequence based on YOLOv3 and DQN, calculate the intersection cycle correlation, and design the signal coordination control method. Experimental results: The average queue waiting time of the intelligent traffic signal coordination control method in this paper is 648.37 s, which is 64.22 s and 41.05 s lower than the other two intelligent traffic signal coordination control methods respectively, indicating that the designed intelligent traffic signal coordination control method is better after combining YOLOv3 and DQN algorithms.

Shanghai Rail Transit Line 10 is the first large capacity line in China that adopts the unattended train operation technology. Limited by the land scale of the station, Jilong Road Station that is the terminal station needs to provide the maximum turn-back capacity designed under the existed civil engineering and operation design conditions. The thirdparty simulation analysis of the operation capacity of the signal system design scheme suggested that the turn-back capacity of one of turn-back modes is low, only 20 pairs / h. Therefore, without changing the station wiring, station location, dwell time and other restrictions, with the goal of tapping the operation potential of electromechanical equipment, six strategies for turn-back capacity promotion based on safety braking model optimization are proposed, and the applicability comparative analysis is completed. Finally, the project adopts the strategy of reducing ATO (i.e., automatic train operation) stop braking rate to shorten the protection distance of JJL2 platform from 35m to 18m. The simulation analysis shows that the turn-back capacity is increased to 33 pairs / h. Compared with the original design, the turn-back capacity is increased by 65%. It meets the construction needs of the project, avoids ten million RMB or even billions of RMB in transformation investment due to insufficient turn-back capacity, and provides important engineering experience for the follow-up work of operation capacity promotion for existed lines.

In this paper, the principle of speech aerodynamics is used, and the acoustic spectrogram and pronunciation airflow waveform are one-to-one corresponded by means of signal processing. The results show that from the airflow waveform, it can be clearly seen that the cooperative pronunciation influence mode between the plosives and vowels /a/ is cis assimilation. The plosives will not only affect the size of vowel airflow and airflow velocity, but also affect the trend of vowel formant. When the aspirated stop is pronounced, the greater the pressure under the glottis results in a longer duration and a higher average airflow velocity, which is about twice as long as the unaspirated stop and more than four times as fast as the unaspirated stop; The stability of the duration of the air flow of the aspirated stop is worse as the voice position is more and more backward, and the stability of the average air flow velocity of the aspirated stop is worse than that of the unaspirated stop.

Conformal array radar system is an important trend for the development of airborne radar system, since the geometric structure of conformal array is consistent with the shape of carrier aircraft, it can expand the antenna aperture and improve the detection performance of the system. In addition, compared with the plane array installed on the platform, it does not change the aerodynamic performance of the aircraft and can effectively improve the flight time and range. This paper analyzes the advantages and disadvantages of airborne conformal antenna array, expounds the development trend of airborne conformal array radar system from the perspective of technical development, and discusses the key technologies and solutions of airborne conformal array radar system associated with the engineering practice.

UAV-assisted wireless communication has the advantages of flexible deployment and wide coverage. However, it still has some disadvantages such as high path loss, low throughput and low transmission capability. The beamforming technology is introduced into UAV-assisted wireless communication, which can effectively compensate the path loss and alleviate the interference. Therefore, a UAV-assisted wireless communication system model is constructed in this paper, which is equipped with uniform planar array on the ground base station and UAV to achieve beamforming, so as to maximize the capacity of target nodes. In order to solve the non-convex problem with high-dimensional and highly-coupled optimization variables, firstly, the block coordinate descent method is used to transform the original problem into two sub-problems: position optimization of relay UAV and beamforming vector optimization. And then the continuous convex approximation method is used to transform two sub-problems into convex optimization problems for solving. The simulation results show that the proposed method can effectively improve the capacity of the target node.

Considering that traditional statistical learning methods are vulnerable to illumination interference and difficult to obtain accurate human eye centers when locating human eyes, this paper proposed a human eye detection and location algorithm based on SIFT registration and a modified CNN model. Firstly, face images were registered using SIFT. Secondly, the processed images were transferred into the modified CNN model to extract left and right eye anchor on the human faces and determine whether there were human eyes in the anchor. Finally, the size of the left and right eye anchor was calculated to obtain the accurate human eye center in the anchor. Tests based on JAFFE face database and CMU Multi-PIE face database showed that the proposed algorithm had a strong anti-interference ability to complex illumination, irregular face images and facial expression transformation, and could obtain accurate human eye center points.

A single-inductor triple-source energy harvesting system that can harvest thermoelectric, photovoltaic and vibration energy was presented. With an integrated self-starting circuit, the system can self-supply while don’t require an external battery. The hybrid modulation of pulse-frequency modulation (PFM) and pulse-width modulation (PWM) is performed to realize adaptive asynchronous control for input and output, achieving higher system stability and better energy aggregation. An adaptive maximum power point tracking (MPPT) was employed to generate inductor charging control signals with different frequencies and duty cycles according to different input sources, enabling the inductor to operate in discontinuous mode (DCM) under different input conditions. The energy harvesting system was designed in a 0.18-μm CMOS process. The simulation results show that the system can realize low-voltage self-startup as low as 200 mV and can provide a stable output of 1.4 V for the load and a 1 V harvested voltage for self-supply. The conversion efficiency of the system is 76.43%@Rload = 10 kΩ, and the average power consumption in sleep mode is 2.38 μW.

In this paper, the research on the dual-station direction finding and cross positioning technology of photoelectric imaging equipment based on high-precision Beidou is carried out. The northeast sky coordinate system is the positioning coordinate system. Calculate according to the location information of the site and the azimuth and elevation angles of the base station and the target, so as to obtain the location information of the target. At the same time, the relationship between the positioning error and the positioning accuracy of the base station, the angle measurement accuracy of the optoelectronic equipment, and the deployment of the actual engineering application are analyzed by simulation experiments. In this paper, the validity and practicability of the method are verified by conducting field tests on the basis of establishing mathematical model analysis.

With an increasing demand of spectrum resources, multiple-input-multiple-output (MIMO) has been widely used in various communication architectures. As the frequency band goes up, line-of-sight (LoS) channel becomes the most common scenario because of the obvious straight beam path. However, the undesirable channel capacity turns into a bottleneck in LoS-MIMO system. Due to the weak correlation of vortex waves, the negative effect of low system capacity can be improved to some extent via a special OAM antenna. Compared with traditional MIMO, our simulation results indicate that channel capacities are increased by about 60-180 percent under OAM based LoS-MIMO system. Further, the wireless systems are expected to acquire a higher capacity gain through antennas number addition, spacing expansion, multi-mode gathered, etc.

Metal detectors are widely used in military, police and civilian fields, especially it can detect both magnetic and non-magnetic metals have a broad market prospect. Aiming at the problems that the performance of electromagnetic equilibrium detectors is affected by electromagnetic interference of environment and the closed-loop adaptive processing time is too long. A rapid suppression technique for Electromagnetic Interference of electromagnetic detectors is studied in this paper, which is a combination of closed-loop adaptive processing and open-loop adaptive processing. This method can greatly shorten electromagnetic interference of environment suppression processing time (such as 5 second to 1 second). It can meet the requirement of rapid data output rate of the detector. The out filed test’s results show that this method greatly reduces the noise base level of the detector by more than 12dB, and the open-loop adaptive is completed within 1 second, which improves the detection and response ability of the electromagnetic detector effectively.

With the automatic verification of intelligent power meters, manpower is saved and the verification efficiency is improved. However, the large-scale automatic verification method can not meet the parameter ratio specified in the verification regulation because the verification environment is difficult to control. The environmental factors in the verification are difficult to be accurately controlled, which leads to inaccurate verification results. There are a large number of watt-hour meters in the process of centralized verification. If the verification results are not accurate, it will lead to unnecessary economic losses. In this paper, a T distribution test model is established to consider the influence of the heat of the external shunt and the internal power chip on the temperature of the metering chip in different environments. According to the error fitting curve corresponding to the chip, the measurement error of the DC energy meter in different environments is obtained. The formula is imported into MATLAB for simulation analysis to study the factors affecting the accuracy of the metering algorithm. The error formula and influencing factors are derived when the effective value method is adopted in fast charging mode, and the simulation analysis is carried out.

The passive localization of the underwater target is conducive to accurately track the underwater target and obtain their radiation characteristics. While, in practical application, the algorithm is affected by many uncertain factors. Such as the target strength, complex shallow water environment, input parameters and so on. So, the robustness of ultra-short-range localization performance is relatively poor. The passive location algorithm of magnetic field based on field modeling and optimized search method is not affected by the adverse factors such as shallow water environment and acoustic array. It can achieve high-precision location performance. The location accuracy is about 0.1 time of the ship’s length. However, due to the influence of the fast attenuation of magnetic field, this algorithm can only realize the precise positioning of ultra-short-range and high stability according to the magnitude of magnetic field. This paper attempts to take advantage of the focused beamforming location algorithm and magnetic field high robustness positioning algorithm for the first time. And the adaptive filtering algorithm is adopted to effectively process the passive localization of acoustic and magnetic field results, so as to overcome the adverse effect of the conical ambiguity surface of scalar acoustic array on focused beamforming acoustic image passive location algorithm, and realize high precision and high stability passive localization results of the ship in near-field online.

In mobile underwater acoustic (UWA) communication, serious Doppler effects will deteriorate the performance of UWA communication system. The traditional narrowband receiver based on phase-locked loop (PLL) technology cannot handle the received signal experienced wideband Doppler effects. In this paper, we propose a new receiver structure to eliminate the wideband Doppler effect and enhance the SER performance with a receiver array. Firstly, we utilize an adaptive pre-combiner to reduce the number of receiver channels, which leads to a lower complexity receiver. Secondly, a variable forgetting factor RLS (VFF-RLS) algorithm instead of the conventional RLS is used to estimate the time-varying channel impulse response (CIR). Experimental results show that improved performance can be achieved by proposed receiver with the VFF-RLS algorithm compared to that of receiver with the conventional RLS algorithm.

With the development of various positioning technologies, new requirements have been put forward for indoor mobile terminal positioning. How to use more superior algorithms to improve the accuracy of indoor positioning, especially the application of WIFI based indoor mobile terminal positioning technology, is very beneficial to improve indoor mobile terminal positioning, and has become the focus of people's attention and the key of mobile terminal positioning technology. By analyzing the overall architecture, server function design and mobile terminal function design of indoor mobile terminal positioning system, this paper explores the key technologies of WIFI based mobile terminal indoor positioning system, analyzes the advantages of KNN algorithm and WKNN algorithm of WIFI based indoor positioning system, and finds that WKNN algorithm is obviously superior to KNN algorithm. It has a wide range of application value in the design of WIFI - based mobile terminal indoor positioning system.

Antenna cross-polarization isolation is one of the important specifications of radar systems using the phase-comparison angular measurement algorithm. In this paper, the effect of antenna cross-polarization isolation on the angular measurement error of this phase-comparison algorithm is modeled and analyzed. First, a mathematical model of the phase-comparison angle measurement algorithm with antenna cross-polarization components is developed. Subsequently, the detailed effects of antenna cross-polarization isolation on the angular measurement accuracy are given in the context of system engineering experience, and then quantitative conclusions are given. The proposed analytical model and quantitative results can be used in the measurement accuracy analysis of the phase-comparison monopulse angle measurement algorithm for radar systems. And these results can be also used as a criterion in specifying the antenna cross-polarization isolation in a radar system design.

For UAV swarms, accurate cooperative localization capability is essential for many applications, especially in GNSSdenied environments where absolute localization information is jammed. The accuracy of cooperative localization is related to the power of the signal transmitted by UAVs. Therefore, for the lifetime and localization accuracy of each UAV, power allocation strategies are crucial issues. The power allocation scheme is looked at in this paper to obtain the optimal resource allocation for UAV swarms. First, in both non-jammed and jammed localization scenarios, the Cramer- Rao lower bound (CRLB) is used as the evaluation index for cooperative localization accuracy. Then, according to the characteristics of the power allocation issue, it is converted into a semidefinite programming (SDP) problem with the limitation of minimum localization accuracy, i.e., the maximum CRLB. The power allocation issues in either scenario can be resolved by SDP. The outcomes of the numerical simulation demonstrate that the proposed optimal scheme can reduce resource consumption while still achieving the needed localization accuracy.

Fault diagnosis, location and maintenance of industrial networks are extremely important components of network management, which can effectively improve network availability, shorten network failure time, help improve network performance, satisfy network service quality and ensure network security. This paper proposes a communication fault detection method based on the analysis of communication protocol message flow for POWERLINK, an industrial Ethernet protocol widely used in the field of industrial communication at present. By monitoring network status and building fault judgment logic, it can analyze the rationality of message sequence and protocol consistency in the message flow so as to effectively detect the dynamic upper layer protocol fault in the network and determine the location of the fault. In order to verify the effectiveness of this method, we designed a POWERLINK network communication fault diagnosis system, which can correctly identify and analyze the wrong data frame sequence in the data message flow and finally achieve the diagnosis and location of POWERLINK network communication fault.

Mesoscale ocean front is a special category of oceanographic phenomena, and has complex effects on the propagation of underwater sound, and meanwhile the complex topographic variations of the seafloor also have an impact on the underwater sound propagation characteristics. In this paper, we simulate the communication located at between the distant and the near, and use the Bellhop numerical model of hydro acoustics to calculate the transmission loss and analyze the effects of mesoscale ocean fronts and topography on deep-sea sound. The result shows that when the sound crosses the ocean front, the deflection of the sound line causes the spatial distribution of the sound field to vary; when the sound travels from offshore to the distance, the constraint on the topography makes the reflected sound waves of the seafloor into the surface water, which will limit the energy loss; when the sound travels from the distance to offshore, the topography makes the energy in the deep-sea duct leak, which compensates the transmission loss caused by the ocean front. The combined effects of topography and hydrographic environment make the acoustic duct exhibit complex change characteristics. Topographic condition plays a dominant role in the sound field pattern, and the non-uniform hydrographic environment caused by the ocean front is the key factor that causes abnormal changes in the spatial distribution of acoustic energy.

In today's society, information leakage events continue to occur, and various innovative and high-complexity network attacks emerge in endlessly. In order to ensure the integrity of data information and the practicability and authenticity in the transmission process, cryptosystems are usually adopted to ensure the security of information. In this paper, an asymmetric short integer parallel optimization algorithm is proposed, which combines a double Logistic chaotic system with a hash algorithm, and applies extensible precision chaos to reduce the rounding error of the chaotic mapping system by setting the number of iterations of the chaotic system and the accuracy of sequence calculation, which is to prevent the degradation of excellent chaotic characteristics and increase the mapping space. It improves the complexity of the chaotic system, thereby enhancing the security of the algorithm. Through the analysis of comparative experiments, it is proved that the algorithm in this paper has the smallest impact on computing power compared with several commonly used algorithms in different topologies, and further improves the network transmission effect. The invention can reduce the influence caused by the encrypted transmission of the network communication distribution service data. It can also maintain the bearing capacity of the link and improve the resource utilization rate.

A power allocation model of MIMO radar networking system is proposed in this paper to meet the requirement of radio frequency stealth performance of airborne radar in modern warfare. Based on the KL divergence criterion and the SNR criterion, the model optimizes the transmitting power between the networked radars so as to minimize the total transmitting power of the system under the condition that the detection performance meets the threshold. Besides, the concave-convex-programming algorithm in sequential convex programming is used to solve the non-convex constrained optimization problem. The effectiveness of CCCP algorithm is verified by comparing with other heuristic algorithms.

Global Positioning System (GPS)/Inertial Measurement Unit (IMU) fusion improves positioning accuracy. Existing fusion algorithms are mainly based on the Bayesian filtering framework. Still, the algorithm performs poorly in applying historical states, defining the state transition model, and correcting IMU computation errors during GPS interruption. This paper proposes GPS/IMU fusion localization based on Attention-based Long Short Term Memory (Attention-LSTM) Networks and sliding windows to solve these problems. we use Attention-LSTM networks to fuse GPS and IMU information to build a nonlinear model that fits the current noisy environment by training the model. In addition, the sliding window size can determine the number of historical state information utilized. We experimentally demonstrate that the method's positioning accuracy is improved under interrupted and uninterrupted GPS conditions.

In this paper, a new OTFS modulation channel estimation technique under high mobility conditions is proposed to face the main demand for stability in high-throughput communication networks in the new century, and the scheme proves to be significantly effective in solving the fast fading problem of time-frequency dual-selective channels, a new technique for multiplexing transmission data over the time-delay-Doppler domain. The LS channel estimation algorithm, the LMMSE channel estimation algorithm and the OMP channel estimation algorithm adapted to OTFS channels are used to investigate the channel estimation of OTFS systems, to estimate the channel environment parameters under high maneuvering conditions, and to select new modulation techniques.

In this letter, a cancelling-type structure (CTS) for decoupling compact (center-to-center spacing of 0.3 λ₀) microstrip antenna is proposed. The CTS consists of an array-antenna decoupling array(ADS) which is above the antenna and a decoupling probe(DP) which is connected to the ground plane. The decoupling mechanism is that by adjusting the size and position of CTS, the reflected wave generated by CTS can cancel the coupling wave. To verify the idea, the antennas with and without CTS are proposed and simulated. The simulated results show that the maximum mutual coupling reduction is 30 dB for S₁₂ at the center frequency. In the 1.2-1.3 frequency band, the S₁₂ is reduced to less than -20dB. Meanwhile, the impedance bandwidth remains almost unchanged. The CTS can also obviously correct the pattern distortion caused by coupling, making the pattern curve of the upper half space of the antenna smoother.

With the economic development, large buildings near the airport are being built in succession, and the terminals are also being built higher and higher. All these have brought great challenges to surface surveillance of the tower. If the monitoring is not in place, it is very easy for aircraft to collide with other aircraft and vehicles. In view of the complex airport environment, the monitoring methods of traditional tower can’t meet the requirements of airport security operation. In order to reduce the conflict events in the airport activity area, improving the detection probability of the existing equipment support methods, this paper introduces one surface movement radar based on the detection technology of moving target. Its signal processing design mainly adopts phase-coherent accumulation operation and CFAR (Constant false-alarm Rate) detection technology, which improves the detection probability of moving target in the airport active area and reduces the probability of False Alarm. The anti-interference ability of the radar system is improved, which not only ensures the flight safety of aircraft effectively, but also improves the operation efficiency of the airport.

Infrared imaging technology is widely used in national defense, industry, medical and other fields. High performance infrared imaging technology is highly valued by all countries in the world. However, the inhomogeneity, the inherent characteristic of infrared image, will seriously affect the real information of the image and seriously restrict the performance of infrared imaging system. In this paper, we proposed a shutter-less non-uniformity correction (SLNUC) algorithm based on ambient temperature. The SLNUC is based on the non-uniformity correction of the collected image of HgCdTE mid-wave infrared detector. The experimental results show that the SLNUC correction algorithm can adapt to the working requirements of a wide temperature range, without affecting the output of video stream, and the non-uniformity reaches 0.17% after correction, which lays a foundation for the development of new equipment.

At present, artificial intelligence and big data are flourishing, driving wireless communication services to a more efficient and intelligent direction, communication devices are increasing dramatically, and the communication environment is becoming increasingly complex, so the decision-making link is crucial to ensure communication performance as much as possible. To address the problems that existing waveform parameter decision algorithms rely on high a priori knowledge, lack of compatibility, and low decision efficiency, a reinforcement learning-based waveform parameter decision method is proposed. The method introduces a dynamic ε mechanism based on the hill-climbing strategy (PHC) under the architecture of the reinforcement learning algorithm and proposes a dynamic ε Q-learning intelligent decision algorithm, which enables the decision model to select ε values more optimally according to the state of the decision network and improves the convergence speed and decision success rate. The algorithm makes full use of the interaction between reinforcement learning and the environment and generates waveform parameter combinations which are suitable for the current channel environment in real-time through online learning. The decision model is based on a multi-carrier spread spectrum (MC-SS) communication system. The simulation results show that the new decision algorithm does not rely on a priori knowledge and has higher decision efficiency, which not only gives suitable decision results in Gaussian channels but also adapts to various fading channels and outperforms the mapping results provided by the Modulation and Coding Scheme (MCS) index table.

In this project, people can determine the location of the pig and lock the pig through the sound signal of the pig. Experienced keepers can also judge the cause of the sound according to the sound of the locked pig and determine the health status of the pig. This paper mainly studies how to enrich the audio spectrum features extracted from the original voice file and minimize the loss of signals in the process of generating spectrum features, especially the spectrum features that play a key role in downstream tasks. This paper proposes the fusion of audio spectrum features based on CNN model and MFCC audio spectrum features with different frame shifts and frame lengths, and serves as the input layer of feature representation and backbone model.

With the promotion of "ocean power" strategy, more and more scholars have paid attention to the development and application of underwater acoustic communication technology. In order to achieve high-speed and stable underwater acoustic communication and reduce complex data transmission. Based on underwater acoustic communication mechanism, this paper designs and expounds a set of underwater acoustic communication system and signal processing system, data interface module, underwater acoustic transducer, underwater electronic module, power supply module and many other components. This system effectively ensures the stability and simplicity of data transmission process, expands good universality, and is of great significance to the universal system application of underwater acoustic communication machine.

The advancement of the electronic information system represents the development level of the communication satellite, and has a significant impact on the communication satellite platform. This paper is the use of the "integrated electronic" design concept, based on efficient distribution of network system, strengthen electronic information system as a bridge between satellite and ground, satellite and satellite communication management, covering data processing, power supply and distribution management, satellite management, independent troubleshooting, meet the information of satellite platform data collection, management needs, ensure the security and reliability of satellite platform work, help satellite implement intelligent and autonomous management and control, improve the work of the quality and efficiency of satellite platform.

This study aimed to use the combined advantages of Space–Air–Ground Integrated Network (SAGIN) and BeiDou short-message technology to realize information interaction between various units within the system perception layer, considering the difficulties of poor communication in remote areas, limited coverage of existing ground networks, and the inability to send monitoring information of high-voltage transmission lines in a timely manner. The sensing terminal was used for terminal analysis and calculation, and the key exception and alarm information were edited into short messages to compensate for the disadvantages of BeiDou short-message communication technology. External communication used the BeiDou short-message communication technology. This method could effectively solve the transmission problem of key data of online monitoring systems in remote areas.

Operation box belongs to execution units of relay protection, with complex wiring, powerful function, and high flexibility. Based on two live examples, this paper focuses on the analysis of the selection of the low-pressure lockout circuit and the switch body. The hidden hazards caused by eliminating the low-pressure lockout circuit of the operation box have been inspected, and improvements have been put forward. Moreover, this paper discusses the countermeasures and the actual elimination of the power switching circuit of the operation box in the field, thus solving the hidden problem of inoperable reclose due to wrong wiring.

Widely used in telecom, television, music players, and controllers, a digital-to-analog converter (DAC) is a device that converts a digital signal into an analog signal. As integrated circuits (ICs) technologies develop, the modulation process has shifted from analog circuits to digital circuits, and the signal has been converted from a digital signal field to an analog signal field. Generally, the design of digital circuits is simpler than that of analog circuits, so the transferring can decline the requirement and difficulty of the system design. How the final DAC performs on the printed circuit board (PCB) directly affects the system's specifications. Therefore, verifying the final parameters of the DAC becomes an important task for the whole test. The DAC chip static parameters DNLity (DNL) and integral nonlinearity (INL) require measuring all points to get the calculated results. It will cost a long time if reading and recording are done by humans. This paper presents a fast and effective method to measure the static parameters of the DAC chip. The test time depends on the measurement accuracy and data. In this paper, the chip independently developed by Beijing Smart-Chip Microelectronics Technology Co. is used as a test case for testing. The results show that the DAC testing method mentioned in this paper can greatly shorten the testing time compared with manual testing.

Bursting disc is a safety relief device which is widely used in industrial production, especially in the arms industry, energy and other fields because of its' simple structure and the high safe capability which is better than safety valves, so the safe and stable operation of bursting discs has attracted widespread attention. However, the research both at home and abroad about bursting discs are concentrated on design and abnormal failure of bursting discs, and the research on the pressure relief test devices of bursting discs has less been published publicly. In order to solve problem of the pressure relief test of bursting discs, in this paper, the RBF neural network is used to compensate the temperature of the pressure sensor, it is of significance in strengthening the safety test of bursting disc.

A new BIST (Built-In Self-Test) method to test static parameters of a DAC (Digital to Analog Converter) is proposed in this paper. The BIST method employs a ramp generator and two voltage references to test static parameters of a DAC: offset error, gain error, INL (Integral Non-Linearity) and DNL (Differential Non-Linearity). The optimization of calculating static parameters and the components sharing can reduce the BIST circuitry. The simulation result shows that the method is able to detect the static errors with the simple BIST structure.

As an effective supplement to underwater acoustic signal, seismic wave excited by underwater acoustic sources in shallow water has gradually become a hot issue for scholars. Scholte wave has the characteristics of very low frequency, slow attenuation, long propagation distance and little influence by underwater sound velocity profile, which has the potential of realizing remote passive detection of underwater targets. The composition of seabed seismic wave excited by underwater acoustic source is complex, thus how to extract Scholte wave with high signal noise ratio from seabed seismic wave field is very important. In order to extract seabed Scholte wave with high resolution in shallow water-bottom interface, based on the polarization characteristics of seabed seismic wave, the time-domain polarization analysis method in seismology was improved in terms of filter to separate Scholte wave from other components of seabed seismic wave field, which is simulated based on the finite-infinite element coupling method. The Scholte wave in seabed seismic wave field is extracted and its background noise is suppressed using the improved method. Results of both simulations under the circumstances of hard and soft seabed show the effectiveness of the method, which also show that the length of time window can be determined by the input signal, and selection of the empirical coefficient is related to the type of seabed, and it increases with the increment of the shear wave velocity.

Motor imagery electroencephalogram (MI-EEG) classification is a vital task for brain computer interface (BCI) system. But the classification accuracy is not satisfactory, which hinders its generalizability. This study proposes a Filter-Bank Long Short-Term Memory (LSTM) Network (FBLSTM), which adopts a series of band-pass filters to obtain the different frequency information in EEG signals, and a convolution neural network to obtain spatial information. Moreover, a LSTM with attention mechanism is employed to process time series information. The open BCI Competition IV dataset 2a is applied to validate the performance of the proposed FBLSTM. Compared with recent methods, our method shows advantages on the within-subject and cross-subject 4-class classification performance and outperformed existing models, achieving an average accuracy of 72.4% and 53.6%, respectively.

The length of long reads produced by third-generation sequencing technologies is tens to hundreds of kbps which benefits genomic research. Still, the high error rate of long reads seriously limits the downstream analysis. Only by preserving the length advantage and reducing the error rate of long reads can the effectiveness of the downstream analysis be improved. Here propose LocPatcH: an accurate, efficient, and universal hybrid error correction algorithm based on local machine learning. LocPatcH constructs a profile hidden Markov model for each region in a long read which is aligned with abundant accurate short reads produced by the second-generation sequencing technologies, and then uses the alignment information of the short reads to train the model and finishes the correction. As for the rest of the aligned regions with lower coverage depths, the idea referred to as “patching” is used to complete the correction. The proposed method outperforms mainstream hybrid error correction methods in continuity and memory usage on real Pacbio and Nanopore sequencing datasets.

The intrusion detection model based on deep learning has become a research hotspot in the field of network security, aiming at the high FPR and low DR of current intrusion detection systems, this paper proposes an anomaly detection model based on deep learning. The system is first optimized by particle swarm optimization. Data feature extraction, and then through the fusion of MLP's bidirectional GRU model to better mine the information contained in the data set and its correlation, and improve the efficiency and quality of feature extraction. Finally, experiments show that the model has good performance, with a detection readiness rate of 98.74%, a false positive rate of 0.029, and a detection rate of 98.42%. The results show that the detection effect of PSO+BI-GRU+MLP is better than other algorithm models.

Aiming at the popularization of intelligent information collection technology and the demand of fairness and objectivity of human resource assessment, a human resource assessment management system based on K-means clustering is proposed. This paper introduces the principle and steps of K-means clustering algorithm, designs its functional modules and overall architecture in combination with the requirements and indicators of human resources assessment, designs and codes the realization of each functional module in detail, and gives the interface diagrams of some functional modules. Through the design, it provides a new reference for the objective evaluation of human resources assessment, and further strengthens the practical application of information technology in enterprise management.

The passenger throughput of a regional airport is an important indicator for measuring the development of a region and the basis for achieving the effective allocation of airport resources. Considering the factors affecting the passenger throughput of a western regional airport in China, this paper analyzed the correlation and significance between the factors and the passenger throughput. The stepwise regression method was used to reduce the dimension of original data and obtain the optimal regression model, and the gray model was used to predict the feature parameters. To verify the rationality of the passenger throughput prediction model, multiple linear regression and grey models were used to predict the passenger throughput in the same year and to compare them. The RMSE, MAPE and R² calculations indicate that: in terms of prediction stability, accuracy and fitting accuracy, the stepwise regression model and multiple linear regression model are better than the gray model, but the multiple linear regression prediction results are large. It means that the combination of stepwise regression and gray model can overcome the shortcomings of a single model, and make the prediction results more scientific and reasonable.

Sparse recovery space-time adaptive processing requires the use of dictionaries composed of redundant basis vectors, but sometimes the dictionary division is inaccurate. At this time, the off network effect will occur..Therefore, this paper studies the elimination of off-grid effect by two-dimensional iterative adaptive dictionary correction algorithm. In this method, the covariance matrix of clutter plus noise is estimated by the 2D IAA-STAP algorithm, and then the atoms most relevant to clutter points are obtained. The maximum joint likelihood function is used to search the clutter Doppler frequency atoms in the local area. Finally, a new space-time oriented dictionary is formed by iteration. The experimental simulation shows that the proposed method can make the clutter points attach to the clutter ridge better and reduce the off-grid effect more effectively.

When using Network Function Virtualization (NFV) technology, the deployment of Virtual Network Function (VNF) is dynamic and fast. The time-varying nature of service requests from users requires the resource allocation of the underlying infrastructure to be flexible in response to changes. If only horizontal scaling is used to dynamically add or delete virtual machine (VM) copies, it will lead to over-provisioning of some unnecessary resources and high reconfiguration costs, so how to effectively use the reconfiguration method to reduce resource consumption and reducing the corresponding reconfiguration cost is challenging. Therefore, this paper adopts a Hybrid Reconfiguration Strategy (HRS) that jointly considers horizontal scaling and vertical scaling to reduce the resource consumption and reconfiguration cost of VNF deployment, and establishes a corresponding integer linear programming model. Experimental results show that compared with existing methods, the total cost can be reduced, while ensuring a higher acceptance rate of service requests.

It is a nonlinear programming problem to solve the radar waveform design model of maximum mutual information under the condition of limited signal to clutter ratio. The complex solving process of particle swarm optimization - sequential quadratic programming algorithm leads to a large increase in the time consumed to solve the model and poor real-time performance. To solve the above problems, this paper adopts an adaptive strategy to update the speed and position of particle swarm optimization algorithm, and dynamically adjusts the inertia weight and learning factor to solve the problems of falling into local optimal and slow convergence. The experimental results show that the improved particle swarm optimization algorithm, while retaining the efficiency of the original algorithm, has faster convergence, higher mutual information, stronger robustness and reduces the time spent in solving the model in the process of solving the optimal radar signal waveform, thus achieving the purpose of improving the performance of radar system target detection, estimation and tracking.

Aiming at the underdetermined two-dimensional(2-D) direction-of-arrival(DOA) and polarization parameter estimation of L-shaped array, this paper established a data receiving model of three-quadrature dipole compression coprime array. In order to separate DOA and polarization parameter estimation, the self-covariance matrix without polarization parameter is constructed, and the virtual array without holes is obtained by vectorization operation, which greatly improves the continuous freedom of the array. At the same time, the 2-D DOA estimation is reduced to one dimensional(1-D) space, and two 1-D DOA are obtained by sparse representation method. An angle matching method based on Khatri-Rao product is proposed under the condition of degree of freedom(DOF) extension of polarization sensitive array. On this basis, the polarization parameter of the signal are obtained by the least square method. The simulation results show that the proposed L-shaped coprime array can achieve the array DOF extension. And the joint estimation of DOA and polarization parameters has higher direction finding accuracy and angle resolution than LV-MUSIC algorithm under the condition of low SNR.

In order to find the global optimal solution of the loop-free path with constraints, to solve the problems of low efficiency, and only local optimum in the traditional segmented splicing algorithm, an optimal calculation of the path with constraints based on an improved recurrent neural network is proposed. The embedded graph attention network structure, by specifying different attention weights for neighbor nodes, learns the graph embedding features of different constructed network topologies, and uses the learned graph features as the internal state of the recurrent neural network to predict according to the recurrent time sequence, the optimal order of the constraint segmentation is obtained, and finally the segmentation path is obtained according to the constraint segmentation, and the global optimal path is obtained by splicing. The experimental results show that, compared with the traditional segmentation splicing algorithm, the calculation amount of the algorithm in this paper is significantly reduced, the operating efficiency is significantly improved, and the global optimal solution can be obtained, and the generalization ability of the unknown network topology is extremely strong. In practical engineering applications It has very high practical value.

Continuous phase modulation, as the constant-envelop digital modulation scheme, has high spectral and power efficiency. However, how to realize the low-complexity demodulation for continuous phase modulation scheme is a challenge. In this paper, we give a new demodulation method based on maximum a posteriori probability and tilted-phase algorithm for continuous phase modulation scheme. Compared to the typical Viterbi algorithm, the proposed new demodulation algorithm can obtain similar demodulation performance. Moreover, the new algorithm does not require the redundancy modulation symbol to provide auxiliary information. The transmission efficiency is higher.

The frequency diverse array (FDA) radar has range-dependent and time-varying characteristics, which can eliminate the limitation of the phased array whose beampattern is angle-dependent only. FDA can create more possibilities due to its unique features. However, the range-angle coupling characteristic of the conventional FDA makes it difficult to be an effective way in anti-jamming and detection areas. In this paper, adopting adaptive genetic algorithm based thinned array, a beamforming method of FDA with non-linear frequency offset is proposed, which can form a sharp dot-shaped beampattern. Simulations demonstrate the validity of the method.

The space-time adaptive processing(STAP) technique adaptively suppresses clutter jointly in the space domain and time domain, but in practice, due to the lack of sample quantity, the performance of this technique suffers a serious loss. In contrast, the knowledge-aided space-time adaptive processing(KA-STAP) algorithm can improve the estimation accuracy of the clutter covariance matrix by using a priori knowledge. In this paper, by using the cycle characteristic of the clutter covariance matrix as the priori knowledge, the spatial circular matrix, the temporal circular matrix and the spatial-temporal circular matrix are constructed. And use the Least-Mean-Square(LMS) criterion to calculate the coefficients and then integrate the above matrices. Compared with the traditional STAP method, the KA-STAP using the cyclic property as a priori knowledge has about 1 dB of output signal-to-clutter-noise(SCNR) improvement.

Ultra-wideband radar has great application prospects in target detection and target imaging with the advantages of strong penetration, high range resolution, and precise target recognition. Since it is difficult to separate the target from the detection background in the B-scan image of ultra-wideband radar in a complex environment, this paper presents a method of object recognition based on edge detection and region segmentation. The Canny edge detection algorithm is used to extract the target signal in the complex background, and then the image region segmentation method is used to segment multiple targets in the dataset. To verify the feasibility of the algorithm, radar imaging experiments are carried out, and the experimental results show that the proposed method can better realize the target recognition and extraction of the B-scan image of ultra-wideband radar in complex environments.

To address the issue of tracking failure facing tracking algorithms due to object occlusion or deformation, an object tracking algorithm is proposed on the basis of positional joint filter bank and inter-frame prior information estimation. Based on the tracking framework of discriminant scale space tracking (DSST), A set of filters are introduced to track specific sub-regions of the tracking target to solve the tracking jitter problem of target due to partial occlusion under a single filter tracking; At the same time, the inter-frame transcendental information of video target is introduced to reckon location of target to solve tracking failure, which effectively addresses the problem of fully occlusion of target, and then different tracking strategies and update models strategy are selected according to the average peak correlation energy (APCE); Finally, in comparison with other tracking algorithms on dataset OTB2013，for example, benchmark tracker, the proposed algorithm improved the overlap success rate by 23.6%, and average accuracy by 31.7% under deformation and occlusion, which can effectively solve the issues of target occlusion and scale challenge and improve accuracy of target tracking.

Most unmanned clusters are typical mobile self-organizing networks (MANETs) with high dynamics and energy constraints. Deep reinforcement learning (DRL) is an emerging hotspot in the field of machine learning because it can make decisions without relying on models and is very suitable for dynamic unmanned cluster communication systems with time-varying network conditions. DRL -based unmanned cluster routing protocols are of great research value and significance. This paper introduces the fundamental characteristics of unmanned clusters and dynamic routing protocols; briefly describes the fundamental elements of DRL algorithms and their applicability in dynamic routing protocols; reviews the applications of DRL algorithms in unmanned clusters routing protocols; and summarizes the challenges and future development directions of unmanned clusters routing protocols based on DRL.

Empirical wavelet transform combines the idea of adaptive decomposition and the compact support frame of wavelet transform theory, and can decompose the signal into several modal components adaptively. This method is prone to modal aliasing and invalid components when dealing with non-stationary and strong noise signals. Then, this paper proposed a novel method combining the improved empirical wavelet transform the unbiased autocorrelation theory and applies it to the feature extraction of bearing faults. The new method employs order statistic filter instead of scale-space representation to segment the Fourier spectrum. The minimum points of the spectral envelope obtained by the maximum order statistic filter are regarded as boundaries of the spectrum. Then the filter banks are constructed according to boundaries and Meyer wavelet and the corresponding modes are screened. Finally, the unbiased autocorrelation theory is used to filter the component containing bearing fault characteristic to realize fault diagnosis. Simulation signals and rolling bearing fault signals verify the effectiveness of the method.

Forward looking imaging has broad application prospects, array radar is widely applied to angular super-resolution in forward-looking imaging. Based on the derived array radar forward-looking echo model, forward-looking imaging algorithms based on MUSIC estimation and single snapshot data sparse decomposition are given. Also, an algorithm based on joint sparse decomposition of multi-pulse amplitude and phase data is proposed. The imaging performance of the above three algorithms are analyzed. Experimental results demonstrated that the multi-pulse amplitude and phase data joint sparse decomposition algorithm has the highest resolution and is suitable for complex scenes.

In order to explore the realization method of network security in the logical security domain of primary and secondary integration devices of edge power, a realization method of network security in the logical security domain based on primary and secondary integration devices of edge power is proposed. This method recommended key technical problems and solutions based on information represented by edge power, and explored the research of network security in the logical security domain of edge power primary and secondary integration equipment. The research shows that the efficiency of realizing network security in the logical security domain based on the primary-secondary integration of edge power equipment is about 30% higher than that of traditional methods. With the development of intelligent distribution network, distribution automation has been popularized, and distribution network will be closer to transmission network in automation degree, which makes the theory of distribution system security domain (DSSR) become a reality.

Jamming attacks can significantly degrade the performance of wireless communication systems and plays a very important role in electronic countermeasures and radio resource management. Most of the relevant research in recent years has been conducted to analyze the optimal jamming of digital modulation signal under the condition of constellation diagram without distortion. However, in non-ideal reception conditions, the optimal jamming scheme will no longer be the conventional one due to the constellation distortion. Therefore, aiming at the optimal jamming scheme for non-ideal received multiple quadrature amplitude modulation (MQAM) signals, a grey wolf optimization-orthogonal decomposition (GWO-OD) based optimal jamming waveform construction algorithm is proposed in this paper. Based on the orthogonal decomposition algorithm, this method can freely distribute the power between the in-phase dimension and the orthogonal dimension. The optimal allocation scheme is selected by using the GWO algorithm, and various unconventional jamming modulation signals are constructed. The simulation results show that the proposed algorithm can intelligently select the jamming scheme with the best jamming effect. In the case of constellation distortion, the jamming modulation scheme constructed by this method has a better jamming effect than the conventional scheme, which is better than the existing optimal jamming methods.

Aiming at the problem of random generation of Extreme Learning Machine (ELM) parameters, an intrusion detection model based on GA-ELM of K-fold stratified cross-validation is proposed. Genetic Algorithm (GA) is used to optimize the parameters of ELM. The simulation experiment is carried out on NSL-KDD data set, and the GA-ELM model is trained by K-fold stratified cross-validation method. The experimental results show that the proposed model has a higher detection rate than ELM, GA-ELM and GA-ELM of nonstratified cross-validation.

The essence of grayscale is to convert a three-dimensional color image into a one-dimensional grayscale image, and information loss is inevitable, so it is necessary to preserve the contrast, detail and structure of the original color image as much as possible. In this paper, we propose a new contrast retention grayscale method based on weighted guided filtering. Firstly, considering the different sensitivity of the human visual system to the three color channels of R, G, and B and the different information contained in the images of each channel, a new contrast feature coefficient is defined to better retain the contrast characteristics of the original color image. Secondly, in order to make the grayscale image have a better sense of hierarchy, the saturation feature holding term is designed. Finally, the discrete search strategy is used to optimize the proposed new model, and on this basis, weighted guided filtering is used to enhance the edge details of grayscale images. A large number of experimental results on the three datasets show that the grayscale effect of the proposed method is better than that of the traditional grayscale method.

The sound signal of power transformers is a convenient and reliable data source for determining machine faults and can be used to monitor the working status in real-time. There are several deep learning methods trained with labeled data for anomalous sound detection, which can identify a limited number of fault types. However, in real scenes, the large number of unlabeled anomalous sounds makes it difficult for existing methods to detect accurately. To solve the above challenges, this paper designs a self-training learning algorithm for anomalous sound detection for supercomputing platforms and expands the training datasets with pseudo-labeled data to enhance the migration capability of the detection algorithm in complex scenes. The performance metrics are evaluated based on the MIMII DG dataset and the local dataset from the DCASE 2022 task II. The results show that the baseline model with the self-training algorithm improves the AUC significantly over the initial baseline model across scenarios. At the same time, relying on the powerful computing power of the supercomputing platform, the iteration time of model self-training learning can be further compressed, which can quickly get the detection model adapted to new scenes.

The power Internet of Things technology plays an important role in ensuring the normal operation of power grid. In order to improve the reliability of information perception communication of distribution network, this paper studies the key technologies of power line and wireless communication fusion, so that the two communication modes complement each other, improve the reliability of data transmission and communication coverage of distribution network, and adopt orthogonal frequency division multiplexing(Orthogonal Frequency Division Multiplexing, OFDM) to optimize resource allocation and enhance network performance. In this paper, a fusion method of low-voltage power line and micro-power wireless communication is proposed. By establishing a unified communication protocol in the medium access control layer of the hybrid communication network, realizing the optimal communication path networking in the network layer, and realizing the sub-service flow allocation based on the bit error rate demand factor in the business layer, the cross-layer fusion of low-voltage power line communication and micro-power wireless communication is realized in the service layer. The simulation results show that the performance of the hybrid communication network is better than that of other comparison networks. This paper presents an OFDM cross-layer resource allocation algorithm in multi-hop relay powerline communication networks and a power line and wireless communication fusion method including OFDM resource allocation.

Considering the fluid nonlinearity, a calculation model based on quasi-linear solution of Westervelt’s equation is proposed in this paper. A more accurate approximation of Rayleigh integral is derived and velocity distribution in parametric loudspeaker surface is fitting by a set of Gaussian functions. Gaussian beam expansion is used to simply five-fold integral in calculation difference-frequency wavefield. Spatial sound pressure distribution of an annular array composed of serval rigid disks are presented. It can be found that uniform distribution of difference-frequency wave in xoy-plane is realized by annular array.