A Susceptibility‐Aware Joint Planning and Cloud–Edge Collaborative Control Framework for Emergency Evacuation Under Geohazards in Mountainous Towns: A Case Study of Fengyi Town, Mao County, China

Evacuation in mountainous towns is constrained by geohazard susceptibility, shelter capacity, and limited alternative corridors. We developed a cloud–edge framework linking susceptibility rasters, directed‐road attributes, event states, and trajectory exposure. Before disruption, an ant colony system (ACS) generated feasible routes, and particle swarm optimization (PSO) allocated demand under conservation and shelter‐capacity constraints. Afterward, network costs guided vehicle rerouting. Across 125 static runs, joint ACS–PSO planning reduced mean planned susceptibility exposure by 20.2% relative to shortest paths. A matched ablation attributed most of this reduction to the PSO route–flow update. Across 750 dynamic runs, cloud–edge collaborative dynamic control (CEDC) and matched single‐layer control completed evacuation in D4, while CEDC reduced degraded‐road occupant time. In bridge‐throat D2, lower degraded‐road occupant time coincided with greater congestion. Dynamic effects varied with route executability and residual‐road loading across the tested events; online control cannot replace residual‐network capacity.