The Research Of Fault Diagnosis Of Analog Circuits Based On Parameter Identification

The research on the fault diagnosis of analog system is one of the most popular foreland problem at present time and even in the future, that is because the reliability of the electronic system is dependent on the anlog circuits, which has a very important practical significance.At first, this paper presents the basic conception and technique of the fault diagnosis of analog circuits. With a detailed presentation of its characteristic, it gives the difficulties and the development trend, and then discusses the general diagnosis method and pattern identification technique. The main purpose of this paper is to research on the diagnosis of the analog circuit with tolerance based on parameter identification, it try to find a novel method of the fault diagnosis of the analog system. In this paper, a procedure for the multifrequency fault diagnosis of analog circuits is presented.Generally speaking, for the diagnosis equations are nonlinear, it's very difficult to solve them with high precision. So it is necessary to apply the Neural Network(NN) and Genetic Arithmetic (GA) to the solving of these equations, it can simplify the diagnosis process and reduce the testing time. This paper introduces the basic GA and its development, and provide an arithmetic to the diagnosis equations with GA. This paper also have a research on BP and Hopfield network, by using the NN, it provide a mechanism to deal with the diagnosis equations. The stability, astringency and the feasibility of this method can be strictly proved. The simulation results shows that the proposed methods can perform correct diagnosis in a very short time.This paper provides a novel method for fault diagnosis of large-scale analog circuits with tolerance, which has academic and practical signification.