Improved Meta-Heuristic for Multi-Objective Integrated Procurement, Production and Distribution Problem of Supply Chain Network Under Fuzziness Uncertainties

Abstract

In this study, we devoted a design under uncertainty of a supply chain network composed of multiple suppliers, multiple plants, multiple distributors and multiple customers. The proposed model is a bi- objective mixed integer linear programming which consider several constraints and optimize the total costs including the procurement, production, storage and distribution costs as well as maximize on-time deliveries. In order to model a similar problem to the real planning problems, the objective function coefficients (e.g., procurement cost, production cost and inventory holding) and other parameters (e.g., demand and production capacity), are all supposed triangular fuzzy numbers. Besides, a hybrid credibility-based model is constructed for the problem, i.e., expected value and chance constrained models. Moreover, the construction of the crisp equivalent model is based on different properties of the credibility theory. The resulted crisp equivalent model is a bi-objective MILP. To transform the crisp BOMILP to a single objective MILP model, we propose the LP-metric method. To solve the planned model for largest instances, we tend to propose a hybrid algorithm based on a recent metaheuristic called Hunger Game Search (HGS). This algorithm employs the so-called chaotic search to prevent premature convergence and entrapment in local optima. Finally, the efficiency and sensitivity of the proposed model are justified through the numerical results.