Interception Detection And Parameter Estimation Based On Time-frequency Image Of Underwater Acoustic Pulse Signal

The interception and parameter estimation of non-cooperative underwater acoustic pulse signal is one of the key techniques of underwater acoustic information countermeasure.Due to the complexity of ocean background noise and sonar signal,the interception of non-cooperative underwater acoustic pulse signal and parameter estimation are becoming more and more challenging.Because acoustic pulses are mostly non-stationary,and the signal form and parameter are very diverse,so,in this paper,based on the time-frequency distribution of the signal,the detection and parameter estimation of the non-cooperative underwater acoustic pulse signal are studied.The main work of this paper is as follows:For the problem that quadratic time-frequency distribution will produce serious crossterms,an adaptive time-frequency analysis method based on feature parameter extraction in ambiguity domain is proposed.Based on the analysis of the ambiguity domain characteristics of the underwater acoustic pulse signal,the parameters of the kernel function are adaptively adjusted by estimating the radial angle and radial length of the water acoustic pulse signal in the ambiguity domain,the optimal matching between the kernel function and the signal is to retain the auto-terms of the signal to the maximum extent and to suppress the cross-terms to the maximum extent.Then the obtained kernel function is used to analyze the time-frequency of the acoustic impulse signal.From the perspective of objective evaluation,the experimental results of the simulated and measured data are obtained.Compared with other time-frequency distribution methods,the time-frequency distribution of the acoustic pulse signal obtained by this method can effectively improve the time-frequency resolution and weaken the cross-term interference.Aiming at the problem that time-frequency image is interfered by ocean background noise,a time-frequency image denoising method based on the combination of background equalization and adaptive sparse representation is studied.This method firstly carries out background equalization for time-frequency distribution of signals,and then uses the adaptive sparse representation method based on over-complete dictionary to carry out denoising for timefrequency images.Experimental results show that this method can effectively improve the PSNR of time-frequency image by at least 10 dB.By using the above methods,time-frequency images of underwater acoustic pulse signals with high time-frequency resolution and weak cross-terms and little interference from background noise can be obtained.In this way,detection probability can be effectively improved when intercepting and detecting underwater acoustic pulse signals based on timefrequency images.In order to extract instantaneous frequency curves from time-frequency images of underwater acoustic pulse signals.We study a blind separation algorithm which can automatically classify and merge the sub-components of the signal.The method extracts the instantaneous frequency curve from the perspective of the energy of the time-frequency distribution.In order to better suppress the interference of reverberation and other factors,this paper designs an algorithm to estimate the instantaneous frequency from the perspective of image——the instantaneous frequency curve reconstruction algorithm based on neighborhood set.Simulation experiments and sea trial data processing results show that the method can effectively extract the instantaneous frequency curve of underwater acoustic pulse signal.