Sound Quality Prediction Of Noise At Power Coupling Mechanism Of HEV

At present,as a transitional product,the hybrid electrical vehicle(HEV)occupies a large market in the new energy automotive industry.However,because multiple power sources in the power coupling mechanism and frequently switch of mode,the sound quality of the vehicle has been affected and this has been one of the main topics of NVH in the vehicle industry.The objective evaluation method,the subjective evaluation method and the evaluation model of noise sound quality at power coupling mechanism of HEV are studied in this paper.Complementary ensemble empirical mode decomposition(CEEMD)and the Hilbert transform(HT)are used to extract the objective parameters of the sound samples.The subjective preference evaluation is carried out by the paired comparison method,and the relevance vector machine(RVM)evaluation model is finally established.The noise sound quality at power coupling mechanism can be predicted by this evaluation model.Firstly,the subjective evaluation method of noise sound quality at power coupling mechanism of HEV is researched.Noise signals at the power coupling mechanism under steady and non-steady conditions are collected in the newly built road which has no other vehicles;the sound samples are preprocessed using Artemis;the subjective preference evaluation is carried out by paired comparison method;the correlation analysis of the subjective evaluation results are carried out by the SPSS software.Then five psychoacoustic objective parameters including tonality,fluctuation,loudness,roughness and sharpness of each sound sample are calculated using Artemis;the correlation between subjective preference and each psychoacoustic objective parameter is calculated.Then,an objective evaluation method of noise sound quality at power coupling mechanism of HEV is proposed.Based on the basic theory research of empirical mode decomposition(EMD),ensemble empirical mode decomposition(EEMD)and HT,a set of simulation signals are used to prove that CEEMD can effectively reduce the modal mixing of EMD and the reconstruction error of EEMD;a set of simulation signals are used to prove that HT is superior to the continuous wavelet transform(CWT)for the time or frequency location in time-frequency joint domain;the collected sound samples are carried out filtering and CEEMD in the objective evaluation method,then the instantaneous frequency of each intrinsic modal function(IMF)after decomposition is calculated by HT,and then critical frequency band weighting is carried out.The critical frequency band weighted energy values are used as the new objective parameters for the sound quality evaluation.Finally,the evaluation model of noise sound quality at the power coupling mechanism of HEV is studied.Based on the proposed objective parameters,three sound quality evaluation models including multiple linear regression(MLR),back propagation(BP)neural network and RVM are established,and then the prediction results are compared;two RVM sound quality evaluation models are established based on the psychoacoustic objective parameters and the proposed objective parameters respectively,and then the prediction results are compared.The final results show that the sound quality evaluation model based on CEEMD-HT and RVM has higher prediction accuracy and can more fully reflect the characteristics of the noise signals at power coupling mechanism of HEV,and are more corresponded to subjective perception of the evaluators.