Research On Weak Fault Recognition And Intelligent Diagnosis Of Automotive Transmission

As the core component of automotive transmission system,performance of automotive transmission has direct influence on the servicelife and running condition of automobile, and the faults of automotivetransmission parts, which are often strongly shown as the characteristics ofnonlinear and non-stationary, will inevitably occur caused by its complicatedstructure and poor working conditions. So the method to extract fault featureinformation effectively from complex machinery equipment that similar toautomotive transmission is the key to fault diagnosis, especially in early weakfault diagnosis and forecast, which plays an important role to improve thesafety and economy of automobile mechanical equipment. The main researchachievements are showed as follows:Firstly, on the basis of large number of literature researches and practiceinvestigations, we build the platform of vibration and noise fault diagnosis ofautomotive transmission. After analysis of the common faults of automotivetransmission and the corresponding signal analysis method, the papercontinues data preprocessing through acquisition of vibration signal data bythe experiment of normal and different faults of automotive transmission.Secondly, with the study of weak fault identification based on stochasticresonance(SR), this paper expounds the stochastic resonance mechanism andthe model of monostable stochastic resonance and bistable stochasticresonance, realizing the fault feature effective extraction of automotivetransmission synchronizer combined with the form of cascade and scaletransformation.Thirdly, the method of gear fault feature extraction of automotivetransmission based on Teager-Huang transform is discussed. In the premiseof noise reduction by stochastic resonance, Teager energy operator(TEO)demodulation that combined with the method of empirical mode decomposition(EMD), is applied in the weak fault feature extraction of gearin automotive transmission.Finally, the paper introduced the theory of support vector machine(SVM)and its application in fault intelligent diagnosis. On basis of the crossvalidation(CV) method, the paper respectively uses grid search method andparticle swarm optimization(PSO) to search and optimize importantparameters in SVM model, thus improving the performance of SVMclassifier.