The Research On Method And Theory Of Large-scale Analog Circuit Fault Diagnosis

The research of analog system test and diagnosis began at 1960's. Traditional analog circuits were always diagnosed by manual work for its'small scale and low integration. With the development of electronic industry, the scale of electric circuits becomes larger and more complex, it is urgent and necessary to realize the auto test and fault diagnosis of large-scale analog circuit. Then, the method of large-scale analog circuit fault diagnosis becomes a research focus in circuit and system of the world.This paper concentrated on the algorithmic research of large-scale analog circuits fault diagnosis. An improved neural network method was proposed combined fuzzy reasoning location system to locate fault area, which put large analog circuit method to system level.Some representative neural networks were discussed for analog circuit fault diagnosis in this paper. Then, on the basis of large-scale network diagnosis method——network decomposition, a neural network method was developed to diagnose fault in large-scale analog circuit including tolerance situation, at the same time, sub network decompose course and data processing can be realized also.After the situation of pattern recognition applied into analog circuit fault diagnosis was been researched deeply, and on the base of neural network and fuzzy reasoning system integrating, a system network diagnosis method was proposed. Fault assumption,verification,test condition and result judge were integrated in the method, which can realize diagnosis systemize. Simulation confessed: the method has high location precision and fast decomposition rapid, which can be applied to diagnose large-scale analog circuit fault.As another work in this paper, module diagnosis method was proposed. The method combined fault dictionary diagnosis with fault tree analysis. The fault dictionary is harder to be constructed if analog circuit has a large scale. Now, this problem can be relieved by module diagnosis method, meanwhile, the workload before test will be reduced in a certain extent. Analog simulation shows that the method has many advantages such as: isolating and minimizing fault location range and so on.