Hybrid Algorithm For Vehicle Routing Problem

With the continuous development of e-commerce, modern logistics, logistics and distribution in the vehicle routing is reasonable or not the speed of the logistics, cost, capacity equipped as well as distribution costs and the effective impact is enormous, the use of scientific and rational approach toMake sure the vehicle path will be the logistics and distribution is an important task.Ant colony algorithm, ant colony algorithm has the following advantages: strong robustness, essentially parallel, easy combination with other methods. Although the ant colony algorithm has many advantages, yet at the same time there are some flaws, such as:(1) Compared with other methods, the algorithm requires a long search time.(2) is prone to stagnation appear to produce a local optimal solution. Genetic algorithm with a limit of not function constraints, with the advantages of implicit parallel search feature. The algorithm is not the solution itself and a single individual can greatly avoid falling into local optimum possible. The disadvantage of the GA is the choice of the initial population is often directly affect the solution quality and efficiency of the algorithm is very sensitive to the initial population. Optimization problem for the complex structure of the combination of the search space, search for a long time, will always be a premature convergence. Genetic algorithm, ant colony algorithm, the same advantages of integration and their own characteristics, their integration can better solve the VRP problem.This article describes the vehicle routing problem VRP models and algorithms, as well as the status quo at home and abroad, and lists the basic principles of the various algorithms. Focus on the basic principles of genetic algorithms, ant colony algorithm and genetic algorithm, ant colony algorithm to construct the hybrid algorithm of the VRP problem, and with a large number of TSPLIB and VRPLIB, to do the simulation of each algorithm to get the the GAAA algorithm is superior to other algorithms in this article.