Genetic Algorithm (GA) Optimized Neural Networks For Analog Circuit Fault Diagnosis

On the base of traditional diagnosis and theory, a second-order voltage control voltage source band-pass filter circuit was taken as studying object, BP Neural Network and genetic algorithm (GA) were applied in the analog circuit fault diagnosis in this paper.First of all , this paper analyzed the advantages of NN which is applied in fault diagnosis, and introduced the forming, development and study actuality of intelligent diagnosis use NN. And then, this paper described BP NN model architecture and studying mechanism significantly. The problems of BP NN and several improved algorithms were analyzed and studied. In order to solve the problems of BP NN, genetic algorithm was used to optimize BP NN, form to GA-BP network. The original weights and biases set by the traditional BP NN were taken place by the optimized data set by GA, and then the network is trained by improved BP algorithm—LM algorithm. After comparing simulation results worked by MATLAB, we can find that the train epochs of GA optimized BP NN decreased a lot. Many disadvantages of traditional BP NN were overcome by using GA-BP. The GA-BP method improved BP NN a lot, made the accuracy rate of fault diagnosis to the band-pass filter circuit much higher.