Path planning method for maritime dynamic target search based on improved GBNN

Abstract

To address the issues of low discovery probability, inefficient search, and antagonistic targets during the process of dynamic target search in the ocean, a dynamic target search path planning method based on the Glasius biologically-inspired neural network (GBNN) in combination with marine environmental information is proposed. Firstly, the motion model of the searcher and the capability model of sonar detection are established, and the dynamic motion characteristics of the target are analyzed. The Beta distribution is employed to characterize the variation of the target velocity, and the distribution probability map of the target position alterations over time is obtained. Then GBNN is presented and the marine environment information is integrated to enhance the calculation approach of the internal connection weights of the network. Moreover, the update rule of the activity value of the neural network is reconfigured. The influence of the peak of the dynamic target distribution probability on the activity value of the neuron is regarded as the external incentive element. According to the turning limitation of the searcher and the activity of GBNN neurons, the search path points are determined smoothly. The paper's algorithm, validated through 10,000 Monte Carlo simulations with real maritime data, significantly outperforms traditional search methods in the discovery probability and search efficiency.