Application of Particle Swarm Optimization and RBF Neural Network in Fault Diagnosis of Analogue Circuits

BP neural network has the shortcoming of over-fitting, local optimal solution, which affects the practicability of BP neural network. RBF neural network is a feedforward neural network, which has the global optimal closing ability. However, the parameters in RBF neural network need determination. Particle swarm optimization is presented to choose the parameters of RBF neural network. The particle swarm optimization-RBF neural network method has high classification performance, and is applied to fault diagnosis of analogue circuits. Finally, the result of fault diagnosis cases shows that the particle swarm optimization RBF neural network method has higher classification than BP neural network.