Synchronizing victim evacuation and debris removal: A data-driven robust prediction approach

This study introduces a new perspective in disaster management's response and post-disaster phases to synchronize multiple vehicles for victim evacuation and debris removal processes. A broad range of interrelated scheduling and routing operations and various synchronization aspects of heterogeneous vehicles are considered in this regard. A novel bi-objective mixed-integer programming model is presented, where the first objective function aims to minimize the total costs of the relief logistics network, and the second one minimizes the total operation times of vehicles. Moreover, due to extensive empirical and analytical errors, preliminary travel and service times are inexact and unreliable. Hence, a novel two-stage data-driven approach is rendered to predict reliable travel and service times. In the first stage, a new hybrid machine learning model is rendered to predict these times, and in the second stage, the distributionally robust optimization with φ-divergence is employed to surmount the unreliability of predicted times. A real case study is examined to illustrate the validity of the proposed model and solution approach. In addition, several simulation experiments are conducted to demonstrate the superiority of the proposed solution method in terms of robustness. Finally, the proposed framework can improve the planning by rendering meaningful insights concerning significant parameters' influence over the schedule and routing consequences.