Research On The DOA Estimation And Localization Technology To Acoustic Target For MEMS Vector Hydrophone Array

The MEMS technology is a new attempt to design hydrophone. MEMS vectorhydrophone is a novel sensor researched independently by the North University of China,which has the advantages of small size, vector, lot manufacturing with well consistent, lowcost and so on. With the growing maturity of production process and performance, it is veryimportant for engineering applications that the research on the DOA estimation andlocalization technology to acoustic target for MEMS vector hydrophone array.In this thesis, we studied the signal processing for acoustic vector sensor array. Throughmathematical modeling,theoretical analysis,algorithm simulation and experimental dataprocessing of MEMS vector hydrophone array and so on, we test the performance of differentalgorithms and the engineering practicability of MEMS vector hydrophone. The mainresearch results are follows.(1)According to Gerschgorin disks theory, we propose the new signal numberestimation algorithm(GDE-V) and modified form(MGDE-V) for acoustic vector sensorarray, and given the criterion to detect the number of signals, in which the covariance matrixhas been adjusted for the coherent signals. Simulation experiments show that the GDE-Vmethod and MGDE-V method can efficiently exert the advantages of vector sensor array, andthey have better estimation performance for the port/starboard signals detecting, coherentsignals detecting, low SNR and few snapshot, and can distinguish more signals than acousticpressure sensor array with the same array aperture size.(2)Through re-construction for the covariance matrix of acoustic vector sensor array andself-adaptive selecting for the lead orientation, we propose a new root-MUSIC algorithm andits real-valued form for acoustic vector sensor array. The theoretical derivation and simulationexperiments show that two new algorithms have better DOA estimation performance in lowSNR and few snapshot than traditional root-MUSIC algorithm for acoustic pressure sensorarray. In the lake trials of MEMS vector hydrophone array, the azimuth of acoustic source hasbeen accurately estimated and the running tracks of moto boats have been successfullyfollowed by two new algorithms.(3)We propose a novel MUSIC algorithm with pressure and particle velocity combinedprocessing(PV-MUSIC) for acoustic vector sensor array, and make DOA estimation byeigen-decomposing for the cross-covariance matrix which is obtained by selecting for the lead orientation and projecting for the particle velocity of acoustic vector sensor. The simulationexperiments show the better DOA estimation performance of PV-MUSIC algorithm thantraditional MUSIC algorithm in isotropic noise field, and the PV-MUSIC algorithm has beensuccessfully application in lake trials of MEMS vector hydrophone array.(4)According to the characteristics of two-dimensional MEMS vector hydrophone, wederivate the expression of various kinds of array errors, and build the vector array signalmodel under the influence of the error, further proposes self-calibration algorithm of acousticvector array error, and verify the algorithm by the simulation experiment and the measureddata in lake trials of MEMS vector hydrophone array.(5)To solve the DOA estimation in non-uniform vector sensor array, we derivate thetime delay expression of four non-uniform vector sensor array and the array direction vector,and proposed the MUSIC algorithm for non-uniform vector array. Through selecting for thearray with optimization performance, we take statistic to the change of probability of successand RMSE with SNR and snapshots, simulate experiment shows that the performance ofnon-uniform line array is best, it has better DOA estimation performance in low SNR and fewsnapshots.(6)Though the experimental data of MEMS vector hydrophone array in differentenvironments, we analysis and discuss about data preprocessing, measuring for environmentalnoise, the correlation of pressure and particle velocity, the gain of vector sensor array, DOAestimation and so on, and verify the feasibility of the proposed algorithms in this thesis, andprovide technical support for the engineering application of MEMS vector hydrophone.The main innovations in the paper are follows.(1)We propose the Gerschgorin disks algorithm estimating the signal number, which cangreatly improve the estimation performance for acoustic vector sensor array.(2)We propose the Root-MUSIC algorithm and its real-valued form, which can improvethe DOA estimation accuracy and effectively reduce the computational complexity, and hasbeen successfully applied in the experiment of MEMS vector hydrophone array.(3)We proposes the self-calibration algorithm of MEMS vector hydrophone arrayerror,which can improve the engineering practicability of various algortithms.The engineering application of MEMS vector hydrophone is a huge systematic project,there are still a lot of work to research, we hope the work in this thesis are helpful to theresearch and development of acoustic vector array signal processing technology, and promotethe MEMS vector hydrophone more extensive engineering application.