Solving The Min-Cost Pronlem Of Multi-Modal Teransport With Time Windows

Taking Just-In-Time distribution and cost decrease into consideration, the minimum cost of Multi-Modal Transport with Time Windows (MMTTW) was studied in this paper according to the key issue of logistics commonness technology and actual cases of the traffic network.The paper summarized the academic researches have proposed various MMT that was related with time windows so far. They mainly are shortest viable path in multimodal networks with the theory of genetic algorithm, shortest time taking of MMT problem, vehicle routing problem with time windows etc. And then, the paper has summarized variant algorithms for as above problems. They include the exact algorithms, the classical heuristic and meta-heuristics. In these algorithms, the size of problems that the exact algorithms can solve is too small to satisfy the need of the problem. And the main shortcoming of the heuristics is too slow constringency speed and too long computational time to fulfill the task of dynamic dispatching.This paper focus on the works bellow:(1) Based on the theory of constraint programming, the modeling and the solving of MMTTW was investigated with Natural Constraint Language (NCL) . The constraint of time windows is circumstantiated from the point of view of customers and the point of view of managers. This paper introduces the concept of the transport time constraint, departing time constraint and the capability constraint of vehicles. A mathematical model of the problem is proposed based on the constraint programming. Furthermore, the searching strategy of this model id discussed.(2) For illustrating and comparing the solving capability of the model, this paper investigates genetic algorithm to study the MMTTW problem. The issue is divided into two steps to solve. The first step enumeration practicable path according to the route information; the second step using genetic algorithm, paths in the first step are generated as gene, all the paths of orders constitute chromosome. Mathematical model was established considering the weight, quantity and size restrictions. Then the solutions to two methods given in this paper are compared. As the solution to big-scale problem from constraint programming is better, the first model is chosen to solve the great-scale problems.(3) In order to increase the speed, the constraint programming model has to be changed. The focus is to combine the routes which have the same starting and ending points and the different transport modes and carriers as a class. Then the ships in one route class are arranged as one-dimension al array. This step can reduce the branch in searching. Because of the model more compact and flat, search strategy is changed, which took much less time to find the solution. Finally, some tests based on specific data improve that the algorithm is effective.