Research On Sound Quality And Noise Source Characteristics Of High-power Tractors

The road conditions of tractors are complicated,and the noise problem is more obvious than other vehicles,especially for high-power tractors.To study the sound quality and noise source characteristics of tractors is not only helpful to improve the driver’s ride comfort experience,but also of great significance to improve the international competitiveness of tractors.This paper takes a high-power tractor as the research object.Based on the subjective and objective evaluation test of the noise in common working conditions of tractors,model s for predicting the sound quality of tractors were established.The main noise sources of tractors were identified and separated by independent component analysis.On the basis of sound quality value and noise source frequency domain characteristics,the noise reduction strategy and noise analysis software for tractor driver’s ear noise were studied and explored.Based on the above research ideas,relevant research work was summarized as follows:(1)Subjective and objective evaluation of tractor noise.The multi-channel noise acquisition program was designed by Lab VIEW software,and the noise data in common working conditions of tractors were collected,and 42 groups of tractor noi se evaluation samples were obtained by interception.Spearman correlation coefficient and K-means clustering analysis were used to test the evaluation results,and 42 groups of subjective and objective evaluation data of tractor noise were obtained.(2)The prediction model of tractor sound quality is established.Taking the subjective and objective evaluation data as training samples,the prediction model of tractor sound quality is established by using the methods of multiple linear regression(MLR),back propagation(BP)neural network and support vector regression(SVR).The global optimization feature of genetic algorithm(GA)optimizes the key parameters of BP neural network and SVR algorithm.The sound quality prediction models of GA-BP neural network and GA-SVR are established,and the prediction results of the models are compared through multiple evaluation indicators.(3)Tractor noise source separation and frequency domain characteristic analysis.Fast independent component analysis(FICA)algorithm was used to identify and separate the noise source of the tractor’s noise signal,and obtain the independent component signals of the fan,engine,muffler,mechanical transmission and cab.The contribution of the five independent components to the driver’s ear noise is calculated through the mixing matrix,and the signal energy of the main noise source and the energy contribution of bark band are calculated under the conditions of road driving and field operation.The frequency characteristics of the main noise sources of the tractor under different working conditions are obtained.(4)Based on the App Designer platform of MATLAB,the tractor sound quality and sound source characteristic analysis software was developed.The software includes login interfa ce,sound acquisition and playback interface,psychoacoustic parameter interface,sound quality evaluation interface,sound source identification and feature interface.(5)Tractor noise optimization test.Based on the calculation result of the contribution degree of the noise around the ear,the sound quality evaluation result is used as the reference standard,and the noise reduction target of the driver’s position is used.The passive and active noise reduction schemes of the tractor ware designed respectively,and the noise reduction effect of the tractor under different working conditions was analyzed.