The Study On Sound Quality Evaluation Of Fuel Cell Vehicle Cooling Fan Based On BP Neural Network

The development of new energy vehicles changes rapidly.The new energy vehicles gradually become the current development trend of the automotive industry.In addition to energy saving,safety and environmental friendliness,consumers want their cars to be more comfortable and quieter.The concept of sound quality put forward that the sound design of cars and users' needs are closely linked together and is further integrated into the car's various parts and sound design.Using the sound quality as the evaluation standard of sound and establishing quieter and more comfortable sound environment for consumers are the most important part of the study on sound in automobiles today.They will be the future direction of development.As one of the most intuitive indicators of the quality of car,the sound quality of the car deeply influences the purchase behavior of the customers.Fuel cell vehicles which are of high efficiency,low noise and no pollutant emissions,are considered of the ideal direction for the development of the automotive industry.Although the noise level in fuel cell vehicles is lower than in the gasoline cars,the noise of fuel cell vehicles is more complicated.The linearity of noise is not as good as that of conventional internal combustion engines cars.Its sound quality is poor.The cooling demand of fuel cell vehicles is higher than of the traditional car.Cooling fan of fuel cell vehicles is one of the major noise sources.The purpose of this paper is to study the predictive model of sound quality for noise reduction in fuel cell vehicles.In order to evaluate and control the noise quality of cooling fans for fuel cell vehicles more quickly and accurately,BP neural network technology is used to establish the fan sound quality prediction model.First of all,using the fuel cell vehicle cooling fan model as the research object,the paired comparison method is adopted to subjectively evaluate the collected sound samples and analyze the reliability and consistency of the test data of the samples.According to the evaluators' subjective evaluation results,we get the subjective evaluation of each sample.Then,objective parameters of psychoacoustic model were usedto calculate four objective parameters.They are the loudness,sharpness,roughness and fluctuation strength.Finally,using the objective parameters of psychoacoustic as input and the subjective evaluation of the sound quality as output,we establish the prediction model of sound quality based on the BP neural network.