基于随机森林的异步电动机转子断条故障诊断

摘要

为了快速准确识别异步电动机转子断条故障,提出一种随机森林算法(Random Forests,RF)的异步电动机转子断条故障诊断方法.该方法以Park矢量模平方函数为故障特征提取手段,然后采用随机森林进行故障模式自动识别.Park矢量模平方函数方法可以较好地削弱定子电流中基频信号的影响,便于故障特征量的准确提取.而随机森林泛化能力好、训练时间短,提高了故障识别的成功率.实例验证表明,基于随机森林的异步电动机转子断条故障诊断方法性能良好.%To identified broken rotor bar faults in induction motors accurately and rapidly,a novel method was illustrated to diagnose broken rotor bar fault on the basis of random forests.Park's vector modulus could reduce the influence of the fundamental frequency signal in stator current,which could help extract the feature vector accurately.Random forests had better tolerance,lesser training time consuming and strengthening the classification correctness.As a result,the experiment reminds the effectiveness and superiority of the proposed method.