Vehicle Routing Problem With Time Window Research Based On The Advanced Genetic Algorithm

Along with the consummation of modern enterprises' management idea and the development of market economy, as well as the E-Commerce irritably emerges with the vigorous development. Logistics cost control, which is one of the most important competition domains, have been influencing the Economy and it will be playing the pivotal role. As an achievement new specialization division of labor domain, Logistics had already been developed into an important part of industrial department and the national economy.Regarding to the majority enterprise, the cost of the transportation usually is the biggest single item cost. However, the Vehicle Routing Problems (VRP) is precisely the foundation of transports. The Vehicle Routing Problems (VRP) is a typical question of combination optimization. The question is that: How to use the limited quantity of the vehicles satisfied customer to meet the goods of the demand delivers, each customer can be visited only one time, each time the vehicles transportation total length is no surpass its maximum travel distance, simultaneously its load carrying capacity can't surpass its maximum load-carrying capacity, the goal is to make the delivers'cost minimum, namely making the length of vehicles travel shortest.If the customer has the certain time- limit to the goals delivers. Namely, the goods must be transported in the certain time, which is called Vehicle Routing Problems with Time Window (VRPTW), and it is what this article studies.According to the time restraint, the Vehicle Routing Problems (VRP) with Time Window (VRPTW) is mainly divided into: Vehicle Routing Problems (VRP) with Soft Time Window (VRPSTW) and Vehicle Routing Problems (VRP) with Hard Time Window (VRPHTW). However, through adjusting the penalty value of the VRPSTW to realize the hard time restraint. Therefore what this paper discusses is mainly the VRPSTW. According to this question, how to make the improvement to the Genetic Algorithm (GA) and make it adapt to the application background, thus we can gain the satisfactory solution effect, which is a key point the article pays attention to.In the foundation of the predecessor studies, this paper proposes an auto-adapted Genetic Algorithm (GA) based on mixed strategy through each method comparison and the analysis and does the correlation analysis. The auto-adapted mixed Genetic Algorithm (GA) is on the base of traditional Genetic Algorithm (GA), which is in view of the shortcomings of Genetic Algorithm (GA) to solve the problem.The advanced Genetic Algorithm (GA) has differences in parameter setting and algorithm structure, which many put up with following kinds of advanced strategy:This paper put forward the concept-"outstanding group". Because the individuals of the outstanding group are the most superior in the evolution, so they have the much better overall average performances.To fully use the partial search function of Genetic Algorithm (GA), simultaneously, avoid the"earlier precocious"phenomenon, proposes two heredities. Although, it has spent a certain time, it does very big improvement to the search success ratio and the search resultThrough the auto-adapted Maladjustment of the Genetic Algorithm (GA) is based on the evolution and speed up the search satisfaction solutions.It proposes the new Evolution Operator and Crosser Operator.It mixes other partial search algorithm.Because of these improvements, the function of Genetic Algorithm (GA) have got great improvement, not only could avoid the flaw about the earlier precocious effectively and the later searches agnates and so on. But also have improvement at the overall optimal solution ability as well as the logarithm astringency and ordination of population individual between multiplicity aspect and so on. Compared with the former Genetic Algorithm (GA) in performance, the speed of the advanced Genetic Algorithm (GA) is slowly, but the stability is much batter after many times operations. Moreover, the cost is acceptable in time.The operator generally can act according to the concrete situation in the actual operation; they establish the satisfactory solution, which may be accepted. Thus, if the result achieves the user request, the operator may terminate the operation. Certainly, we can obtain the final satisfactory solution if we have adequate time.From the above analysis, we can obtain the following collusion:Compared with the former Genetic Algorithm (GA), the advanced Genetic Algorithm (GA) can get superior solution.The stability of advanced Genetic Algorithm (GA) is much better, though many times operations.Because of stochastically produced initial community, the ratio of the local convergence is much smaller.We can obtain much better solution in the relative bigger pliability by using the mixed Genetic Algorithm (GA). But that is not absolution, although that is what we hoped.