Abstract: Since the spare parts demand is almost the intermittent demand in real spare parts support, the supply of spare parts is usually a multi-stage dynamic optimization problem. Focusing on this, a multi-stage mathematical model of spare parts supply is constructed. In order to solve this kind of dynamic optimization problem, a meta-heuristic dynamic optimization algorithm is proposed. Firstly, an environment change detector and an environment change response strategy are added to the classical differential evolution algorithm, which enables the differential evolution algorithm to solve the dynamic optimization problem when the environment changes. Secondly, a self-adaptive Levy flight strategy is proposed, which enables the algorithm to maintain a good global exploration and local exploitation capability when the environment changes. Empirical test shows that the proposed dynamic self-adaptive difference algorithm can obtain the optimal feasible solution of the model, and the distribution and convergence of the algorithm are greatly improved.

Key words: spare parts supply, dynamic optimization, differential evolutionary algorithm, Levy flight, self-adaptive