Simulated Annealing for Multi Objective Stochastic Optimization

Abstract

This paper addresses the multi-objective stochastic optimization problem that arises in many real-world applications, especially in supply chain management and optimization.To this end, a simulated annealing algorithm is presented and used for solving this problem. The algorithm uses the hill-climbing criterion in order to escape from local minimality trap. The paper also introduces a new Pareto set for stochastic optimization problems and demonstrates the application of simulated annealing on this Pareto set. Finally, the proposed algorithm is applied on an inventory example that is solved by optimizing three objectives. Numerical results indicate that the algorithm is capable of constructing a Pareto set of non-dominated solutions.