Research On Prediction And Control Technology Of Vehicle Interior Sound Quality

With the continuous development of society and economy, vehicle interiornoise gradually become one of the most important indicators to measure thequality of vehicles, and it is become an important factor to evaluate and buying acar. Reducing interior noise to increase acoustic comfort has also become animportant part of the automotive product development design, and the relatedresearches become the hotspot in automotive industry. At present, A-weightedsound level is taken as standard and optimization objective in vehicles interiornoise evaluation study. However, A sound level can not fully reflect the physicalcharacteristics of interior noise and people subjective feeling, simply reduce thecar sound level A method has been unable to meet the demand of passengers.Consumers began to pay attention to sound quality. So creating a comfortable interior acoustic environment and taking sound quality as vehicles interioroptimized design target become an important research and future developmentdirection.Works in this thesis are based on the projec（t20100361）form Jilin ProvinceScience and Technology Development. In the paper, steady-state interior noisewas focused as research targets, subjective evaluation test was performed forannoyance grade. Based on the subjective and objective evaluation data, a soundquality evaluation model was established by neural network. By weight analysisand model reconstruction, the objective parameters that have importantimplications to subjective evaluation results were found out. Combine thetransmission path synthesis technology with sound quality evaluation mode, avirtual car sound quality equivalent model is established. In the model, mainnoise source that have important influence on sound quality can be identified bytransfer path analysis. In the model, sound quality simulation was preformed by modify the transmission characteristic of the primary path. With the analysisovercome, improvement was applied on original car, and achieved good results.The main contents are as follows:Firstly, subjective evaluation test was carried out and the objectiveparameters were calculated for interior noise of the steady-state operatingconditions based on the theories of vehicle noise and psychoacoustics. Theobjective parameters as input, the subjective value as output, a GA-BP neuralnetwork was adopted to establish a sound quality evaluation model. Comparedwith BP model, the model convergence rate nearly three times the averageprediction error is reduced by3.29%, and gave good predictions of highcorrelation (0.928) and low error (±8%). It proves that the proposed model hashigh accuracy and effectiveness. Weight analysis and reconstruction modelshowed that loudness, sharpness and roughness are the most influentialparameters in vehicle interior sound quality, other objective parameters have little effect.Next, with the composing analysis of interior noise, combining the interiornoise model, a vehicle interior noise simulation model based on transfer path analysis is established. In order to improve model accuracy, the m sequencecorrelative identification method of airborne noise transfer function wasproposed, and matrix inverse method was used to calculate excitation ofstructure-born noise. The model prediction spectrum is same as experiments data,the error of A-weighted sound level is less than2.0%, the error of loudnessis less than6.8%, the relative error of sharpness is4.7%, the relative error ofroughness is less than6.0%. This shows that the established synthesis model hashigh prediction accuracy, the synthetic interior noise can reflect the actual carnoise characteristics. Combine noise synthetic model with sound qualityevaluation model, a virtual car sound quality equivalent model was established.In the model, the sound quality as a result of the output and the final destination, the interior noise synthesis model extends to a prediction of sound quality,objective parameters can be calculated and directly to obtain annoyance values.Then, the contribution of each transmission path is calculated, main noisesource transfer path is identified as rear suspension. In order to simulate designthe sound quality, transfer characteristics of engine rear mounting Z andX-direction paths that have great influence in sound quality are modified inprediction model. The simulation results show that, just take effective noisereduction measures in main noise source path can get good sound qualityimprovement effect. These works provide a guide for sound qualityimprovement.Finally, the technique of vibration isolation and decrease interior noise areresearch and some situations are applied on real vehicle. The body mountbracket system is optimized to improve stiffness and vibration resistance byfinite element dynamic analysis method. Real vehicle tests show that, applying the improvement on original car, the sound quality is enhanced, andcompared with foreign car the gap is less than0.5. The result verify theeffectiveness of the optimization measures.In this thesis, a sound quality prediction model and a virtual car soundquality equivalent model are established. The research gives an integratedtechnique of vehicle sound quality prediction, evaluation and control and have ahigh theoretical significance and a good prospect.