Many‐Objective Evolutionary Emergency Assistance Decision‐Making for Highway Tunnels Considering Time Factors

ABSTRACT Emergency response planning increasingly relies on data‐driven decision support models. However, historical emergency cases are predominantly stored as unstructured text, and real‐world rescue conditions exhibit significant dynamism, rendering traditional methods ill‐suited for complex emergency scenarios. To address this, this paper proposes a time‐aware multi‐objective evolutionary emergency aid scheduling method (MEEAS‐T) to provide more relevant decision support for current emergencies based on historical cases. This method constructs a structured representation of emergency knowledge, integrating factors such as case similarity, scenario diversity, processing time, resource consumption, and temporal relevance into a unified multi‐objective optimization framework. Furthermore, it introduces a partition‐based search mechanism and applies a vector‐angle‐based guided probabilistic environment selection strategy to mitigate individual overcrowding and excessive dispersion. This approach ensures solution diversity while reducing computational complexity. Experimental results demonstrate that the proposed method more effectively screens historical cases highly relevant to current events, reduces redundant recommendations, and significantly enhances decision support quality in dynamic emergency scenarios.