Research On Fault Noise Detection Method Of Electric Vehicle Gear

Empirical mode decomposition（EMD）is an effective method for gear fault diagnosis under unsteady conditions,but for noise under different loads,the signal decomposition and feature extraction of EMD are not ideal.The accurate extraction of gear noise signal is a prerequisite to improve the rate of gear fault diagnosis.In view of the above problems,this paper takes the gears and noise signals of electric vehicles as the research object.And carries out research on the method of extracting the optimal feature vector of gear noise signals by optimizing the parameters of variational modal decomposition through ant lion optimization algorithm,aiming at improving the diagnosis rate of gear faults of electric vehicles under varying conditions.The mechanism of gear meshing noise in electric vehicle was analyzed theoretically,the dynamics model of gear pair was established,and the eigenmodal components of noise signals of different gear faults were extracted.A method based on ant-lion optimization variational modal decomposition combined with energy entropy and generalized neural network was proposed.Firstly,the Intrinsic Mode Functions（IMF）are used to decompose the noise signals of five kinds of faulty gears with different loads using the method of Variational Mode Decomposition（VMD）.IMF component is the finite bandwidth component of different central frequencies such as gear meshing noise,fault noise and mechanical natural frequency.Secondly,in order to avoid the over-decomposition and mode mixing of gear noise signal caused by incorrect values of intrinsic modal components K and penalty factor α,ant-Lion Optimization algorithm（ALO）was introduced to optimize the parameters of variational modal decomposition.The global search capability and adaptive weight change of ant Lion optimization algorithm were utilized to obtain the optimal parameter combination [K,α],and the feature extraction of the optimal modal IMF component was carried out.Finally,the Generalized Regression Neural Network（GRNN）is introduced to realize fault type recognition based on the feature vector data extracted by the above method.In this paper,based on the basic theoretical analysis,combined with simulation experiment,gear noise module experiment and gear noise comprehensive experiment,the fault diagnosis of five kinds of faulty gear noise of electric vehicle with different damage degree is carried out.Through training and testing,noise feature extraction and fault diagnosis of gears under different working conditions are realized,and the feasibility and effectiveness of this method are verified.