Study On An Optimization Of Dynamic Vehicle Distribution Route Based On GPS

The paper started from related concepts of the vehicle routing problem, analyzed the elements that constitute the vehicle routing problem, and then led to a dynamic vehicle routing problem; on the basis of analyzing various uncertain information of dynamic vehicle routing problem, this problem is divided into six types, and cleared the research direction:the uncertainty of customer demand aroused dynamic vehicle routing problem; then compared the different characteristics of dynamic and static vehicle routing problem; according to the research process, summed up related solving algorithm of the vehicle routing problem, and outlined basic principles, advantages and disadvantages of various algorithms.Through comparative analysis, determined to use the GPS, GPRS and GIS technology as the supporting technologies for dynamic vehicle scheduling system, described the network architecture of dynamic vehicle scheduling system, and created part of the building block design.Determined a two-stage solution strategy based on the description and analysis of related problems of the hypothetical distribution enterprises:static optimization stage, which is a pre-optimization procedure for pre-reserved applying customers; dynamic optimization stage, which is a further optimization procedure that combined with some optimized distribution strategy for real-time applying customers. According to business goals, established a mathematical model of the dynamic vehicle routing problem with soft time window.At the dynamic optimization stage, subdivided the real-time customer demand for five types:newly increased customer demand, increase of original customer demand, decrease of original customer demand, cancel of original customer demand and change of time window of original customer demand. Designed separate sub-strategy for each type; designed improved genetic algorithm for solving dynamic routing optimization problem through the improvement of fitness function and crossover operation.Conducted simulation experiment for hypothetical distribution cases by means of Matlab and C# program, tested and verified feasibility of the proposed algorithmic while simulating to realize the dynamic distribution strategies.