Research On Key Techniques For Multibeam Echo Sounder In The Unmanned Platform

With the continuous development of various surface and underwater unmanned sports platforms,multi-beam detection technology based on unmanned platform has unique advantages in the fields of marine surveying and mapping,marine mineral mining,fishery survey,post-disaster rescue and military,especially in harsh and dangerous environments,and has become an important development direction of marine detection technology.However,in the unmanned platform,due to the limitation of hardware scale and the inability to manually monitor the beam data conditions and control parameters in real time,the measurement accuracy and outliers of terrain results are usually difficult to guarantee.At the same time,the diverse detection tasks of the MBES of the unmanned platform,such as pipeline tracking monitoring,underwater sediment detection and platform terrain-aided navigation,have improved data accuracy and ability to suppress outliers.Therefore,how to obtain finer,more accurate and cleaner result data has become a key issue for MBES on unmanned platforms.This paper conducts an in-depth study on how to solve the problem of data quality from the main signal processing flow of MBES.According to the characteristics of near bottom fine measurement task of MBES on unmanned platforms and the demand for higher signal-to-noise ratio of beam data to improve the quality of subsequent detection results,a dynamic background suppression and highresolution direction estimation techniques is proposed.First,the azimuth spectrum of the Multiple Signal Classification(MUSIC)algorithm is further processed with the deconvolution algorithm.Based on the output azimuth spectrum of the point sound source in SNR 0d B of MUSIC,an adaptive point scattering function is constructed through the eigenvalues of each moment.The algorithm could basically keep the algorithm results of the consistent array gain under different signal-to-noise ratios.Afterwards,the simulation and experimental data processing are compared,and the effectiveness of the algorithm is proved.To avoid the false targets caused by unmanned monitoring beam data and adjusting system parameters,a two-dimensional(2D)double selectivity index-constant false alarm rate(DSICFAR)algorihm is proposed.By fitting the real beam data with various distributions,the algorithm accurately establishes the clutter power level model with exponential distribution,which simplifies the clutter power level estimation process and ensures the detection performance of DSI-CFAR.At the same time,a two-dimensional cross sliding window including reference cells and guarding cells is adopted.This method improves the detection performance by adopting appropriate clutter power level estimation strategies in different directions.To reduce the amount of calculation of the algorithm and improve engineering practice,this paper adopts a fast region of interest(ROI)selection algorithm based on global threshold.The DSI-CFAR detector performs only in the region of interest,which significantly reduces the number of pixels that must be calculated.Finally,the algorithm is verified by simulation and measured data,which proves that the algorithm can maintain a lower false alarm rate on the premise of the same detection rate.Even if better-performing DOA and detection algorithms can be used,outliers in MBES data cannot be completely avoided due to the limitations of various factors such as sea conditions,acoustic echo intensity of the subfloor,severe terrain fluctuations,and water propagation interference,and the manual control threshold and post-processing manual erasure mode in the onboard mode can not meet the task requirements of the unmanned platform.Aiming at this problem,a multi-beam inter-frame terrain robust tracking technology is proposed.Firstly,by simplifying the terrain tracking state model,the multi-beam terrain tracking is modeled as a single target tracking problem in a single beam fixed direction,avoiding the nonlinear error caused by the polar coordinate transformation of terrain data.At the same time,to solve the common non-uniform observation noise in multi-beam data,especially the influence of cluster noise on tracking results,a robust tracking algorithm combining Support Vector Machine(SVM)and M estimation theory is used to realize multi-beam Accurate tracking of underwater terrain greatly improves the quality of the final output terrain data.The simulation results show that the algorithm can better eliminate the influence of outliers and output terrain results accurately.The algorithm is verified by multi-beam measured data.The output terrain is not only not affected by isolated outliers,but also maintains an accurate and good tracking effect for the continuous outliers in the edge beam area.The above introduces and analyzes the main signal processing flow based on the unmanned platform from three aspects: DOA azimuth estimation,terrain constant false alarm detection and terrain tracking,and proves these algorithms can improve the final data quality to a certain extent.To verify the validity and advancement of the signal processing of the overall algorithm flow,three processing algorithms are formed into a new multi-beam signal processing flow,and the simulation and experimental data are used to verify it.The Ultrasound Toolbox(USTB)is used to invert the echo signals of each channel through the 3D terrain point cloud model established in the previous stage,and the corresponding channel original signals are obtained according to the preset multi-beam system parameters.At the same time,the experimental data is obtained by the MBES carried by the AUV "Hai Xun No.1".Using different signal processing procedures to process simulation and experimental data,it is proved that the new processing procedure proposed in this paper can greatly improve the quality of terrain data and improve the operation capability of MBES under unmanned platforms.