Study And Application On Sound Quality Of Road Noise Of Passenger Vehicle

In the past 10 years,China automobile industry has developed rapidly,and consumer’s demands increase significantly.Sound quality of road noise is one of the most important driving performance,which is treated very seriously by scientific research institutes and automobile manufacturers.At present,the problems on sound quality of road noise research are:(1)It is difficult for automobile research and development institutions to grasp customer’s needs,accurately at the pre-research stage,and there are always showed that some market problems,that the huge deviate distance exists between the needs of customers and the target of research and development department.(2)In engineering fields,there is also lack of effective means to evaluate the sound quality of road noise,so it is really hard to judge whether the NVH performance of a product meets customer’s expectations,which have wasted much time and labor.Based on the actual needs of road noise development,this thesis collected and compiled the objective test data of road noise from 31 vehicles with different manufacturers,and organized 63 persons to conduct subjective evaluation.Combined with BP neural network Algorithm theory model,to predict the sound quality of road noise.The model with a three-layer structure that contains: The input layer,hidden layer and the output layer.Using the SPL,sharpness,roughness,articulation index,and loudness as five input factors in input layer.Then the input layer factors will be processed by hidden layer,at last the output value will be predicted.This model can be used to solve time and labor wasting problems which are usually caused by nobody make decision when facing complexed conditions.In the aspect of model validation,this thesis uses the subjective results and also the objective test results,which were tested and collected from other four cars.The verification results show that the model can accurately predict the customer’s subjective perception,and the error between the prediction result and the actual result is no more than 0.5(10-point scale).In the aspect of engineering application,to solve the problem of terrible subjective feeling of road noise,which occurred on a rough asphalt road with the speed of 60 km /h,a road noise quality prediction model was rebuilt in this thesis.First,the model was used to confirm the target of vehicle NVH performance,and then the model was used to confirm that which frequency was the root cause frequency of the problem.and then after the useful engineering strategy were applicated,The road noise quality prediction model were used to confirm whether it achieved the target.As a result,the score of subjective evaluation was improved from 4.5 to 6.5,which verified the engineering applicational value of road noise subjective prediction model.