Research On Localization Of Aeroacoustic Noise Based On Compressed Sensing

This paper mainly works on the research of aeroacoutic noise localization based on compressed sensing.As the number of civil aviation airliners and the density of aircraft movements increase,the contradiction between aircraft noise and environmental requirements becomes prominent.Due to the complicated geometry shape,strong perceived noise is gener-ated by landing gears during departures and landings,which will bring trouble to the nomal work and rest of residents near the airport.Therefore,landing gear noise has become a world wide issue.The most commonly used methods for aeroacoustic noise research nowadays in-clude calculation methods and microphone array based methods.The main drawbacks of cal-culation method are tests need to be done for calibration,models are overly simplified and computationally inefficient.As to microphone array based method,beamformers are mainly used to obtain source map.However,one or more poblems like low spatial resolution,com-putational inefficient and some assumpation need to be made are suffered by beamformers.In order to overcome the drawbacks of current methods,a compressed sensing based method for aeroacoustic noise localization is proposed.This method,which is mainly based on the spatial sparsity of noise,applies compressed sensing to aeroacoustic noise localization.Simulation and aeroacoutic noise experiments are conducted for narrowband and wideband noise localization respectively.Firstly,common geometry used for planar microphone array are analyzed.Based on far-field delay and sum beamforming method,spatial resolution and max sidelobe level of microphone arrays are studied,which will lay foundation for the selection of array geometry.Secondly,observation model is built for narrow band aeroacoustic noise localization.?₁-minimization method and?₁-Singular Value Decomposition??₁-SVD?are also introduced.The performance of?₁-SVD method in noise localization is analyzed and compared with Conventional Beamforming method?CBF?and?₁-minimization method through simulation and aeroacoustic noise experiment.Lastly,based on joint sparsity model,wide band observa-tion model is built.Besides,wide band Basis Pursuit?BP?and Distributed Compressed Sens-ing-Simultaneous Orthogonal Matching Pursuit?DCS-SOMP?algorithm are introduced and alalyzed through simulation and experiment.Seen from the simulation and experiment results,?₁-SVD method is found to be robust to noise for narrow band noise localization,apart from super resolution.Exact result can al-ways be obtained in aeroacoustic noise localization by?₁-SVD method,as the Restricted Isemetry Property?RIP?holds for the measurement matrix.The parameters needed for?₁-SVD method are also analyzed.It is found that incorrect estimation of the source number does not have much effect on the results.When the singal to noise ratio?SNR?is unknown,promising results can also be obtained by selecting a large value as the estimation of SNR.As to wide band noise localization,computational efficiency is improved by joining narrow band together.DCS-SOMP algorithm is very efficient and is more robust to noise than wide band BP algorithm.The wide band BP algorithm is also bounded by RIP.RIP can hardly be hold for measurements in low frequency range.But,by joining low range and high range together,exact results can also be obtained through wide band BP.DCS-SOMP algorithm is not strictly bounded by RIP,but its performance is not stable in low frequency range.Compared with narrow band localization method,the maximum number of souces that can be exactly ob-tained is improved by wide band BP algorithm.Therefore,compressed sensing based aeroa-coustic noise localization method can overcome the drawbacks of current method efficiently and has a promising application prospect.