Intermittent Fault Diagnosis of Analog Circuit Based on Enhanced One-Dimensional Vision Transformer and Transfer Learning Strategy

作     者：Wang, Shengdong Liu, Zhenbao Jia, Zhen Vong, Chi-Man Li, Zihao 

作者机构：School of Civil Aviation Northwestern Polytechnical University Xi'An710072 China School of Mechanical and Electrical Engineering Xi'an University of Architecture and Technology Xi'An710055 China Dept. of Computer and Information Science University of Macau 999078 China 

出 版 物：《SSRN》 

年 卷 期：2023年

核心收录：

主　　题：Fault detection 

摘      要：Intermittent faults of analog circuits are major causes of built-in test (BIT) false alarms. To enhance the reliability and maintainability of electronic system, an end-to-end approach based on enhanced one-dimensional Vision Transformer (1DViT) is constructed to realize intelligent diagnosis of intermittent faults. One multiscale convolution fusion module (MSC) is introduced into 1DViT to mine fault features on multiple time scales to enhance the interaction between global and local information. In addition, due to the complex operation and measurement conditions, it is difficult and time-consuming to obtain a large amount of data from practical test. To cope with this problem, transfer learning is introduced. The model will be first pre-trained with adequate simulation data which is readily accessible, and then fine-tuned with actual test data to match the practical distribution. Experiments on two typical circuits demonstrate that the proposed method could achieve excellent diagnostic result in practical test. © 2023, The Authors. All rights reserved.