Research On Differential Evolution And Its Application In Container Terminal Logistic Scheduling

Optimization problems of equipment and resource widely exist in manufacturing and logistics systems. Because the optimizing quality and effectiveness are directly related to the operating costs and production efficiency of the system, how to design an efficient optimization algorithm has become a hot issue commonly concerned in academia and industry. Optimization problems can be divided into continuous and discrete ones according to the values of variables. For these two different typical optimization problems, basics and application of intelligent optimization methods are systematically studied, respectively, in this dissertation.Taking the nonlinear programming problem as a continuous optimization problem and the resource-constrained project scheduling problem as a discrete one, two differential evolution (DE) algorithms for solving continuous optimization problems and discrete ones are studied in this dissertation, respectively. For the nonlinear programming problem, a new DE algorithm based on an individual-dependent mechanism (IDE) is proposed, then the application research for stowage plan of the container ship is studied; for the resource-constrained project scheduling problem (RCPSP), a new pointer-based discrete DE algorithm (PDDE) is presented, and the application research of PDDE for container port berth and quay integrated scheduling problem is researched; a logistics optimization decision support system is developed in the context of an actual container port. The main work is summarized as follows:1) A new DE algorithm with an individual-dependent mechanism (IDE) is proposed for the nonlinear programming problems as a continuous optimization one. Different from classical differential evolution algorithm, in IDE, an adjusting parameter method based on the fitness difference between individuals is developed, various mutation strategies are designed, and random elements are introduced in the mutation strategies to avoid falling into local optimum, to improve the global searching ability. According to the experiments based on the international benchmark functions, IDE is the best DE for solving such problems up to now.(2) The IDE application is researched with the container ship stowage plan problem. The task of the problem is to decide the stacking order of outbound containers in each region of the container ship segments to maximize the stability of the ships. To solve this problem, a dividing continuous encode to determine the stacking priority of each container is designed, and a two-stage hybrid algorithm based on IDE is proposed. Experimental results based on actual data show that IDE can obtain near-optimal solutions of the problem within a short time.(3) In the background of resource-constrained project scheduling problem (RCPSP), discrete optimization problems are studied, and a new pointer-based discrete differential evolution algorithm (PDDE) is proposed. For discrete optimization problems, a ranking code is designed, and a new discrete mutation and crossover operators are defined to ensure that the algorithm is iterated in the feasible region. Through the observation of the solutions of RCPSP generated in the iterative process, we find a similarity law (pseudo theorem) between different solutions of the first (or last) several activities, and a reduced-dimension search strategies for holding similarity activities unchanged and evolving non-similarity ones is designed. Based on the solutions obtained in latter iterations, the common law of relevance characteristics between activities of the most of solutions has been found, and a local search strategy is proposed to keep the activities relevance. According to the experimental results based on the benchmark test data of resource constrained project scheduling problems, the proposed algorithm is superior to the current popular algorithms for solving RCPSP.(4) Application research of PDDE algorithm for berth and quay integrated scheduling problem in container ports is studied. The task of the problem is to decide the start time of each ship with satisfying the resource constraints on berth and quays, as well as the precedence relationships constraint of ships, the goal is to minimize the total finish time of serving all ships within a period. To solve this problem, through the perspective of resource-constrained project scheduling, we design ordering discrete encoding for determining the start time of each ship, then adopt PDDE to solve the problem. Experimental results based on the actual data and simulated data indicate that, PDDE is rapid and effective.(5) In a background of National Development Demonstration Zone GuangXi Qinzhou Port Container Terminal, based on the new continuous and discrete differential evolution algorithms, a decision support system for optimizing the scheduling problems in container terminal is developed. The functional modules of the system includes berth & quay crane allocation, container storage planning, container reshuffling optimizing, container ship stowage planning, and so on. Through online test at the port, the algorithm and system can effectively optimize the scheduling problems in container port.