High-speed train running gear fault recognition based on information fusion of multi-resources
ZHU Jian-qu1,2, JIN Wei-dong1, ZHENG Gao3, ZHU Bin1,4
1.School of Electric Engineering, Southwest Jiaotong University, Chengdu 610031;2. Chongqing University of Science and Technology, Chongqing 401331;3. Department of Electromechanical Management, China Maritime Police Academy, Ningbo, Zhejiang 315801;4. Changjiang Normal University, Chongqing 408100
Abstract:To solve the problem that it is difficult to identify the fault of high-speed train running gear caused by multiple detection points and large amounts of data, a running gear fault recognition method with multi-feature and multi-source information fusion is proposed based on the fuzzy evidence theory. Firstly, the information fusion degrees of sensors are obtained by calculating the difference of membership degrees which can reflect a certain feature of different sensors belonging to different fault modes. The fusion degrees are used to determine the weights of different sensors during fusion, thus the membership degrees between different sensors with similar characteristics can be obtained after information fusion. Then, the basic probability assignment functions are converted from the fused membership degrees. Finally, the information of different characteristics is fused on the basis of evidence theory. Experimental results show that four cases, including the normal, loss air of air spring, without anti-yaw shock absorber and without transverse shock absorber, can be identified effectively, and the satisfactory fault identification rates are obtained at different speeds, all these verify the validity of new method.