Research on Electric-vehicle Status Recognition Based on Sound and Vibration Signals
At present,electric vehicle accidents always attract a lot of attention and discussion from the public.This pa-per designs a recognition scheme based on sound and vibration signals to judge the main status of electric vehicles,which can provide data support for vehicle conditions,driving habits,accident identification,etc.In the aspect of signal acquisition,a device for real-time acquisition of sound signals and vibration signals is designed and developed.In the aspect of state rec-ognition,a method of wavelet packet energy entropy and improved Grey Wolf Optimization(GWO)to optimize the parame-ters of Support Vector Machine(SVM)is studied to find the best penalty coefficient and kernel function of SVM,so as to de-termine the state recognition model of the SVM.Finally,the experimental tests of electric vehicles in different states are per-formed.It shows that compared with the empirical mode decomposition(EMD),the variational mode decomposition(VMD),the grid search algorithm(GridSearch)and the gray wolf search algorithm,The method of optimizing support vector machine based on wavelet packet energy entropy and improved Grey Wolf Optimization has the advantages of high accuracy and strong stability for electric vehicle operation status recognition.
vibration and waveelectric vehiclestatus recognitionsupport vector machine
万方数据
维普
