Analog circuit fault diagnosis based on fuzzy support vector machine and kernel density estimation

Because analog circuits such as abnormal noise contained in the information, to the support vector machine to build up the optimal classification brings difficulties, this paper proposes a new method for analog circuit fault diagnosis. First of all, time-domain signal extraction circuit statistical parameters, a set of fault characteristics and then use kernel density estimation method, proposed a form of fuzzy membership function construction, to eliminate the impact of noise characteristics. The establishment of such a membership functions with fuzzy support vector machines on the circuit fault diagnosis. Through the training of support vector machine fault diagnosis model was to achieve single-fault and multi-circuit fault diagnostic classification. The method is applied on CSTV filter circuit, the simulation experiment results show that the method can highlight the different characteristics of fault can be diagnosed correctly and effectively multi-fault types, comprehensive diagnostic accuracy of 95%, and the method for analog circuit fault diagnosis a new way. This technology has good prospects for engineering applications.