Simulation-based multi-objective optimization towards proactive evacuation planning at metro stations
Kai Guo,Limao Zhang,Maozhi Wu
DOI: https://doi.org/10.1016/j.engappai.2023.105858
IF: 8
2023-01-22
Engineering Applications of Artificial Intelligence
Abstract:Effective evacuation management is crucial in response to an emergency at metro stations. Due to the unpredictability and high complexity at metro stations, great challenges exist for evacuation management. A hybrid approach with the integration of building information modeling (BIM), simulation tool (Anylogic), and machine learning algorithms is proposed in this research to realize the evacuation event simulation and proactive evacuation management. A case study is performed to test the applicability and effectiveness of the proposed approach. It is found in the case study that: (1) The constructed simulation model could successfully perform the prediction of the evacuation process for the target metro station, and numbers of congestion areas can be identified (i.e., 7, 7, 9, 11 congestion areas for the four typical scenarios, respectively); (2) A proactive evacuation guiding strategy is proposed from the hybrid approach, which could realize a much better improvement for the evacuation events (at least 15.3% and 39.3% could be achieved for objectives of the evacuation time and the evacuation over-density rate, respectively), compared to the conventional guiding strategies; (3) The proposed proactive guiding strategy is the only one, in all three guiding strategies, that could shorten the evacuation time to the maximum extent and remove the congestion areas entirely. The novelty of the proposed approach lies in that: (i) The proposed hybrid approach could be able to accurately predict the evacuation conditions under different scenarios by incorporating the LightGBM algorithm; (ii) A proactive guiding strategy, along with the proposal of the innovative over-density rate rule, is provided with the ability of significantly improving the evacuation efficiency. This proposed approach not only presents an efficient tool for the evaluation of evacuations, but also greatly enriches the field of proactive evacuation management at metro stations.
automation & control systems,computer science, artificial intelligence,engineering, electrical & electronic, multidisciplinary