Localization and characterization of noise sources in enclosed space by microphone array

Abstract

Sound source localization has been an important research topic with applications in many fields. During decades, an enormous variety of localization techniques have been developed using microphone arrays including the Time Difference of Arrival (TDOA) method, the beamforming technique, the Near-field Acoustic Holography (NAH), and the Time Reversal Acoustics (TRA). However, the localization of sound sources in enclosures with reverberation and background noise is still a challenging problem. In this study, firstly, an approach for multiple sound sources localization and characterization based on the time-domain beamforming is developed. Main sources in the sound field are localized and the contribution of each source is reconstructed in a sequence. The reconstruction of the sound sources are further refined with a correction procedure. The performance of this approach has been verified numerically and experimentally in the free-field condition and in a moderate reverberant environment. Thereafter, the problem of sound source localization in reverberant environment is studied and a dereverberation method based on the beamforming technique and statistical characteristics of the signals is proposed. In order to eliminate the reverberation effects, multiple sub-arrays are formed by extracting certain number of microphones from the whole microphone array. The probability distribution of the sound source position is then estimated according to each beamforming map of these sub-arrays. The combination of the sub-array average and the weighting of beamforming power standard deviation is then used to estimate the probability distribution of the source location. The dereverberation ability of this approach is validated through numerical simulation and experiment verification. Moreover, a structure failure sound source localization system is developed based on the time difference of arrival method. This system localizes the structural crack which radiates impulsive sound wave by using four microphone in each side of a cubic. In order to increase accuracy and eliminate the interference effect by background noise, the low-quefrency filtering technique is adopted. This has been validated by a series of experiments which includes experimental rig and satellite model.