基于多征兆域的异构集成故障诊断策略

Study on Fault Diagnosis Based on Heterogeneous Integration of Multi-symptom Domains

为了充分利用机械设备振动信号的信息,并避免学习的复杂性,提出多征兆域异构集成故障诊断方法,即提取信号在不同征兆域的故障特征,利用异构分类器分别学习,集成各分类器的结论.该方法充分利用了多征兆域信息的冗余性、互补性和辅助性,以及异构分类器方法之间的互补性,提高了故障诊断的准确性和确定性.文中方法在转子系统故障诊断应用中得到验证,结果表明,多征兆域异构集成方法明显优于单征兆域单分类器方法与单征兆域集成方法.

In order to make full use of the information of vibration signals of mechanical equipment and avoid the complicated learning,a heterogeneous integration method of multi-symptom domains was put forward.That is,the fault characteristics of different symptom domains were extracted,and heterogeneous classifiers were used to study different symptom domains separately,and their conclusions were integrated.This method makes full use of the redundancy,complementarity and auxiliary of information of different symptom domains and the complementarity of heterogeneous classifiers,so the accuracy and certainty of fault diagnosis were improved.This method was used to diagnose the faults of rotor system.It is shown in the results that heterogeneous integration of multi-symptom domains is better than single classifier and heterogeneous integration of single symptom domain.