Abstract: Aiming at the low detection rate and long diagnosis time of the traditional neural network methods, particle swarm optimization (PSO) is used to train neural network and optimize connection weights. It is applied to fault diagnosis of optical-electronic detection equipment. Compared to BP and GA, the experiment results show that PSO neural network can improve the fault detection rate, decrease the false alarm rate of optical-electronic detection equipment and fault diagnosis time, and be realized more easily. Therefore, It gets better effects of fault diagnosis.

Key words: Back propagation (BP), Particle swarm optimization (PSO), Genetic algorithm (GA), Neutral network, Optical-electronic detection, Fault diagnosis