A Kalman Filter-Based Submersible Position Prediction Model and a Multi-Target Dynamic Search and Rescue Scheme

Aiming at the deep-sea submarine search and rescue problem, this paper proposes a universally applicable submarine position prediction model and search model. The position prediction of the submersible is realized by Kalman filtering and dynamics modeling. Meanwhile, Monte Carlo method is used to construct the initialized population of the motion trajectory and optimize the search path step by step by iterative genetic algorithm. In this paper, the optimal search path is taken as the objective, the main ship SAR trajectory is simulated, and the probability of the diver is fitted as a function of time and cumulative search results. For the multiple submersible search and rescue problem, this paper proposes two possible search and rescue schemes and compares the search and rescue time. Then we propose the optimal multi-objective dynamic search and rescue method. In addition, this paper adopts the G1-entropy weight-coefficient of variation combined assignment method to evaluate different search devices and creatively adopts the CRITIC method to secondary assign the weights obtained from different evaluation methods to select the optimal search device. Finally, sensitivity and robustness analysis are also carried out in this paper.