基于GA-IBP神经网络的故障诊断方法

Fault Diagnosis Method Based on GA-IBP Neural Network

针对BP(Back Propagation)神经网络在进行故障诊断时准确度低、收敛速度慢等问题,设计了一种基于误差指针改进的BP(Improved Back Propagation,IBP)神经网络,并通过遗传算法(Genetic Algorithm,GA)对这种改进后的神经网络进行优化,从而建立了基于GA-IBP神经网络的故障诊断模型.使用典型三相逆变电路中IGBT开路故障数据作为样本,对所设计的模型进行了仿真分析.结果表明:改进后的网络模型收敛速度优于典型BP神经网络和基于GA算法优化的典型BP神经网络,故障诊断精度分别提高15%和4.5%.

Aiming at the problems of low learning efficiency and slow convergence speed of BP(Back Propagation) neural network in fault diagnosis, an Improved BP neural(IBP) network based on error pointer was designed;Genetic Algorithm(GA) was used to optimize the initial parameters of the improved network, and the GA-IBP neural network fault diagnosis model was established. The IGBT open circuit fault data in typical three-phase inverter circuits were used as samples to simulate and analyze the designed model. The results show that the convergence speed of the improved network model is better than the typical BP neural network and the typical BP neural network based on GA algorithm optimization, and the fault diagnosis accuracy is improved by 15% and 4.5% respectively.