Sonar Echo Simulation Technology Based on Array Phase Weight Estimation

Abstract

Sonar locates underwater targets by receiving reflected sound waves. However, the complex marine environment makes it difficult to set targets in appropriate locations and conditions. The sonar echo simulator has the function of simulating sonar detection target echo signals. As a cutting-edge technology, underwater backscatter has led to the emergence of array based acoustic reflection systems. The research on sonar echo simulators based on backscatter technology has promoted the solution of problems such as target echo modeling, sonar arrival direction estimation, and echo directional transmission. In response to the above problems, this paper designs an end-to-end sonar echo simulator system array phase estimation sonar echo simulation system (APE-SESS), which can independently complete high-resolution real-time direction of arrival (DOA) estimation and generate directional simulation echo based on the array structure. Dual branch convolutional neural network (DB-CNN) is proposed in the system to estimate the direction of the signal array and directly obtain the phase weights containing azimuth information. Comparing DB-CNN with conventional methods and classic underwater DOA network models based on classification problems, the results show that DB-CNN exhibits stability, small error, and high real-time performance under different signal-to-noise ratio (SNR). The proposed APE-SESS has end-to-end characteristics, real-time angle estimation, and azimuth simulation functions.