Research On Optimizing Logistic Distribution Routing Algorithm And It's Application

In the storm of economic globalization and informatization, the Logistic Distribution industry has expanded from offering traditional transportation service to integrated Logistic Distribution industry system afforded by modern technology, management and information. The development of Logistic Distribution industry has become a new point of growth of national economy, the scientific and logical Logistic Distribution industry as"the third source of profit"is an important part of continual economy development. And the Logistic Distribution is the direct link with customers in the Logistic Distributions. In the various costs of Logistic Distribution, the cost of Logistic Distribution is a large proportion. Whether the arrangement of dispatching routes is rational or not affects dispatching rate, costs and profits largely ,especially it is more complex in confirming dispatching routes of multi-user. Optimize dispatching routes using scientific and logic methods is an important work of dispatching.Logistic Distribution Optimize Problem is a typical Combination Optimization Problem, is NP-hard and has very high computation complexity. As the boom of market economy and fast development of Logistic Distribution industry, more and more enterprises aware of the importance of Logistic Distribution in enterprises'production and distribution flow. Traditional manual selection of dispatching routes depends absolutely on the experience and knowledge of workers, exhausting much time and energy. As the enterprises'scale broadening, and follows the business scale broadening, the number of dispatching spots increases, and dispatching routes arrangement becomes more and more complex. Therefore, manual dispatching routes can not satisfy the needs of enterprise business, and arranging routes using computer becomes imperative. Arranging routes using computers is fast and efficient, and also can be integrated with the ERPsystem of enterprises. The aim of this method is to improving efficiency of physical dispatching and reducing the costs of production. Whether the arrangement of dispatching routes is logical or not impacts largely on the rate, costs and profit of dispatching. Therefore, confirming dispatching routes using scientific and logic methods is a very important task in dispatching activity. The regular models of Logistic Distribution are as follows:(1) Point-to-point dispatching, also named single-point dispatching, this is a relatively simple model.(2) Many points dispatching, also named multi-points dispatching, is also a relatively simple model. This model mainly solves the balance problem of transportation and dispatch among several points, this is also the problems that both starting points and ending points are not only.(3) The single-loop dispatching problem is: there is nodes set D in routes optimization, selecting an appropriate route that it goes by all points and it has to be closed. As far as less constraints of practical problems are concerned, this model is relatively simple ,and can simply problems.TSP(Traveling Salesman Problem) is a typical model of single-loop dispatching problem, and also is a typical NP-hard problem. It can not obtain optimized solution in large scale route optimization problem, and only can choose more optimized solutions in all approximate solutions.(4) The multi-loop dispatching problem is a regular dispatching problem in practice, especially for the entities which have numerous service objects. For example, a tobacco or a pure-water company which has thousands of clients, the dispatching problem exists. The core problem is how to minimize the dispatching cost using least number of vehicles or making the most of existed vehicles, and also have to offer to clients on time. VRP(Vehicle Routing Problem) is a most basic model of resolving problems of large dispatching factories or the dispatching entities (vehicles), and many problems should be considered on base of it.(5) Many small-cargo dispatching companies, such as tobacco, pure water, and express, should deal with the problems involving thinning routesand many influencing factors. Thus, the selection of dispatching center also affects dispatching scheme and service quality.In this paper, we advance the solutions for single-loop and multi-loop dispatching problems respectively. As for single-loop dispatching problem, we compare the performances of branch bound method, nearest intervention method and genetic algorithm. The running time of nearest intervention method is short in solving large-scale route programming problem and it has much superiority to others. Whereas, the result of this method is not perfect. We present a heuristic algorithm on the base of nearest intervention method.This algorithm has two steps:Step 1: Obtain initial solutions using nearest intervention method; Step 2: optimize the initial solution obtained in Step 1 using local search Engines.At present, this algorithm is verified by the ADMS system of TYBB. The solutions of heuristic is much improved compared to those of nearest intervention. As for multi-loop dispatching problem, there are many solutions because of various requirements of clients. On the base of practical application and requirement of most client, we build mathematic model for it. Under this model, we suggest many solutions for multi-loop dispatching problem combining packing algorithm. First, as for a depot, we should obtain the destinations set that the goods in it are to be shipped to, calculate the TSP paths that start from this depot and go by all nodes of destination set, confirm the vehicle-loading sequence of goods based on obtained paths, and confirm the usage sequence of vehicles based on vehicle-loading sequence. And next load the goods of this storage according to the sequence of vehicles. And calculate the route again of the loaded vehicles of this storage: As for a loaded vehicle, obtain the destination set that it should go by and calculate the TSP paths that start from this storage and go by all nodes of destination set. We can use branch bound method according to situation. From the compare data of experiment, we find the branch bound method not only can obtain optimized solution, but has high performance efficiency. Last set thepath of this vehicle again according to the route obtained. And next calculate the vehicle loading of next storage. This algorithm is also applied to the ADMS system of TYBB, and also applied by many physical distributing company.The Logistic Distribution is a complex systemic work, and many optimize algorithms have to be optimized in the days ahead.