Genetic-algorithm-based heatmap optimization of maritime search and rescue unit deployment considering time-step dynamics

Establishing a rapid and appropriate search plan is crucial to minimize casualties during a maritime distress incident. Developing an optimal search plan for many particles, as predicted by the particle drift prediction model in the Monte Carlo simulation method, poses significant challenges and requires considerable time and physical resources. This paper proposes a weighted heatmap-based optimization framework for the deployment of maritime search and rescue units (SRUs), which incorporates time-step dynamics for adaptive planning and accounts for the relative motion between SRUs and drifting particles during the dynamic updating stage. The framework consists of three integrated modules: (1) a time-dependent weighted heatmap generation module that represents the probability-density distribution of particles corresponding to each search object type, generated at regular intervals from the commence search time (CST) to the end search time (EST) to reduce computational cost; (2) a genetic-algorithm–based optimization module for efficient deployment of SRUs over time-evolving search areas; and (3) a dynamic updating module that iteratively adjusts coverage factor by incorporating time-step dynamics and the relative motion between SRUs and drifting particles to maximize the cumulative probability of success ( P O S c ). Through a case study, we demonstrate the applicability and effectiveness of the proposed search plan optimization framework for an actual maritime SAR mission. • A weighted heatmap-based optimization framework is proposed for maritime search and rescue (SAR) missions, where the heatmap assigns weights to different search-object types according to their relative importance. • The framework incorporates time-step dynamics and relative motion between search and rescue units (SRUs) and drifting particles. • Genetic algorithm (GA)-based optimization is employed to deploy SRUs efficiently across time-evolving search areas. • The framework improves search efficiency and cumulative probability of success ( P O S c ) compared with conventional allocation methods. • A case study demonstrates the applicability and effectiveness of the proposed approach to actual maritime SAR operations.