Vector Array Direction-of-arrival Estimation Algorithm Based On Sparse Reconstruction

Direction-of-Arrival estimation is one of the important contents of underwater active target detection,which is the prerequisite and foundation for underwater target positioning and tracking,and has important research significance.Underwater active target direction estimation mainly faces two problems:firstly,the active echo signal at long distances is weak,the SNR is low,and there is very little available target information;Secondly,during close range and covert detection,the short pulse width detection signal results in less available echo snapshots.Aiming at the above problems of low SNR and low snapshot,the research on high-performance bearing estimation with low SNR and low snapshot is carried out based on vector hydrophone and sparse reconstruction algorithm,as follows:Firstly,based on the vector spatial array model,the estimation performance of convex optimization and sparse Bayesian algorithms in low SNR and low snapshot scenarios was studied.The simulation results of the sparse Bayesian algorithm under the Nehorai processing framework were presented and compared with traditional subspace class methods.Secondly,based on the uncorrelated properties of isotropic noise in the sound pressure and vibration velocity channels,the article proposes a method for constructing isotropic noise fields based on numerical simulation,the processing advantages of the combined sound pressure and vibration velocity processing compared to the Nehorai framework were analyzed.By formally transforming the complete dictionary set in the spatial domain,the equation of the covariance domain sparse Bayesian(p v_c-SBLalgorithm based on joint processing is derived,and simulation comparative research is conducted.The results indicate that due to the joint processing gain ofpv_c,pv_c-SBLcan achieve better estimation performance under low snapshot and low SNR.Then,using the principle of frequency division,the method is extended to the off grid scenario of broadband received signals,and a multi vector observation model(MMV that can process coherent signals is introduced.The off grid sparse Bayesian direction estimation method for broadband signal joint processing under the MMV model is presented(p v_c-w PSBL.The p v_c-w PSBLalgorithm has been verified through simulation to have good estimation performance under low SNR,low snapshot,and coherent source conditions,and has good correction ability for quantization errors caused by different spatial sampling intervals.Once again,in response to the computational efficiency shortcomings of sparse Bayesian algorithms,a beam domain sparse Bayesian method based on preprocessing is proposed based on the above research.The computational complexity analysis and simulation experiments have shown that the proposed method can ensure good estimation performance while reducing the complexity and average running time of the algorithm.Finally,the performance of the proposed algorithm was verified through lake trial data processing.