Research On Direction Estimation Based On Vector Hydrophone Array

For a long time,sonar arrays in our country have adopted scalar hydrophone arrays,and a complete sound field contains not only scalar information,but also vector information.The emergence of vector hydrophones has made up for this gap.Vector hydrophones In addition to receiving the sound pressure information of a certain sound field,it can also receive particle vibration velocity information.The rich sound field information provides more possibilities for the(direction of arrival DOA)algorithm.The DOA estimation algorithm has been fully developed in the application field of scalar arrays.Among them,the subspace algorithm is the most representative,which has already achieved super-resolution azimuth estimation,but for vector array signal processing,there are some new features: such as thorough Solve the problem of ambiguity on the port and port,and a larger search range can be obtained under free-field conditions;the inconsistency of the noise power between the sound pressure channel and the vibration velocity channel causes the inaccuracy of the subspace division,which leads to the performance of the algorithm;the vector hydrophone The influence of the sound field scattered by the shell on the performance of the array when the device is installed in the carrier shell.The above-mentioned problems put forward requirements for improvement when the traditional position estimation algorithm is applied to the vector hydrophone array,but also caused researchers to think about the development direction of the new vector hydrophone array position estimation algorithm.Based on the National Natural Science Foundation of China,this paper takes the uniform linear array of vector hydrophones as the research object.Under free-field conditions,the practical application of low signal-to-noise ratio and fast position estimation algorithms for low snapshots are studied.Under the condition of the plate,the azimuth estimation under the unknown frequency of the incoming wave is studied,and the spectral peak search problem in the azimuth estimation is studied.The specific content includes:(1)Under the condition of free field,the smooth L0 algorithm has poor signal reconstruction effect in low-noise environment,and the problem of low success rate of azimuth estimation and slow operation speed is improved,and a faster and more accurate azimuth estimation algorithm is proposed.(2)Under the condition of the baffle,a mathematical model of azimuth estimation based on the reflection coefficient of the baffle is established,and the influence of the baffle scattering sound field on the scalar array and the vector array is respectively explained.Aiming at the problem that the reflection coefficient is affected by the incoming wave frequency,use Comsol and Matlab to obtain training data,use Tensorflow to build a convolutional neural network model,and use network training to extract the orientation information in the array output covariance matrix,and get the unknown incoming wave frequency condition The bearing estimation result below.(3)When there are multiple signal sources in the space,the traditional spectral peak search method is slow,the search tends to fall into the local optimum,and the traditional swarm intelligence algorithm can only converge to the global optimum,so that only one of the azimuths can be found.The existing group intelligence algorithm makes it possible to realize multi-spectral peak search.