Research On Analysis And Evaluation Of Vehicle Interior Sound Quality For Pure Electric Vehicle

With the continuous development of new energy automotive industry all over theworld, China is also starting to focus on the development and research of new energyvehicles.As an important type of new energy vehicles, pure electric vehicles havegradually become a hot research at home and abroad.But the research on pure electricvehicle has mainly focused on power systems and battery systems, paid less attentionto the noise characteristics.The study on sound quality of pure electric vehicles is notdocumented,so it is necessary to conduct in-depth research.The topic comes from the major projects of the2012Jilin Province Science andTechnology Development Program-the research on analysis,evaluation and controltechnology for FAW pure electric vehicles.We put pure electric vehicles byindependent research and manufacture as object,and mainly study the noise frommotor system of pure electric vehicles.By collection and processing of the motor noisesamples,we conducted a subjective evaluation test on sound quality and analyzed thetest results.Through the study of neural networks and genetic algorithms,weestablished a prediction model based on genetic algorithm to optimize BP neuralnetwork of vehicle interior sound quality for pure electric vehicles.We finally identifythe main factors which have main impact on the vehicle interior sound quality forpure electric vehicles, and help to further control and improve the sound quality.Thisstudy will have important practical value and significance on improvement of acousticdesign and noise control for pure electric vehicles and enhancement of our electricvehicle industry competitiveness in international markets.The main contents of this article is divided into two parts:(1) Collect noise samples of vehicle interior,conduct objective/subjectiveevaluation and research on sound quality.Through the semi-anechoic laboratory,we carry through sample collection ofvehicle interior noise for four different models of pure electric vehicles on the rotating hub tester.The test uses digital artificial head to collect noise signal on five uniformconditions of a vehicle in20km/h,40km/h,80km/h,100km/h,120km/h.Thenwe began to subjective evaluation test of sound quality,using level score method toscore the irritability on the sound samples of pure electric vehicle interior noise.UsingArtemis sound quality software to calculate the psychoacoustic objective parametersof noise samples, including loudness, sharpness, fluctuation, roughness.Finally wemake correlation analysis of the value of subjective evaluation and objectiveparameters, and select the psychoacoustic parameters with a larger correlation withthe subjective evaluation value.(2) Establish a prediction model of vehicle interior sound quality for pure electricvehicle based on neural network.First, the structure of the BP neural network model,transfer functions andalgorithm principles are described.Then we introduce the related concepts, processesand operations of GA-Genetic Algorithm.By combination with Genetic Algorithm andBP algorithms to establish the GA-BP neural network model.The introduction ofsubjective and objective evaluation of test results to determine the structure of BPneural network for vehicle interior sound quality. Set the number of network layersand neurons, and Determine the transfer function.Using genetic algorithm to optimizethe network established.Finally, using Matlab software to establish a prediction modelbased on GA-BP neural network of vehicle interior sound quality(annoyance) for pureelectric vehicles.Through training and testing the model, fully prove the convergence,stability and prediction accuracy of the model.Besides,we take the measures to controlvehicle interior sound quality for pure electric vehicles.Finally,we verify the effect ofimproving sound quality through the prediction model based on GA-BP neuralnetwork and subjective evaluation test.