基于改进粒子群算法神经网络的电力电子电路故障诊断

Fault diagnosis of power electronic circuits based on improved PSO algorithm neural network

粒子群优化(PSO)算法是基于群智能的全局优化技术,它通过粒子间的相互作用,对解空间进行智能搜索,从而发现最优解。对基本粒子群算法进行改进,并将改进粒子群优化算法与误差反向传播(BP)算法结合起来构成的混合算法用于训练人工神经网络,对电力电子电路故障进行在线诊断。仿真结果表明,改进PSO-BP算法有效地解决常规BP算法学习网络权值和阈值收敛速度慢、易陷入局部极小等问题,具有较快的收敛速度和较高的诊断精度。

Particle Swarm Optimization （PSO） algorithm is a global optimization technology based on the group intelligence, it carries on the intelligent search for the solution space through mutual effect in order to discover the optimal solution. The improvement to the basic particle swarm algorithm is carried out to create a mix algorithm, which is the mix of combining the improved PSO algorithm with the erroneous reverse dissemination （BP） algorithm. This mix algorithm can be used for training the artificial neural network, diagnosising the fault for power electronic circuits .The simulation results show that the improved PSO-BP algorithm solves effectively the problems of the conventional BP algorithm network convergence rate slow and easily falling into partial minimum, it works with quicker convergence rate and the higher forecast precision.