Research On Direction Finding Technology Of Embedded Planar Microphone Array

Relying on the National Natural Science Foundation of China(61401203)and the Fundamental Research Funds of the Central Universities(30918012203),this thesis carry out research on the direction of arrival(DOA)estimation of embedded planar microphone arrays.However,the outer casing blocks part of the array elements of the embedded array,causing error.In order to optimize the embedded array miniaturization design and improve the DOA estimation accuracy,this thesis studies the influence of embedded installation on DOA estimation and algorithm improvement.From the study of the embedded array TDOA and DOA,the diffraction,refraction and reflection were analyzed to establish a model.Based on the above analysis,the model-based and data-based DOA estimation algorithm are proposed.Firstly,the angle compensation is adopted,but considering that the DOA estimation error is not a one-to-one mapping,we give a model-based DOA algorithm and propose to decompose the complex array into a rectangle and an arc.In addition,we limit the search range by combining the conventional direction finding results.Experiments show that the method significantly reduces the DOA estimation error and time.We also propose a data-based DOA estimation method that attempts to achieve accurate direction finding based on line-of-sight(LOS)TDOA.The existing literature often screens the ideal TDOA through the cost function.Based on the characteristics of the embedded array,we propose a DOA estimation algorithm for screening the LOS microphone.Among them,the algorithm using the DOA estimation result algorithm has a non-one-to-one mapping problem.We solve the problem by using diffraction attenuation and TOA,and further improve the accuracy.Considering the selection method to discard rather than correct TDOA,the DOA estimation method using low-rank and geometric constraint is proposed.Experiments show that the low-rank constraint suppresses the random noise in TDOA,the geometric constraint suppresses the diffraction propagation deviation,and the DOA estimation method based on low-rank and geometric constraint significantly reduces the direction finding error,which has broad prospects.