Transformer fault diagnosis model based on FI-CNN method

Nan Lin; Zhengwei Guo

doi:10.1117/12.2628003

14 March 2022 Transformer fault diagnosis model based on FI-CNN method

Nan Lin, Zhengwei Guo

Proceedings Volume 12165, International Conference on Intelligent Traffic Systems and Smart City (ITSSC 2021); 1216522 (2022) https://doi.org/10.1117/12.2628003
Event: 2021 International Conference on Intelligent Traffic Systems and Smart City, 2021, Zhengzhou, China

Abstract

At present, the three-ratio method and artificial intelligence algorithm are widely used in transformer fault diagnosis. However, the three-ratio method based on DGA has problems such as lack of coding and insufficient anti-interference ability of various artificial intelligence methods, resulting in a high rate of misdiagnosis. This paper proposes a transformer fault diagnosis model based on FI (Feature Image)-CNN. It takes the percentage of dissolved gas in the oil as an input parameter, constructs a feature vector, and extracts the two most widely distributed data as two-dimensional coordinates based on the characteristics of the fault. Axis, and then uses the RGB principle to construct a three-layer neuron structure with the remaining gas data to form a “RGB dynamic map” of the training sample; build a convolutional neural network model, combine the faulty DGA data, and use the DNN full-link network for fitting, and finally Realize the fault diagnosis of the power transformer. Finally, it is compared with the diagnostic performance of various optimization algorithms. The results show that the FI-CNN method can obtain high-dimensional curves that divide the fault categories according to the "RGB dynamic map" of the training samples and the corresponding spatial locations, and has higher diagnostic accuracy.

Citation Download Citation

Nan Lin and Zhengwei Guo "Transformer fault diagnosis model based on FI-CNN method", Proc. SPIE 12165, International Conference on Intelligent Traffic Systems and Smart City (ITSSC 2021), 1216522 (14 March 2022); https://doi.org/10.1117/12.2628003

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available