基于加权证据理论的异步电机故障诊断研究

Study of the Induction Motor Fault Diagnosis Based on Weighted Evidential Theory

针对D-S理论不能解决高冲突证据的融合问题,本文提出一种基于证据间距离和先验知识的加权证据理论。利用异步电机失电残余电压和Park矢量模平方各自的故障特征频率分别进行初步故障诊断,诊断结果分别作为证据m1和m2,同时引入已知证据m3,再用本文提出的加权证据理论作最后的决策诊断。实验表明,诊断结果的准确性明显提高并且高冲突证据的融合问题也能有效解决。

Because the D-S theory failed to settle the fusion problem of evidences with high conflict,a kind of weighted evidential theory,which was based on evidences gap and field background knowledge,was proposed.Residual voltage after AC dump and the square of the Park vector modulus were separately used to preliminarily diagnose fault by their respective frequencies of fault features.Their diagnosis was separately used to act as evidences m1 and m2.Then final diagnosis can be made through the weighted evidential theory based on m1,m2 and known evidence m3 which was simultaneously brought in.The experiment demonstrated that accuracy of the fault diagnosis was improved obviously and the fusion problem given above was also solved effectively.