Analog Circuit Fault Diagnosis Based On Particle Swarm Optimization

It has never been able to build a unified and effective fault models in analog circuits because of the influence caused by the factors like component tolerance, non-linearity, temperature drift etc, leading to a low efficiency for traditional diagnostic methods. Neural networks have excellent abilities in many respects including parallel distributed processing, self-adaption, and associative memory etc, which brings new life to analog circuit fault diagnosis.An analog circuit fault diagnosis method based on particle swarm optimization was proposed to solve the soft fault diagnosis. I use the information entropy to describe the uncertainty of the circuit fault status to build the objective function, search an optimal feature subset which carries the most fault information combining with PSO, construct the sample set for neural network and use this set to train RBF neural network to realize the circuit fault diagnosis using the trained neural network. Experimental results showed that the neural network’s diagnostic efficiency could reach up to95%trained by the feature set which searched by PSO algorithm. The results verified the feasibility of this diagnostic method.