| Summary: | 碩士 === 國防管理學院 === 資源管理研究所 === 89 === A Study of Using Experimental Design to Improve
Vehicle Routing Problem with Time Windows
ABSTRACT
This thesis is aims to find the optimal combination of factors in different conditions of the vehicle routing problem with time windows (VRPTW), by using the experimental design and analysis of variance. For simplicity, the author uses graphics to present information of customers and vehicles, and the solution processes.
The solution processes are as follows:
(1)Setting a solution process for solving the VRPTW with literature reviews.
(2)Modifying the initial solution model of the process (1) to be a two-stage mataheuristic with a route neighborhood structure.
(3)Obtaining the high quality initial feasible solution and embeding the improvement procedure in (2).
(4)Using the benchmark problem to test and verify the two-stage mataheuristic (2), and using the fractional factorial design to analyze the different problem sets.
(5)Comparing and analyzing the different combination’s solutions and to find the optimal solution and comparing those with existing solutions in the literature.
The research results are as follows:
(1)The different factorial combinations lead to different solutions, and the different factorial combinations are of no significance among different problem types.
(2)The interaction between the different problem sets and time windows attribution affects the number of vehicles required.
(3)The “Distance” factor is more important than others in the parallel insertion algorithm.
(4)The algorithm designed in this thesis is comparable with other existing algorithms, especially in the real-life VRPTW.
Key words: Experimental design; Time windows; Vehicle routing problem;
Heuristic algorithm; Logistics
|
|---|
