Research And Implementation Of Sound Source Location Algorithm Based On Microphone Arrays

Microphone array has functions of echancing and locating the sound source signal.Sound source localization is a main function of the microphone array.The main content of this thesis includes the introduction of related theoretical knowledge,the introduction and improvement of Generalized Inverse Beamforming algorithm,and the combination of the Generalized Inverse Beamforming algorithm and the compressed sensing theory to propose a sound source localization algorithm based on the sparse representation of eigenmode.Finally,a complete hardware platform is designed.The main work and research results of this thesis is as follows:Firstly,introducing the knowledge related to array signal processing,and introduces several classic positioning algorithms and source number estimation algorithms in array signal processing.It also studies the theory of compressive sensing,focusing on the signal sparse model and sparse reconstruction algorithm.It lays the foundation for the subsequent research based on Generalized Inverse Beamforming algorithm and compressed sensing theory.Secondly,studing the Generalized Inverse Beamforming algorithm.The core idea of the algorithm is to reduce the number of possible locations of the source through iteration,and finally obtain the location of the source.In order to enlarge the difference in sound intensity between the grid points with and without the source,this research proposes a weighted Generalized Inverse Beamforming algorithm.The core idea of this algorithm is to construct a weighted matrix to expand the difference in sound intensity,and extended this algorithm to make it suitable for broadband sources.This thesis analyzes the algorithm flow of the Generalized Inverse Beamforming,which does not consider the results of the last iteration.Therefore,another improved algorithm is proposed.The core idea of the algorithm is to introduce the result of the last iteration into the current iteration through accumulation.Then,the original algorithm and the two improved algorithms are simulated and analyzed to compare the performance differences between these algorithms.Thirdly,analyzing the spatial sparseness of the sound source signal and establishes a corresponding sparse signal model.This model provides a theoretical basis for estimating the sound source position using compressed sensing theory.After that,several kinds of sound source localization algorithms based on data sparse representation are introduced.The main idea of these algorithms is to use the spatial sparseness of the sound source signal and use the sparse signal reconstruction algorithm to achieve the sound source localization.Fourthly,proposed a sound source localization algorithm based on sparse representation of eigenmode.The main idea of the algorithm is to process the received signals of the array and use eigenmode to represent the source signals.After analysis,the eigenmode is consistent with the source signal and is sparse in the airspace,so the sparse reconstruction algorithm can be used to locate the sound source.At the same time,the selection method of atoms in sparse reconstruction algorithm was improved,the maximum sound intensity criterion was used to select atoms.Then the algorithm is simulated,and the simulation results prove the effectiveness and high accuracy of the algorithm.Finally,a complete hardware platform is designed in this thesis,which can realize data collection,processing and transmission.The corresponding computer is used to process the data in the host computer,analyze the performance of the proposed algorithm,and verify the practicability of the hardware system.