Research On Intelligent Processing Technology Of Ship Seismic Wave Signal

This thesis mainly focuses on the research of intelligent processing technology of ship seismic wave signal.Ship seismic wave signal is a signal containing multiple frequencies.In its propagation process,high frequency signals are naturally filtered out,and very low frequency signals are transmitted over a long distance.Its main components include direct waves,longitudinal waves,transverse waves and Scholte waves.They are all mixed together in the signal acquisition process,and the effect of ordinary signal separation and identification methods is poor.This thesis proposes a method to separate the collected ship seismic wave signals by fitting them into ship seismic wave image signals according to the law of propagation.At the same time,the BP neural network with momentum item is used to identify signals,which provides a new method for underwater target detection and recognition.According to the characteristics of small attenuation of very low frequency signal propagation in ship seismic wave signals,two vibration sensors separated by a certain distance are used to collect them.For the interference signal mixed in the acquisition process,Butterworth low-pass filter is used to filter the acquired signal.The filtered signal is simulated to synthesize the ship seismic wave image signal according to the propagation law.The linear Radon transform algorithm is used to intelligently separate and extract the fitted ship seismic wave image signals,transform the image signal from t-x domain to ?-p domain,and eliminate other interference waves according to the p value corresponding to each wave.The simulation proves that the method of fitting images can better extract direct waves,longitudinal waves,transverse waves and Scholte waves.According to the same method,several ship seismic wave signals and non-ship seismic wave signals are simulated and their corresponding components are extracted.The extracted components are used as training samples to train the BP neural network model that drives the quantity item.A part of training samples is randomly selected as test samples,and the trained neural network model is used for intelligent recognition.Through the analysis of simulation data,the recognition rate of ship seismic wave signals can reach 86.7%,which shows that the BP neural network driving the quantity terms has a good recognition effect on ship seismic wave signals.In this thesis,the experimental data are collected by simulating the ship in a site similar to the shallow sea environment,and the data are processed by using the method of fitting image and BP neural network with the momentum term.The experimental results show that the method proposed in this thesis can realize the intelligent processing of ship seismic wave signal.