Acoustic Characteristics Extraction Of Underwater Targets Under Continuous Explosion Interference

The underwater continuous explosion has the characteristics of high power,wide frequency band and long duration,and has been widely concerned as an acoustic interference source.Therefore,the study of underwater continuous explosion characteristics is of great significance to improve the anti-jamming ability of mine fuze.In this paper,the underwater explosion signal is taken as the main research object,and the mechanical and acoustic characteristics of underwater explosion are studied by the method of combining theoretical research and simulation,and the shock wave propagation law and acoustic characteristics law of explosion signal with different blast depth and different charge amount are obtained.The machine learning method is used to train underwater explosion signal,ship target signal and their mixed signals,and to realize the separation of explosion signal and target signal.The main work and achievements of this paper are as follows:(1)The underwater explosion model was established with AUTODYN numerical simulation software.In the study of underwater single explosion,firstly,the propagation law of underwater blast wave is analyzed,and then the simulation results are compared with the empirical formula to verify the accuracy of the simulation.By changing the explosion depth and charge amount,the influence law of shock wave peak pressure is obtained.In the study of underwater continuous explosions,by changing the initiation interval time of continuous explosions,the influence law of different interval time on shock wave propagation is obtained.(2)The acoustic characteristics of underwater continuous explosions are obtained by comparing and analyzing the acoustic signals of underwater single explosion and continuous explosion into corresponding SPL and power spectrum.It can be concluded that underwater explosion has strong sound power,high sound pressure level and wide frequency coverage of explosion signal.Bubble pulsation has a great influence on the energy distribution of sound power level.The more bubble pulsation times,the greater the proportion of low-frequency energy.The SPL of the concave point can be increased by decreasing the initiation time interval between the continuous explosion elements.The sound signal with concentrated energy can be obtained by reasonably controlling the detonation interval time.(3)The ship target signal in water is simulated by Matlab,and time-frequency analysis of explosion signal,target ship signal and their mixed signals is carried out by using wavelet transform,and frequency domain characteristics of two kinds of different signals are obtained.It can be concluded that the frequency coverage of the explosion sound signal is very wide,and mainly distributed in the low frequency.Explosion signal and target signal can be roughly distinguished by time-frequency graph.(4)In this paper,a mixed signal recognition model based on the YOLO-v7 algorithm is established.The network can effectively identify the explosion signal and target signal,and the recognition accuracy of the YOLO-v7 algorithm for the mixed signal of the explosion and target is 79.6%.After comparison with SVM algorithm,it can be concluded that traditional machine classification learning methods like SVM are not suitable for distinguishing sudden signals like explosion signals,while deep learning algorithm can achieve better target detection.