Research On Condition Evaluation And Fault Diagnosis Of Locomotive Wheelset Based On Information Fusion

In recent years,China’s heavy-haul and high-speed railways have developed rapidly,and the problems of dynamics and vibration have become more serious.In the railway industry,the hidden danger brought by the larger vibration to the safety of locomotive operation is also extremely serious.The small one will cause the loss of signal of locomotive control line,the large one will cause the crack and fall off of locomotive running parts,and even may cause the locomotive to fall off and overturn.The state of wheelset represents the size of vibration,so the railway department attaches great importance to the state monitoring and fault diagnosis of locomotive wheelset.In view of the state monitoring and fault diagnosis of wheelset,according to different detection principles,several detection systems have been developed.However,due to the influence of external environmental factors,the accuracy of wheelset state monitoring and fault diagnosis is sometimes difficult to be guaranteed for different detection systems.On this basis,this paper compares and analyzes the advantages and disadvantages of locomotive wheelset status monitoring and fault diagnosis system used in railway industry,and puts forward a wheelset comprehensive evaluation method based on multi-sensor information fusion technology.Based on locomotive wheel of the fault as the research object,the first wheel to the development of fault diagnosis technology are introduced,with the development trend of condition monitoring and fault diagnosis technology are analyzed,at the same time,the application of information fusion technology to develop a simple introduction,and the tread wear,abrasion,stripping and polygon round of common faults are analyzed.Next to the railway industry of the three kinds of locomotive wheel used for condition monitoring and fault diagnosis system principle and technical characteristics are analyzed in comparison,and studied the information fusion technology,conventional algorithms for their contrast analysis of the information fusion principle,technical features,current mature theory is presented in this paper,the application range of the BP neural network algorithm.Then the data analysis process and key points of locomotive wheelset on site are investigated.According to the field investigation and the principle of establishing evaluation index system,the evaluation index system of locomotive wheelset state is constructed by selecting evaluation indexes of locomotive wheelset.Finally,according to the wheelset state evaluation index system established above,sample data are selected in combination with actual cases on site,and the network junction of the neural network model is determined after processing the sample data according to relevant experience and formulas Combined with the training parameters,the comprehensive evaluation model of locomotive wheelset state based on multi-sensor information fusion was constructed and trained.The comprehensive evaluation model of locomotive wheelset state was verified and analyzed with the actual case data as test samples.Finally,by comparing the results of manual analysis and model evaluation,the feasibility and effectiveness of the locomotive wheelset state evaluation model constructed in this paper based on information fusion is verified,which can improve the on-site judgment and disposal of locomotive wheelset state and effectively play a role in maintenance and safe operation.