Research On Classification Method Of Submarine Substrate Type Based On Characteristics Of Sonar Wave

With the development and growth of multi-beam systems,multi-beam sonar data has become more and more.In the process of using multi-beam sonar data,multi-beam sounding data packets and multi-beam sonar pseudo-side scan image data packets are obtained.Effective use.At present,there are many work on the study of submarine sediment classification using multi-beam sonar images,but the reception waveform of multi-beam sonar is not yet effectively utilized.In this paper,the multi-beam submarine echo waveform is analyzed and the appropriate waveform feature vector is extracted to study the submarine sediment classification.Firstly,this paper briefly introduces the research background and significance of submarine sediment classification research,combs the research and development status of multi-beam system at home and abroad,and analyzes the research status of multi-beam system for submarine sediment classification and utilization,as well as domestic and international research.The current status of research on seabed classification using waveform features.Secondly,this paper analyzes the source of sea bottom data and data preprocessing.The propagation process of sound waves in seawater is analyzed,and the correlation between seafloor scattering intensity and seafloor sediment is demonstrated.The preservation format of multi-beam data is introduced in detail.XTF,for parsed multi-beam submarine data,The error source of multi-beam data and the propagation process of backscattering intensity of the seabed are analyzed.The analysis and correction of incident angle,propagation loss and sound line propagation are made in this paper.Thirdly,this paper extracts the features of multi-beam sonar submarine waveform data,mainly the feature extraction methods in time domain and frequency domain.In this paper,by analyzing the shape of the sonar time domain waveforms of different substrates,a seven-dimensional eigenvector is proposed,which can effectively describe the waveform structure of the seafloor echo.In the frequency domain analysis,according to the seafloor echo,the Fourier transform is excluded,and the wavelet packet decomposition and HHT transform are used to extract the features of the seabed echo.This paper analyzes the waveform by wavelet packet decomposition and extracts the waveform.The wavelet packet energy is used as the eigenvector.In order to analyze the characteristics on the instantaneous frequency,the original waveform information is analyzed by Hilbert Huang transform.After simulation,the extracted frequency domain features can effectively reflect the three-dimensional variation of the time-frequency domain of the waveform.Finally,this paper carries out the design work of the submarine sediment classifier,and summarizes the extracted features to form the input vector of the classifier.Firstly,the K-means is used to analyze the feature space,and then the support vector machine and neural network are used to study the submarine sediment.Classification,compared with the previous bottom quality results,the correct rate is as high as 82%,which has the use value,which proves the feasibility of using multi-beam echo for seabed classification..