Study Of Fault Diagnosis System Of Gearbox

It is very difficult to solve the problem of gearbox for the variability of influencingfactors and the diversity of performing behavior. The theoretical part of the gearbox faultdiagnosis involves multiple disciplines, and it is a cross-disciplinary. The fault diagnosis ofgearbox synthetically uses automation, machinery, mathematics, computers and otherdisciplines of knowledge. Gearbox fault diagnosis encounters some problems in development.How to accurately find out the fault features, which are extracted from complex gearboxrunning status information? How to improve the reliability and validity of the fault diagnosis?Firstly, the overall design of the gearbox fault diagnosis system was made. Gearbox faultdiagnosis system was a multi-task information processing system with the functions of statedetection, data storage, working conditions analysis and fault diagnosis. The diagnosticstrategies were flexible and elastic. It used conventional fault diagnosis subsystem for thesimple and common fault diagnosis and intelligent fault diagnosis subsystem withaccumulating the knowledge base.Secondly, the thesis established a model of the gearbox fault diagnosis in theory. Itanalyzed the gearbox vibration mechanism and the vibration signal components, which laidthe foundation for the further research.Thirdly, the conventional diagnostic means was analyzed theoretically. Time domainanalysis, frequency domain analysis and demodulation analysis were introduced in this part.The defects of these diagnostic means were analyzed in theory. It was seen that conventionaldiagnostic means were usually used for simple gear box fault. Simulation was made inconventional means, wavelet analysis and neural network fault diagnosis on gearbox fault.The conventional diagnostic means, which had many shortcomings, had been limited in someapplications. We could check the result from the comparison of the simulation. The thesisused incompact and compact wavelet network synthetically. The frequency energydecomposed by wavelet packet was used as the entry of network. It was passed to the networkto solve the problem of gearbox. Then BP algorithm was used.Through the combination ofwavelet analysis and neural network, the better results for gearbox fault diagnosis were found.The results of wavelet network training and pattern recognition were good. It was found thatthe neural network diagnosis performed better.At last, the gearbox fault diagnosis system had come true. Gearbox fault diagnosis usedconventional analysis and intelligence analysis. In the gearbox fault diagnosis system, the main components were signal analysis, the conventional fault diagnosis and intelligent faultdiagnosis. For different test function, the software would call the corresponding functionmodule according to the specific requirements automatically. The modules called each otherin the way of exchanging news. The practice of use proved that this system was easy tooperate and friendly in human-computer interaction. With the test of actual gearbox fault data,it had been proved that this system’s performance was good. The gearbox fault diagnosissystem could step into the actual use and had the value of popularization.