Research On Dynamic Location-Routing Problem Of Distribution Center

The location-allocation problems (shortened form LAP) and the vehicle routing problems (shortened form VRP) are considered synthetically in the location-routing problems (shortened form LRP). With the concept of integrated logistic management system accepted gradually by more and more people, LRP has been a very pop topic for many subjects currently. Through research on the LRP, not only the optimal quantity of the depot, location, capability and the clients'allocation scheme can be determined, but also the optimal vehicle routing can be sought. Thus, the distribution cost is reduced, and the circulation efficiency is improved, which can make the corporations hold more competition ability. So the research on the LRP is significant to real world. The dissertation selects the LRP as research object, and the main research contents are as follows:After analyzing the location basic aims, criterions and influence factors systematically, the dissertation uses interval analytical hierarchy process (shortened form AHP) to determine the locations of distribution centers with an eye to the indeterminacy of decision. While the process of dealing with the interval numbers, the concept of identity-discrepancy-contrary is introduced about the set pair analysis (shortened form SPA), and the interval number is translated into connection degree. Interval AHP based on SPA is proposed to apply to the location problem.When the static LRP is studied, the concept of vehicle running number is introduced, and one vehicle could serve for more than one route using the different vehicle running number. Combined with the economic order quantity (shortened form EOQ), a static multi-depot LRP model of two-level is built. Due to the LRP is NP hard, a two-phase sequential heuristic solution is designed. The original problem is divided into two correlative sub-problems, i.e., the location-allocation problem and the vehicle routing problem. Whereafter, partheno genetic algorithm and double population genetic algorithm are proposed respectively to solve the two sub-problems.On the basis of the research on the static LRP, it is considered that the demands of clients are variable with the time, and the programming period is divided into some stages, then the dynamic LRP is changed into a multi-stage decision problem. Subsequently, the clients'demands are determined, the static LRP model and algorithm are used in each stage so as to obtain the optimal location scheme and the routing scheme of the corresponding stage. Finally, put all the optimal location schemes of every stage together as a set of feasible decision schemes, and make the multi-stage decision problem as a shortest-path problem. Then the dijkstra algorithm is used to seek an optimal location strategy so as to make the whole cost of all the programming period be minimum.The research above not only improves the practical application effect of the LRP model, but also insures the validity and maneuverability of the algorithm. The production makes the location theory and technique more substantial and integrated, and it is of valuable worth for practical application.