Unsteady Exhaust Noise Sound Quality Prediction And Muffler Structure Optimization

With the progress and development of the times,the research of vehicle sound quality has become one of the main direction in the field of vehicle noise.The exhaust noise is one of the main noise sources of vehicles,and improving the exhaust noise sound quality is of great significance to enhance the NVH performance of the vehicle.Establishing a prediction model which is suitable for unsteady conditions to predict the sound quality of the exhaust noise from domestic mainstream automotive products of SUV,and to predict the sound quality.An optimized engineering design method for muffler to improve the sound quality of unsteady exhaust noise based on this model.The main research contents were as follows:(1)Exhaust noise tests were conducted on mainstream domestic SUV for complete unsteady exhaust noise samples collection.The subjective evaluation test of exhaust noise samples were carried out by the paired comparison method,and got the satisfaction scores of the noise samples.At the same time,the psychoacoustic parameters of unsteady samples were calculated based on the Zwicker time-varying algorithm,and analyzed the association between psychoacoustic parameters and the satisfaction scores.(2)The BP neural network and the Support Vector Regression(SVR)were optimized by Particle swarm optimization(PSO),which were used to establish the sound quality prediction models of unsteady exhaust noise,and the prediction results showed that two models had poor prediction accuracy for unsteady conditions.In view of the lack of prediction accuracy,the Complementary Ensemble Empirical Mode Decomposition(CEEMD)and Sample Entropy were combined to complete the feature extraction of unsteady signals.Principal Component Analysis(PCA)was used to reduced the dimensionality of features and established the sound quality parameters of SQP-CSP(Sound quality parameter base on CEEMD and then proceed SE-PCA).At the same time,the analysis showed that low-frequency noise had a greater impact on sound quality.The roughness and fluctuation which have a poor correlation with subjective satisfaction were replaced by parameters of SQP-CSP.And the parameters of SQP-CSP were combined with loudness,sharpness,A-weighted sound level and kurtosis to be the input of the model,thena model for predicting the sound quality of unsteady exhaust noise based on objective parameters of psychoacoustics and SQP-CSP was proposed.This model had betterprediction accuracy,and it can be used for the evaluation of exhaust sound noise quality in unsteady conditions.(3)The influence of parameters such as sound absorbing material,section shape of muffler,expansion ratio on the low frequency noise was analyzed,and the muffler was optimized.Applying the created automobile exhuast noise sound quality prediction model,and the mulffer optimization and improvement design based on sound quality was performed.After optimization,the sound quality satisfaction was increased by above 10%compared with that before optimization.And the pressure loss of the muffler is not increased.It proved the feasibility of the optimization plan.