Research on the optimization of urban emergency shelters considering flood disaster risks

To enhance the operational efficiency of emergency shelters in disaster management, this study focuses on flood disasters and identifies four critical indicators. These include the hazard posed by disaster-inducing factors, the sensitivity of the disaster-prone environment, the vulnerability of the carrying capacity, and the capability for disaster prevention and mitigation. The Random Forest algorithm is employed to assess the spatial distribution of flood risk, and the resulting risk values are incorporated as weights to address latent threats. The method simultaneously considers the minimization of weighted evacuation distance and accessibility inequality. It addresses multiple objectives—such as equity, efficiency, and balance—within a multi-objective optimization model for shelter location. The model utilizes an enhanced Non-dominated Sorting Genetic Algorithm II (NSGA-II), validated through a case study of shelter allocation in Zhengzhou. The results show that, compared to the current shelter configuration, the optimized plan reduces total evacuation distance, uncovered risk value, and accessibility inequality by 31.46%, 10.73%, and 28.80%, respectively.