Abstract
This paper presents a neural network based fault diagnosis approach for analog circuits, taking the tolerances of circuit elements into account. Specifically, a normalization rule of input information, a pseudo-fault domain border (PFDB) pattern selection method and a new output error function are proposed for training the backpropagation (BP) network to be a fault diagnoser. Experimental results demonstrate that the diagnoser performs as well as or better than any classical approaches in terms of accuracy, and provides at least an order-of-magnitude improvement in post-fault diagnostic speed.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Liu R W. Testing and Diagnosis of Analog Circuits and Systems. Van Nostand Reinhold, New York, 1991, pp. 1–35.
Starzyk J Aet al., Artificial neural networks for testing analog circuits. InProc. of IEEE International Symposium on Circuits and Systems, New Orleans, May 1–3, 1990, pp. 1851–1854.
Haley P Jet al. Extrapolation limitations of multilayer feedforward neural networks. InProc. of IJCNN’92, Baltimore, Vol.IV, pp. 25–30.
Xu Sixinet al. A fast learning method and application for the neural toward-networks.Control and Decision, 1993, 8(4): 284–288.
Author information
Authors and Affiliations
Additional information
This project is supported by the National Natural Science Foundation of China.
Wei Naihong received her B.S. degree from Tsinghua University in 1989 and her M.S. degree from Beijing Institute of Technology in 1992. Now she is a Ph.D. candidate in Electronics at Tsinghua University. Her current research interests include analog circuit fault diagnosis, neural networks and fault tolerant computing.
Yang Shiyuan was born in Shanghai, China, on Nov. 15, 1945. He received his B.S. degree and his M.S. degree from Tsinghua University in 1970 and 1981 respectively. From 1989 to 1995, he was an Associate Professor of Department of Automation, Tsinghua University. Since July 1995, he has been a Professor of the same department. He is the cochairman of the Tolerant Computing Committee of China and a senior member of Chinese Electronic Institute. His current research interest is fault detection and diagnosis in digital and analog circuits, application of ANN to fault diagnosis, reliability design of control system and EMC of electronic equipment.
Tong Shibai was born in Liaoning, China, on Feb. 14, 1920. He received his B.S. degree from the National Southwest Associated University at Kunming, China, in 1946, and his M.S. and Ph.D. degrees in E.E. from the University of Illinois at Urbana, U.S.A., in 1949 and 1951 respectively. He taught Electronics and Control theory at the Polytechnic Institute of Brooklyn, N.Y., from 1952 to 1955. He returned to China in July 1955. Since then, he has worked at the Electrical Engineering Department and Automation Department as a Professor in electronics and automation. His research area is in the field of automatic testing, fault diagnosis and electronic system reliability. Dr. Tong is the Chairman of UN-ESCO INCCA Beijing Computer Application Center, a member of Sigma Xi and chief editor of several textbooks on electronics. He also was the Chairman of Electronic Engineering Department of Shenzhen University from 1983 to 1986, the Chairman of the Electronic Course Directory Group of the State Education Commission from 1985 to 1990, member of the standing Council of China Instrument Society from 1985 to 1990.
Rights and permissions
About this article
Cite this article
Wei, N., Yang, S. & Tong, S. A neural network approach to fault diagnosis in analog circuits. J. of Comput. Sci. & Technol. 11, 542–550 (1996). https://doi.org/10.1007/BF02951617
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF02951617