Application Of Dempster-Shafer Evidence Theory In Fault Diagnosis Of Machine Equipment

In the fault diagnosis of mechanical equipment, it is often difficult to identify the faulttype with the signal measured. Information fusion supply a new approach to diagnose themachine faults with help of varies information. This article focuses on the disadvantage ofD-S evidence and the characteristic of BP neural network. A model combined neural networkand data fusion method are used to recognize the faults occurred in a rolling element bearingand an industrial fan in power plant. Comparing with the approaches that only adopt onemethod, the accuracy of diagnostic results is obviously improved, and the vibration signalsanalysis proved this conclusion. In the paper the computer aided maintenance system of feed water pump in power plantwas discussed and a software system was developed.