Hardware Design Of Multi-beam Imaging Sonar Data Processing System Based On Embedded GPU

Since the proposal of the Ocean Power Strategy,the field of marine exploration has received increasing attention.As one of the key technologies of marine exploration,multi-beam imaging sonar has multiple uses such as sounding,obstacle avoidance,and positioning.It has broad development prospects in the areas of underwater target positioning,marine emergency rescue and salvage.For this reason,this thesis studies and develops the multi-beam imaging sonar data processing system based on embedded GPU.The system takes embedded GPU module Jetson TX1 as the core.As the GPU cooperates with high-performance FPGA,the system can realize synchonous acquisition,transmission and data processing of 400 acoustic signals from four line-arrays.The system includes two parts:data receiving subsystem and signal processing subsystem.First,the analog input signal conditioning and signal synchronous sampling is realized by the data receiving system.The data receiving system then performs DFT calculations through Spartan-6 FPGA and the data is transfered to the signal processing subsystem in parallel mode.The signal processing subsystem realizes the reception and buffering of DFT data through Artix-7 FPGA.Next,the data is uploaded to Jetson TX1 module through the PCIe bus to complete the beamforming calculation and cross-spectrum azimuth estimation.Finally,the system sends the result to the host computer.By testing the designed hardware system,all Indicators are qualified.In the lake test,the system can display 2D underwater acoustic images in real-time,locate and annotate 3D coordinate data of the target.The system has the advantages of clear imaging,strong real-time performance,wide detection range,and high flexibility,which meets design requirements.