Research On Technologies Of Underwater Unmanned Swarm Robust Detection Using Sparse Representation

With the steady implementation of the Ocean Power and Smart Ocean strategies,underwater acoustic monitoring is gradually moving towards three-dimensional data acquisition and intelligent information processing.Underwater unmanned swarm distributed cooperative underwater acoustic detection system,which is mounted on multi-autonomous underwater vehicles,is popular for underwater monitoring in complex environments due to its flexible working mode and fast and comprehensive information acquisition.At the same time,the broad application prospect puts higher demands on the reliability and robustness of underwater unmanned swarm detection technology.However,in the active detection posture of multiple transmitting and receiving static nodes,the active transmitting platforms of underwater unmanned swarm are poorly concealed and easily exposed;the transmitting signals from multiple active platforms suffer from acoustic interferences;and the multipath interference,Gaussian and impulsive noise interference of complex oceanic channels seriously affect the time delay estimation accuracy of the received signals.This thesis aims to enhance the stealthiness and improve the target detection performance of underwater unmanned swarm,and investigates the robust detection techniques of underwater unmanned swarm from three aspects:biological transmitting signal modelling,channel blind deconvolution techniques and environmental decoupling time delay estimation.Based on the sparse representation theory,the theoretical framework of parameter estimation based on linear regression model is established.Due to the sparsity of underwater acoustic signals in different observation "domains",the matched sparse representation model is introduced to solve the specific physical problems,and a variety of parameter estimation algorithms and signal processing approaches for underwater unmanned swarm scenarios are proposed.In order to solve the problems of poor concealment and easy exposure of underwater unmanned swarm active platform,this thesis proposes a biological multi-static transmitting signal model based on LASSO and block-LASSO by imitating dolphin whistles with the idea of bionic.The generated biological multi-static transmitting signal is very similar to the dolphin whistle signal in the time-frequency structure,and the active covert detection of underwater unmanned swarm can be realized by using the biological environmental noise in the ocean.In addition,in order to solve the problem that the actual dolphin whistles data are interfered by impulsive noise,a biological multi-static transmitting signal model based on robust penalty function is proposed to realize the robust estimation of dolphin whistles parameters in the environment of impulsive noise.Finally,through the correlation coefficient analysis of instantaneous frequency curve,it is proved that the biological multi-static transmitting signal has strong concealment,platform resolution and good detection performance.In order to solve the problem of acoustic interference of multi-platform signals emitted by underwater unmanned swarm,the sound mechanism of dolphins in swarm state was studied at the transmitter end,and the code multiplexing of code division channel was realized by using biological multi-static with different simultaneous frequency structure codes.At the receiving end,a kind of unmanned underwater swarm based on Toeplitz matrix channel blind robust deconvolution technique is proposed.The technique transforms double-convex problem into simple convex problem,refactoring emission signal waveform and the generalized channel impulse response,and then complete the received signal source identification,realizes the unmanned underwater cluster bumpless robust detection.In order to solve the dictionary mismatch problem in parameter estimation,a dictionary updating strategy based on a wideband integrated dictionary is proposed,which can improve the reconstruction accuracy of both the transmitted signal and the generalized channel impulse response in continuous parameter space,and further strengthen the robustness of the blind channel deconvolution algorithm.In addition,the proposed robust blind deconvolution technique for underwater unmanned channel has no limitation on the transmitting waveform and has wide applicability.Aiming at the problem that multi-path interference of complex ocean channels seriously affects the detection performance of underwater unmanned swarm,a target echo delay estimation technology based on blind deconvolution and weighted iteration strategy is proposed in this paper.Firstly,blind deconvolution model is used to realize the identification of multisource target echoes.Then the weighted coefficient matrix is introduced,and by applying different weight coefficients to the time delay estimation,the main path is highlighted,the interference of multi-path channel is overcome,and the arrival time of target echo is decoupled.In addition,a weighted iterative delay estimation technique based on robust penalty function is proposed for the impulse noise interference in the target detection process,which can achieve accurate estimation of target echo delay in complex multipath channels under impulse noise background.The effectiveness of the proposed algorithm is verified by numerical simulation and experimental data processing.It provides a new signal processing approach for multi-platform transmitting signal design,blind estimation of underwater acoustic channel and time delay estimation of target echo in complex multi-path channel.