1. A novel fault diagnosis method for analog circuits with noise immunity and generalization ability
- Author
-
Shouda Jiang, Tianyu Gao, and Jingli Yang
- Subjects
0209 industrial biotechnology ,Computer science ,business.industry ,Dimensionality reduction ,Feature extraction ,Pattern recognition ,02 engineering and technology ,Fault (power engineering) ,Linear discriminant analysis ,Autoencoder ,Support vector machine ,020901 industrial engineering & automation ,Artificial Intelligence ,0202 electrical engineering, electronic engineering, information engineering ,020201 artificial intelligence & image processing ,Noise (video) ,Artificial intelligence ,Cluster analysis ,business ,Feature learning ,Software - Abstract
To enhance the reliability of analog circuits in complex electrical systems, a novel fault diagnosis method is presented in this paper. A denoising autoencoder and a sparse autoencoder are combined, producing a feature extraction model named the denoising sparse deep autoencoder (DSDAE) that can obtain effective information from signals contaminated by noise. Compared with traditional feature extraction methods, the DSDAE model can be used to implement adaptive feature learning. Then, linear discriminant analysis is adopted to perform linear dimensionality reduction, thereby obtaining the maximum clustering features of the signals. Finally, a fault diagnosis model based on a support vector machine (SVM) with high versatility and accuracy is developed to identify the fault classes of analog circuits. In addition, the salp swarm algorithm, which is capable of convergence and strong global optimization, is employed to intelligently optimize the SVM classifier. The method is comprehensively evaluated with three typical analog circuits from the ISCAS’97 circuit set. The experimental results illustrate that the proposed fault diagnosis method can achieve excellent fault identification accuracy and generalization performance even under noise interference conditions.
- Published
- 2021