Research On Particle Swarm Optimization And Its Application In Logistics System

The logistics has been considered as the'third source of the profit'. Sicne it is hard to increase profit with using the"natural resources"and the"human resources"now, the potential of the logistics gains more and more attentions. With the optimizations of the logistics enterprises can increase their profit and competitiveness by decreasing cost. Hence, it is significative to optimize logistics systems with optimization algorithms.The particle swarm optimization (PSO) is a stochastic optimization algorithm, it is easy to perform and can converge very fast. The dissertation studied and anlysised PSO deeply to improve performances. Based on which, a new approach was introduced, where the genetic particle swarm optimization (GPSO) was employed to address the inventory optimization problems and the Vehicle Routing Problem (VRP) instead of the traditional PSO. GPSO was derived from the original continuous particle swarm optimization and incorporated with the genetic reproduction mechanisms, namely crossover and mutation. The simulation results have shown that with heuristic algorithms, GPSO presented its effecitiveness and consistence.The achievements of the dissertation in PSO as follows:(1) An adaptive individual inertia weight strategy with mutation operators were proposed for high-dimensional optimization problems. The methods were proved to be effectiveness.(2) Kinds of mutation operators were incorporated to PSO. The results have shown no mutation operator could outperform others on all problems. Hence a multi-mutations operator was introdued, it was more consistent than single mutations, while performed worse on few problems. Based on which an adaptive mutation operator selection strategy was proposed, where the particles were divided in to groups with different probability to employ various mutation operators. The probabilities were adjust by comparision the evolutionary results between groups, the results have shown it outperformed the above mutation strategies. At last, a dual particle swarm optimization (Dual-PSO) was proposed, where the standard PSO and GPSO were incorporated, the simulation results have shown it outperformed both standard PSO and GPSO along with the genetic algorithm, the evolutionary strategy and the differential evolution. The achievements of the dissertation in logistics sytems as follows:(1) An approach for knapsack problem (KP) was proposed based on GPSO with a 1-2opt heuristic algorithm (GPSO-Opt) , the simulation results shows it outperformed the standard PSO and genetic algorithms introduced in the dissertation. A typical storage system was modeled and addressed with Dual-PSO and GPSO-Opt, simulation results have shown the effectiveness of the algorithms, and provided useful references.(2) The dissertation proposed a frame to solve traveling salesman problem (TSP) with GPSO. In the frame, a modified ordere crossover (MOX) was proposed to maintain information mainly from the high quality solutions during the crossover. A mutation operator based on 2-opt was introduced for local search. A two-stage crossover was proposed for the crossover among three chromosomes. Based on the above approach, a novel GPSO frame for TSP was propsed, which is simple, intuitionistic, flexible and easy to impletement. The simulation results have shown its feasibility and effectiveness.(3) A hybrid GPSO method was proposed to solve the capacitated vehicle routing problem (CVRP) , where MOX and 2-opt operator were employed along with a heuristic operator to deal with the constraints. The simulation results have shown its feasibility and effectiveness.(4) A two stage hybrid GPSO method was employed to solve the multi-depot vehicle routing problem (MDVRP). In the first stage the customers were assignmented to the depots and transformed the MDVRP to a set of CVRP problems; in the second stage the hybrid GPSO method was employed to solve the CVRP problems. The simulation results have shown the feasibility. Moreover, the results not only optimized the routings, but also provided supports for the decision-makings of truck purchase and depots placement.