Study on data fusion faulty diagnosis method of hydraulic pump based on time-domain analysis
This paper introduces the signal parameters which are applied to the machinery fault diagnosis in timedomain, builds multi-source data fusion fault diagnosis system based on D-S evidence theory and artificial neural network, selects 8 parameters in time-domain as outputs of the data fusion level module, adopts two collateral neuronal networks whose structures are the same as the feature level local fusion module, uses the evidence theory as the decision fusion level module, presents the combination arithmetic based on matrix and solves a bottleneck problem of evidence theory in the application. The fault diagnosis system is tested on the hydraulic pump, vibration signal of end cap and pressure signal of exit are selected as two data sources, it is shown from the result-of the experiment: the method is able to conquer the shortcoming of neural network and solve the abuse of evidence theory which is difficult to gain the basic probability assignment functions, the system improves the reliability and decreases the uncertainty of the diagnosis.
Hydraulic pump Time-domain parameters Multi-data fusion D-S evidence theory Fault diagnosis
Shengqiang Wu Wanlu Jiang
College of Mechanical Engineering, Yanshan University, Qinhuangdao, Heibei, 066004, China
国际会议
杭州
英文
947-951
2009-04-08（万方平台首次上网日期，不代表论文的发表时间）