다양한 알고리즘을 활용한 지하 역사에서의 최적 대피 경로 분석

초록

This study analyzed optimal evacuation route derivation for subway stations during disasters using Dijkstra, A*, and Greedy algorithms. A hypothetical three-level underground station was modeled, and evacuation routes were analyzed for all three algorithms. The effects of crowd density and disaster-induced movement speed reduction were incorporated. Results demonstrated that the A* algorithm exhibited the best performance, with a maximum evacuation time of 349 seconds. The Dijkstra algorithm recorded 365 seconds, while the Greedy algorithm required 1,545 seconds. Notably, the Dijkstra algorithm, which theoretically guarantees optimal solutions, yielded a longer maximum evacuation time than the A* algorithm. This counterintuitive outcome occurred because Dijkstra’s path optimization performance generated identical routes for clusters located at similar nodes, paradoxically increasing crowd density. This finding suggests that excellent optimization performance can paradoxically intensify crowd density. In contrast, the A* algorithm’s heuristic-based search induced route diversity, producing a natural distribution effect. These results indicate the necessity of designing algorithms that prioritize collective efficiency rather than individual optimality in evacuation route planning.

키워드