Internal Sound Quality Evaluation And Electromagnetic Noise Analysis Of Electric Vehicle

Electric vehicles are driven by motors and electromagnetic noise is one of the unique NVH characteristics of electric vehicles.Aiming at the problem of the interior sound quality of electric vehicles at medium and low speed,taking a domestic brand electric vehicle as an example,according to the characteristics of the interior motor electromagnetic noise,the sound quality evaluation method of electric vehicles was studied.The overall noise in the vehicle was evaluated subjectively and objectively.In terms of subjective evaluation,the grade scoring method was used to evaluate the interior noise subjectively,and the subjective score was used as the dependent variable of the prediction model.In terms of objective evaluation,the Articulation Index,A-weighting sound pressure level,roughness,loudness and sharpness of the overall noise were calculated to provide independent variables for the prediction model.Based on the subjective and objective evaluation of the overall noise in the car,considering the prominent characteristics of high-frequency electromagnetic noise of electric vehicles,the electromagnetic noise of the motor is analyzed and studied.The electromagnetic noise is obtained by order analysis method,and the sharpness,sound pressure level and loudness of the electromagnetic noise were calculated to characterize the characteristics of the electromagnetic noise of the motor.The objective parameters of the overall noise and electromagnetic noise were used as the independent variables of the prediction model.With the subjective score as the dependent variable,and the objective parameters of the overall noise and the electromagnetic noise of the motor as the independent variable,a multivariate nonlinear regression prediction model was constructed,and compared with the multiple linear regression prediction model and XGBoost prediction model.The results show that the multiple nonlinear regression model has the best prediction performance among the test samples,followed by the XGBoost prediction model.The mean absolute error and root mean square error of the multivariate nonlinear regression model were reduced by 21.2% and 24.2%,respectively,compared with the XGBoost model,which performed well in terms of prediction accuracy.In the multivariate nonlinear regression prediction model,the increase of the sharpness of electromagnetic noise will reduce the subjective feeling of the vehicle noise and enhance the influence of Articulation Index on the subjective feeling,which can provide some reference for the research and optimization of sound quality.