High Spatial Resolution Reconstruction of Underwater Acoustic Signal from Drawing Tower Grating with Long Cavity Based on GA-BP

摘要

A method for reconstructing the underwater acoustic signals from a drawing tower grating (DTG) array hydrophone with a cavity length of 50 m is proposed to achieve a spatial resolution of 5 m without changing the hydrophone structure. The genetic algorithm and the backpropagation neural network (GA–BPNN) algorithm are used to create a model to reconstruct 10 sets of underwater acoustic signals with a sensing cavity length of 5 m from a DTG hydrophone with a cavity length of 50 m and achieve a 5 m spatial resolution. Results show that the sensor’s signal with a cavity length of 50 m can reconstruct the signals with a cavity length of 5 m, and the reconstructed signals’ mean absolute error and root mean square error of are between 0.0035–0.0470 and 0.0002–0.0082, respectively. Comparing the reconstructed signals by the GA–BP algorithm with the traditional BPNN algorithm reveals that the relative error of the BPNN algorithm is higher than that of the GA– BPNN algorithm. We can prove that the GA–BPNN algorithm has an excellent reconstruction performance for the hydrophone array system’s high spatial resolution with a large sensing cavity length.