Hybrid Particle Swarm Optimization Algorithm for the Logistics Network Design Problem under Concave Cost Function

An effective hybrid algorithm integrated the heuristic route exploration algorithm (HRE) and particle swarm optimization (PSO) is proposed for logistics network design problem with transportation concave cost function between plants and distribution centers. In this solution, PSO is used to search feasible logistics network structures, while HRE algorithm is used to decide whether to route a shipment through an intermediate consolidation centers or route it directly to its distribution centers. A novel encoding scheme satisfying spontaneously with complicated constraints is designed. Two particle state updating mechanisms are presented to mitigate premature convergence. Simulation results and comparisons demonstrate the effectiveness of the proposed hybrid PSO algorithm.

I. INTRODUTION

In today's increasingly competitive global market, companies more and more attach importance to better design and manage their logistics networks in order to meet higher-level service at the lowest possible cost. To achieve this goal, the optimal logistics network design requires the integrated decision-making of facility location, inventory control and transportation routing, which has become one focus problem of logistics researches. Tian (2004)considered distribution center location problem for chain retail enterprise under stochastic demand and scale economies of transportation [1]. Teo(2004) studied the distribution network design problem integrating transportation and infinite horizon multi-echelon inventory cost function, where the trade-off between inventory costs, direct shipping cost and facility location cost is considered, only small or moderate size models were solved efficiently by column generation [2]. It is proved in practice that the sequence modeling can lead to sub-optimality. However, the integrated modeling for logistics network design problem, which the facility location, inventory policies and transportation routing is considered simultaneously, creates opportunities for efficiently reducing cost, but complicates the mathematical analysis and solution of the location problem.