基于PSO优化的BP神经网络在电动机绝缘剩余寿命预测中的应用

Application of BP Neural Network Based on PSO in Motor Insulation Residual Life Prediction

针对普通的电动机绝缘剩余寿命预测模型收敛速度慢、结果偏差大的缺陷,提出了一种基于粒子群算法(PSO)优化BP神经网络的电动机绝缘剩余寿命预测模型。首先,利用PSO算法全局随机最优解搜索的特性,对传统BP神经网络模型的权值和阈值进行优化设计。其次,为便于预测模型的运算处理,对采集的三相异步电动机的数据进行归一化处理。最后,结合经PSO算法优化的BP神经网络模型对三相异步电动机的绝缘剩余寿命进行试验预测。结果表明,基于PSO优化的BP神经网络比传统BP神经网络有更为精准的预测能力以及更快的收敛速度。

Aiming at the drawbacks of slow convergence and large deviation of the results in general motor insulation residual life prediction model, a motor insulation residual life prediction model based back propagation neural network optimized by particle swarm optimization （PSO） is proposed. First, the features of the global random optimal solution search of PSO are utilized to optimize the weights and thresholds of the traditional back propagation neural network model. Secondly, in order to facilitate the computation of the prediction model, the acquisition data of triple-phase asyn chronous motors are normalized. Finally, back propagation neural network model optimized by PSO is applied to prediction of triple-phase asynchronous motor＇s insulation residual life. The experimental results show that the BP neural network based PSO has more precise prediction ability and faster convergence speed than that of the traditional BP neural network.