Genetic Algorithm Research And Application In The VRP

With the rapid development of the science and technology, the logistics industry is faced with more challenges. Thus the reformation is highly demanded if you want to succeed. Since the delivery expenses occupy more than 50% of the total cost, how to optimize delivery routing is the most concerned topic for the logistics enterprises nowadays. The Vehicle Routing Problem(VRP)is a complicated constitutes optimization, which has been researched increasingly wide-ranging. The delivery route heavily affects not only the dispatching speed but also the delivery expenses and the customers'satisfaction. This paper aims at optimizing VRP by using Genetic algorithm, which acts on cost reduction, customers'satisfaction, service up-grading. At present, the research on VRP is still in exploration stage and there is not a universal method for all problems. Some research is limited to satisfy the vehicle scale constraint without caring for the constraint of timing and customers'satisfaction. The short route is taken into consideration while the numbers of vehicles and the customers'satisfaction are not involved. These may increase delivery cost. Taking timing cost and customers'satisfaction into consideration, this paper explores further how to obtain the maximum economic results at minimum cost.In this paper, the main research contents are as follows:First, it brings a systematic introduction of VRP and the related theoretical knowledge about different arithmetic. It adopts the local-improvement Genetic algorithm to solve the VRP after fully analyzing the problem model and conduct a study on the present methods of solving VRP.Second, it tries to improve and perfect the VRP problem model by taking vehicle scale constraint, timing constraint, and customers'satisfaction constraint into account. The simplification of the model makes the arithmetic easier on the premise of all the previous conditions.It uses natural number codes tactics to make Genetic algorithm satisfy different kinds of model requirements. By using Scanning Algorithm,it effectively avoids initial group's low adaption due to the randomized policy and speeds up theAlgorithm convergence. The probability of passing the highly adaption unit to the next generation is increased by choosing the genetic operator through rank mode. It makes the process of variation more regular with the combination of reversal operator and mutation operator,meanwhile, it effectively avoids the fine gene deletion. As for the arithmetic, every steps try to avoid randomness and local convergence so that to easily get the best solution.It uses two different samples to test its superiority and emulate it through Matlab which makes the application visually and clearly.