Research On Noise Source Identification And Sound Quality Evaluation Method Of Diesel Engine

With the development of automobile industry and the improvement of consumer requirements,the sound quality(SQ)of vehicles and engines has become one of the key performance affecting the market competitiveness.The acoustic research of heavy-duty diesel engine in commercial vehicle has gradually changed from the initial low-noise design to the development of high-level SQ.This paper mainly focuses on the adaptive noise sources separation,rapid evaluation and prediction of SQ in multiple scenes,extraction and diagnosis of “fidgety” features,the main work contents are as follows:According to the characteristics of sound and vibration signals of diesel engine,a parameter adaptive noise source separation method was proposed.In this approach,the weighted fuzzy distribution entropy was constructed to optimize the Variational Mode Decomposition(VMD)to adaptively obtain the optimal parameters,which considers the mutual information between decomposition components and the original signal.The accuracy and robustness of Adaptive VMD(AVMD)method were verified by simulated signals.According to the Intrinsic Mode Function(IMF)of noise signal,combined with time-frequency analysis and partial coherence analysis(PCA)to conduct the noise source separation of diesel engine,and the accuracy of the separation results was verified by motored test and acoustic package,so as to provide guidance for noise source separation of power machinery.The diesel engine has many noise sources,and the sound energy and frequency distribution have high correlation with the operating conditions and measuring points.In this study,the influencing factors such as operating conditions,test points and frequency band distribution were comprehensively considered,and the analytic hierarchy process tree was constructed,a noise source contribution analysis method based on fuzzy comprehensive evaluation method was proposed.In the method,the noise frequency band can be adaptively divided by AVMD,according to the coupling relationship between typical components vibration and engine noise,and the noise spectrum results of different measuring points,the fuzzy consistency matrix and combined weight matrix were constructed based on fuzzy analytic hierarchy process(FAHP),it effectively solves the problem of consistency of judgment matrix,and the noise contribution of different typical components for diesel engine can be obtained,which can provide guidance for the acoustic development.Signal decomposition technology has good extraction effect on low-mid frequency noise,and the characteristics of high-frequency abnormal sound need to be further diagnosed combined with the SQ subjective evaluation method.The SQ of diesel engine varies with scenario and operating condition,it is difficult to quickly evaluate the acoustic level of diesel engine in different application scenarios.In this study,a quick SQ evaluation approach was proposed based on the DNA construction of scenario and operating condition.In this approach,the effectiveness and diversity of sampling were considered to reduce the test scale,the revised grouped pair comparison(RGPC)was developed to improve the evaluation accuracy and efficiency.Finally,the weights of operating conditions in different application scenarios were calculated,and the weighted preference was employed as a comprehensive SQ index.The construction of DNA condition can provide guidance for the SQ rapid evaluation and market positioning in different application scenarios.Current,most of the research on SQ prediction only quantifies the objective parameters from the time domain,which ignores the frequency domain characteristics of the SQ objective parameters,it is difficult to effectively characterize the feature of whistling and piercing.In this study,the frequency domain derived parameters of TNR and PR were constructed,and the objective evaluation system of engine SQ was improved from the perspective of time-frequency domain.The SQ prediction model was constructed by improved SVM algorithm,which can lay a foundation for the extraction and quantification of “irritability” features for diesel engine.The extraction of SQ “fidgety” characteristics is the key to acoustic improvement.However,the generation mechanism of “fidgety” characteristics is unknown,it is difficult to directly guide the SQ improvement.DNA condition samples were selected as the research object,referring to the spectrum distribution of physical acoustics and psychoacoustics parameters,the effect of frequency domain on subjective preference can be obtained by the SQ prediction model,an extraction method of “fidgety” feature was proposed in the paper.Besides,the acoustic diagnosis was conducted by means of noise source identification.The results show that the abnormal noise mainly comes from the turbocharger BPF noise.Through the parameter optimization design of turbocharger trim and large/small diameter of impeller,the compressor BPF noise was reduced by 8.5d B and the engine noise was reduced by 1.2d B.After the subjective jury test of the optimized engine noise,the abnormal sound characteristics of the engine basically disappeared and the SQ was greatly improved.