修正的D-S证据理论在液压故障诊断中的应用

Modified D-S Evidential Theory in Hydraulic System Fault Diagnosis

针对液压系统出现故障隐蔽性强、可变性强的问题,给出基于修正的D-S(Dempster-Shafer)证据理论的多传感器时空域信息融合方法。通过测试待诊断部分的压力、温度、流量并结合经验值,得到各部分信度函数分配,根据D-S组合、决策规则,对多个传感器多个测量周期的时空域信息进行融合,识别出故障部分。实验表明:该方法在证据高度冲突时仍有效果并有效防止了指数爆炸现象,算法简单,能准确定位故障。

In order to solve the problem of hydraulic system fault diagnosis,an information fusion method,which is used for hydraulic system fault diagnosis based on modified D-S evidential theory and time-space domain,is presented.According to the measurement results of the pressure and the temperature and the flux,the belief function assignment of the three sensors was obtained,then the DS rule and decision rule were used to fuse the information in time-space domain,and at last the fault was recognized.The simulated experiment show that the method in conflicted evidence is still correct and can avoid exploded index and exactly fix the fault.