Noise Field Reconstruction In Complex Environment Based On Nonlinear Programming

The inter noise of aircraft,automobile,high-speed train and other vehicles seriously affects the comfort of riding.A method of noise filed reconstruction in complex environment is needed to build a test platform,where tests can be carried out economically and efficiently to control the internal noise.There are two type approaches to reproduce sound field: noise test and audio sound field reproduction.Noise test is a reproduction of the target noise field in frequency domain,and at present there are three kinds of noise test: reverberation filed test,direct field test and traveling wave test.Audio field reconstruction is a reconstruction of the target noise field in time domain,which includes three main methods: wave filed synthesis,HOA and the least squares method.In this paper,the frequency domain acoustic field reproduction and the least square method are combined to present a new method for the reconstruction of noise filed in complex environment.This method can be used for noise filed reconstruction in complex environment,direct field acoustic test control and frequency domain simulation of various background noise to test the performance of products.First of all,the description model of the acoustic system in complex environment is studied in this paper.To avoid the difficulty of accurately measure the acoustic transfer function and deal with its ill-conditioned matrix,a new mathematical model is proposed in the form of energy.The driving signal of each loudspeaker and its response signals at each microphone in noise filed are regarded as random signals,and the superposition relation of signals power is derived by analyzing random signals,and then a definition of gain matrix is proposed to establish the mathematical model of acoustic system with noise filed in complex environment.As the mathematical module of acoustic system,the gain matrix can predict the response power spectrum according to the given excitation power spectrum,and the frequency resolution of gain matrix can be choose flexibly to describe the acoustic system with different precision.Secondly,according to the established acoustic model and the target sound field,the driving signals of each loudspeaker are calculated to realize the sound field reproduction.In response to the unreasonable negative value of driving power spectrums resulted by acoustic environment mismatch,nonlinear programming method is introduced to calculate the driving signals whose power spectrums is bound within a reasonable range.The square sum of sound field reproduction errors of each channel is taken as an objective function,and the target noise field reproduction in complex environment is realized by iterative calculation of driving signals.Thirdly,the over-error problem in noise filed reproduction is analyzed and the control scheme is optimized.The relative error of sound field reproduction of each channel is introduced into the objective function as a weighting coefficient to equalize the error between each channel to avoid a large error in one channel.Gain matrixes of different resolutions are flexibly matched together,for different resolutions correspond to different control accuracy of the driving power spectrums and different calculation difficulty.Finally,three sets of noise filed reproduction tests are carried out to verify the effectiveness of this method in complex environment.The test platform is a half-plane model made with true proportions.Before the tests,the required driving level is estimated by pre-test,then the gain matrix is measured.In the first test,the target filed is measured form the test platform to verify the effect of noise reproduction under the same acoustic environment,while the last two tests are carried to verify under different acoustic environment using real sound field date measured from aircraft in flight.