基于最小二乘支持向量机和能量频谱分析的电机匝间短路故障诊断

针对异步电机的构造与定子匝间短路故障,提出了一种基于最小二乘支持向量机和能量频谱分析的故障诊断方法,该方法通过样本误差来选择贡献量比较大的样本,同时又顾及到其它样本的一些属性,再通过最小二乘支持向量机来进行训练,在保证算法的精度和推广的能力的同时,可以减少训练样本的数量,因而训练速度会很快。此方法具有鲁棒性和快速性,同时它也会削弱部分干扰样本的影响,从而减少误判的概率。实验结果显示,使用最小二乘支持向量机和能量频谱分析的故障诊断方法,对电机故障方式进行诊断具有一定的可行性和有效性。

The motor fault diagnosis of turn-to-turn short circuit based on LS-SVM and energy spectrum analysis

Aiming at the formation of the asynchronous dynamo and the stator inter-turn short circuit faults,a method of fault diagnosis based on LS-SVM and Energy spectrum analysis is proposed.This method chooses samples which contributes more through comparing the sample error,while takes the qualities of other samples into account.Then the LS-SVM is used for sample training.The method diminishes the amount of samples needed and in the meantime guarantees the precision of the algorithm and the generalization ability.Therefore,the training goes fast.This method has advantages on the robustness and training speed,and it also weakens the impact of some interference samples,therefore decreases the probability of erroneous judgment.The results of experience present that the method combines feasibility and effectiveness for the fault diagnosis of inter-turn short circuit of motors.