Research On Time-frequency Distribution Of Underwater Target Geometrical Scattering Echoes

In the detection and identification of active sonar,highlight model of sound-scattering target is very important.It can reflect the feature of target structure well.The point of this model is how to extract highlight characteristics from target echo signal accurately.When active sonar transmits LFM signals,they can be distinguished on time-frequency domain in ideal situation.But due to the cross-term of highlights' echoes,they are not entirely separated either.This thesis focuses on the time-frequency domain about the suppression method of cross-terms among highlight echoes and the improvement of the aggregation of auto-terms,Which can be used in extracting the highlights characteristics and improving the detecting ability.Firstly,combined with the physical acoustical method,the impulse response function of scattering acoustic field of regular geometry target model is exported.Based on the acoustic-path difference among highlights of the target model and analytical expression of incident direction of launching signal,the angle and distance spectrum among target highlights from different incidence angle was obtained,which could be used as theoretical guidance for the following analysis.Secondly,for the LFM signals which active sonar usually used,the echoes remains the form of LFM signals.however,the echoes overlap in the time-frequency domain and the interference of multi-component cross-term in the time-frequency domain makes the resolving ability decline seriously,which has an influence on the features of highlight echoes.In this research,PHAF(Product High-Order Ambiguity Function)of LFM signals is used as prototype of kernel function of the Cohen method.PHAF can enhance auto-terms of multi-component LFM signals,suppress cross-term and noise,and adaptively estimate the frequency modulation rate of received signals combining the Radon transform of PHAF.So,when used as kernel function,it is effective to suppress the cross-terms of highlight echoes of multi-component LFM signals.Thirdly,for the cross-term interference problem which inevitably exists in bilinear transformation decomposition,the method of linear decomposition of multi-component LFM signals is studied.Chirplet atoms also perfectly match with LFM signal.Therefore,combined with OMP(Orthogonal Matching Pursuit)algorithm and with the Chirplet atom decomposition method,sparse representation of the LFM signals can be obtained,which does not have any cross-term interference.Finally,the experimental data of underwater target detection under free field condition is analyzed with the above two methods respectively.And,the time-frequency characteristics of geometry acoustic scattering highlight of the target is extracted.The effectiveness of the methods in dealing with multi-component LFM signals are compared and verified eventually.