Study On Compressed Sensing Method Based On The Active Sonar Signal Of Prior Information

Active sonar signal is one of the effective ways to recognize underwater target.The active sonar transmits the signal,after underwater target scattering,the device receives the echo signal,and then processes the echo signal to recognize the target.The underwater environment is complex,there are all kinds of interference noise,it is difficult to process the echo signal with noise.In order to improve the detection accuracy of the active sonar system,it is necessary to increase the bandwidth of the transmitted signal.The increase of signal bandwidth means the sharp increase of data,which it is a burden to the whole system of signal acquisition,storage,transmission and processing.In order to solve this problem,the paper researches on the compressed sensing processing method based on the prior information of active sonar signal.To solve the problems of echo signal reconstruction accuracy and large amount of sampling data when echo signal with a lot of noise interference.Firstly,introduced the background and significance of underwater active sonar signal and compressed sensing theory,and the research status at home and abroad is briefly summarized.From the principle of active sonar device,the paper introduces the incident signal and target echo signal of active sonar,and constructs the echo model of the target signal.To the active sonar signal,digout the prior informations from the active sonar signal process,it provides the prior information theoretical support to the following chapters.Secondly,to solve the problem of low reconstruction accuracy when underwater echo signal with low SNR.Using the prior information of the incident signal as an "atom" to build an over complete atom library and Combining the "block" sparse characteristics of underwater target echo signal.A compressed sensing algorithm based on time domain prior information is proposed.Dealing with the simulation echo signal of active sonar and real measurement data on the lake to verify the correctness and performance of the algorithm.Then,considering the engineering realization,constructing the undersampling measurement matrix based on undersampling theory and Time domain prior compression sensing method.Combining the matrix with time domain prior compression sensing method to deal with simulated echo signal.Using undersampling system to get the undersampling date which compression ratio is 20% and 10%.The experimental results of signal reconstruction with the algorithm show that the methodhas practical engineering significance.Finally,based on Bayesian theory,integrating statistical prior information into compressed perception theory.A statistical prior compressed sensing method is proposed.Dealing with the simulated echo signal,to build a hierarchical probability model based on the statistical prior information of the echo signal.Then obtain the posterior distribution of the model’s "super parameters".Combining with sparse Bayesian learning iteration to reconstruct the original signal.