基于改进的粒子群优化的神经网络故障诊断方法研究

ON FAULT DIAGNOSIS OF NEURAL NETWORK BASED ON IMPROVED PARTICLE SWARM OPTIMIZATION

针对BP(Back Propagation)神经网络易陷入局部极值的缺点,提出了一种粒子群PSO(Particle Swarm Optimization)神经网络,同时为避免PSO算法早熟,对部分粒子采用变异操作。应用于故障诊断系统的仿真结果表明,该算法能够大大提高故障诊断的精度。

A PSO neural network is proposed for overcoming the disadvantages of BP neural network in its easily being trapped into local maxima.At the same time,in order to avoid premature convergence in basic PSO algorithm,some mutation operations are conducted upon the particles.The results of simulation on applying it to fault diagnosis system show that the improved PSO neural network algorithm can improve accuracy of the fault diagnosis greatly.