The application of simplified swarm optimization in a precautionary evacuation model

• Published in: Swarm and evolutionary computation Vol. 75; p. 101189
• Main Author: Lai, Chyh-Ming
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.12.2022
• Subjects: covering problem; precautionary evacuation; simplified swarm optimization; covering problem; simplified swarm optimization; precautionary evacuation
• ISSN: 2210-6502
• DOI: 10.1016/j.swevo.2022.101189

•A novel covering model for precautionary evacuation is proposed.•A repair mechanism to repair infeasible solutions to be feasible is proposed.•A local search to improve solutions without violation of constraints is proposed.•Statistical results indicate that the proposed method is better than its competitors. This paper introduces a two-level covering problem in the pre-positioning of military forces for precautionary evacuation during a typhoon, which is called the precautionary evacuation covering problem (PECP). PECP is highly constrained by multiple practical constraints, and requires a long time to converge to the optimal or near-optimal solution, as the evolutionary method used to solve it often encounters a population of infeasible solutions. This study proposes a tailored simplified swarm optimization method (SSO) with two novel mechanisms to solve this problem, and utilizes the problem-specific knowledge of PECP to accelerate the solution process and improve the solution quality. The proposed approach (SSOR&D) is empirically verified on a set of randomly generated problems, and compared with widely popular evolutionary approaches. SSOR&D can reduce the average of fitness value by at least 26% performed better than original SSO owing to the two proposed mechanisms. Compared to three other evolutionary methods and their improved versions, SSOR&D outperformed 26 out of 36 instances of PECP in terms of the average of fitness values. Overall, the computational results demonstrate that the proposed approach is highly competitive and performs well in terms of solution quality and convergence rate.