Research On The Evaluation Of Interior Sound Quality Of Pure Electric Vehicle

With the scarcity of fossil energy and environmental pollution becoming increasingly serious,the new energy vehicles represented by pure electric vehicles have been gradually promoted in the world.However,compared with fuel vehicles,pure electric vehicles lack the masking effect of traditional power system noise,which increases the high-frequency squeal noise of the motor.The sound quality in the vehicle is completely different from that of traditional fuel vehicles.With the development of people’s material living standard to a new stage,the evaluation of the interior sound quality of passenger cars has become a hot research topic.In addition,while the pure electric vehicle industry in the ascendant,considering that its sound quality characteristics are quite different from traditional fuel vehicles and there is no relevant evaluation standard as a reference,it is of great significance to evaluate and analyze the interior sound quality of pure electric vehicles and establish an objective acoustic parameter prediction model of the sound quality.The methods of interior sound quality evaluation mainly include subjective evaluation test and establishment of prediction model of sound quality.The subjective evaluation test can more intuitively reflect the subjective feelings of the driver on the noise environment in the car.Then simulating the subjective feelings of the passengers by using multiple objective acoustic parameters,and further establishing the objective acoustic parameter prediction model of the sound quality in the car,is a widely accepted method of quantifing vehicle sound quality.In this paper,the vehicle interior noise of three different positioning pure electric vehicles(Tesla Model s,Chery e Q and Great Wall C30 EV)under the condition of constant speed and full throttle acceleration is taken as the research object,and the sound quality evaluation and analysis are carried out systematically.In the subjective evaluation,considering the age structure and gender characteristics of reviewers,a fuzzy anchored semantic differential method(FASDM)is proposed based on interval grey number theory and anchored semantic differential method.Compared with the traditional semantic differential method and anchored semantic differential method,the accuracy and practicability of the improved method are verified,and the preferences of different age and gender groups for different acoustic quality indicators of pure electric vehicle interior noise are analyzed.In the process of establishing the prediction model of sound quality,the objective acoustic parameters such as A-level,loudness,sharpness and their change rate of noise samples are extracted as training and testing data.According to the results and rules summarized from subjective evaluation,the sound quality prediction models are established respectively based on Back-Propagation(BP)neural network,Radial Basis Function(RBF)neural network and RBF improved by Particle Swarm Optimization(PSO)algorithm.Then the following results can be obtained.Compared with the traditional semantic differential method and the anchored semantic differential method,the FASDM based on the interval grey number theory can more accurately reflect people’s subjective feelings of the interior noise under the premise of maintaining the same workload.Under the condition of constant speed,the correlation between the pleasant degree index of interior noise quality and its tonality or fluctuation strength of pure electric vehicle is much greater than that of traditional fuel vehicle.Different age groups and gender groups have different preferences for the interior sound quality of pure electric vehicles.The prediction model which takes the age and gender characteristics into account is more accurate and practical.Compared with the prediction model based on BP neural network and non-optimized RBF neural network,the generalization ability of RBF neural network optimized by PSO algorithm to the interior noise of pure electric vehicle is more stable and accurate.