Bistatic Sonar Detection And Localization Under Strong Direct Blast Interference

Bistatic sonar has important applications in underwater acoustic detection,but it also faces the problem of strong direct blast interference,which causes conventional bistatic sonar detection technology to form a "detection blind zone" near the transceiver line.Studying the method of bistatic sonar target detection and localization under strong direct blast interference can effectively fill the "detection blind zone",improve the performance and adaptability of the bistatic sonar system.It is of great significance to improve the detection and early warning capabilities of underwater targets.Based on the existing research and work on bistatic sonar detection methods at home and abroad,bistatic sonar target detection and positioning methods under strong direct wave interference are proposed.The related methods have been verified by numerical simulation or experimental data.The main contents and conclusions of this article are as follows:1.Established a bistatic sonar model.According to the application scenario of the bistatic sonar system and the underwater multipath environment,a mathematical model of the bistatic sonar receiving signal is established.The geometrical relationship between the target,the transmitter and receiver of the bistatic sonar is analyzed.The scattering function of the ellipsoidal target at different incident and outgoing angles is analyzed.The Bellhop ray model is used to simulate the acoustic transmission channel,and the numerical simulation results are given.The signal model and simulation method provide the basis for the subsequent research and numerical simulation of detection and localization methods.2.Aiming at the signal model in the underwater acoustic multipath environment,a generalized likelihood ratio test method in the frequency domain of the received signal is proposed for target detection,and the maximum likelihood estimation of unknown parameters in the signal model is given.The test statistic and the closed-form solution of the probability distribution of the test statistic under two conditions of known noise variance and unknown noise variance are derived respectively.On this basis,the detection threshold of generalized likelihood ratio method is obtained,and constant false alarm rate detection is realized.The influence of parameters on the performance of the method is analyzed.Numerical simulation results show that the method can achieve target detection in the dead zone of bistatic sonar under the conditions of low signal-to-noise ratio and low signal-to-direct blast ratio,which confirming the effectiveness of the method.3.Aiming at the aberration caused by the interference of the target scattered signal and the direct blast,an unsupervised learning method based on the idea of anomaly detection is proposed to achieve target detection.This method has the characteristics of being data-driven,irrelevant to the environment and bistatic configuration,and adaptable.Combined with experimental data,the performance of post-processing detection and real-time detection of blind zone target is analyzed.The detection performance of array beamforming and single-array elements,target position distance and signal-to-noise ratio are discussed separately,which confirms the effectiveness of the method.4.Aiming at the problem of target localization under strong direct wave interference,a target localization method based on neural network is proposed.The mechanism of locating targets in the blind zone and the ambiguity of target position about the midpoint of the sonar baseline are studied.A target position regression model based on a convolutional neural network is established.Data preprocessing,model training and regularization acceleration methods are given.The effect of network layers on performance is analyzed.Numerical simulations show that the method can correctly predict the target location.After good training,the network model with several layers of depth configuration can obtain an mean absolute percentage error of less than 20%,and deepening the network can achieve better positioning performance.