Study Of Sound Quality Characteristic And Evaluation Models Of HEV

Comparing with Internal Combustion Engine Vehicles(ICEV),Hybrid Electric Vehicle(HEV)which supplemented with battery packs,motors and other components has more complex structures and operating conditions.As a result,the vibration and noise sources and their characteristics have changed greatly.Therefore,sound quality of interior noise does not get improvement.So this project mainly aims at analyzing the sound quality characteristics of the interior noise in steady conditions and unsteady conditions of HEV;Studying the subjective and objective evaluation methods of the sound quality of HEV to find out the key factors on the sound quality of the vehicle;And establishing a sound quality evaluation prediction model which are suitable for steady conditions and unsteady conditions to predict the sound quality of the interior noise of HEV.Firstly,the power system’s basic structure and several common driving modes are introduced,and the sensing characteristics and structural characteristics of the human ear hearing system are also analyzed.Then,taking a HEV as the test vehicle,the noise signals of vehicle in steady and unsteady driving conditions were collected on the test road.By using reference semantic subdivision method,subjective evaluation tests on the vehicle interior sound quality annoyance index of steady condition and unsteady condition noise signals were performed.The SPSS software was used to analyze the subjective evaluation results,and the unqualified test data was eliminated.And noise signal annoyance values under steady conditions and unsteady conditions were obtained.Several objective parameters of the noise signal under steady conditions and unsteady conditions,including loudness,sharpness,roughness,fluctuation,tone,AI index,sound pressure level,and A-weighted sound pressure,were calculated with Artemis software.The correlation between objective parameters and subjective annoyance was analyzed using SPSS software.Then use 4 models which are Extreme Learning Machine(ELM),BP(Back Propagation,BP)neural network,Support Vector Machine(SVM),and Least Square Support Vector Machine(LSSVM)to establish prediction models based on sound quality annoyance index under steady conditions and unsteady conditions.The results show that the datasets of the four evaluation models under steady conditions show better prediction performance compared with unsteady conditions,which has more poor prediction effect.Then,a feature parameter based on CEEMD and sample entropy is proposed to improve the prediction performance of models under unsteady conditions.The CEEMD is used to decompose the noise signal under unsteady conditions,and the Intrinsic Mode Function(IMF)of the signal is obtained.The correlation between the IMFs and the original signal is calculated by SPSS software and the sample entropy of these IMFs are calculated as the characteristic parameter of the noise signal under the unsteady conditions.As the input of the evaluation model,the prediction model of the sound quality annoyance index under the unsteady condition is established.Compared with the prediction model based on psychoacoustic parameters as input of the models,the results show that the prediction model with sample entropy as the model input has better prediction performance and can be used to predict the sound quality of HEV under unsteady conditions.It also shows that the sample entropy value can reflect the characteristics of the noise signal under unsteady conditions more specifically,and is more suitable for the subjective feeling of the evaluators.