A Multi-Depot Vehicle Routing Problem with Stochastic Road Capacity and Reduced Two-Stage Stochastic Integer Linear Programming Models for Rollout Algorithm

A matheuristic approach based on a reduced two-stage Stochastic Integer Linear Programming (SILP) model is presented. The proposed suitable for obtaining policy constructed dynamically the go during rollout algorithm. algorithm part of Approximate Dynamic (ADP) lookahead solution Markov Decision Processes (MDP) framed Multi-Depot Vehicle Routing Problem with Road Capacity (MDDVRPSRC). First, Deterministic VRP (D-MDVRPRC) Then an extension, MDVRPSRC-2S, presented as offline SILP MDDVRPSRC. These models are validated using small simulated instances CPLEX. Next, two versions MDVRPSRC-2S (MDVRPSRC-2S1 and MDVRPSRC-2S2) derived. They have specific task in routing: replenishment delivering supplies. to be utilised interchangeably depending capacity vehicle, repeatedly execution reinforcement learning. As result, it shown that base consisting exact optimal decision at each epoch can obtained constructively through these stochastic integer linear programming models. results from resulting CPLEX also validate This simple implementation when performing ADP.