Research On Routing Optimization Problem Of 4^th Party Logistics With Stochastic Transportation Time

Along with the development of globalization, people have recognized that optimization and integration of supply chain are playing a more and more important role in competition between modern enterprises. The birth of 4th party logistics (4PL) just complies with this trend. The 4PL provider takes on the task of integrating and optimizing the whole logistics system. It is proved that 4PL has improved the efficiency and has reduced the operation cost sharply, so the relevant research has significant meaning.Generally speaking, the main job of 4PL provider is to select suitable 3PL suppliers from a huge amount to find out a path costs the lowest. In pass researches, researchers always abstracted a multi-graph whose arcs represent the 3PL suppliers from logistics system. Although there has been some progresses in this field, most of them are standing on deterministic problems without considering influences by practical changes. In this thesis, we put objective factors that cause disturbances of 3PL transport time into consideration. At this time, we assume these time disturbances follow normal distribution and have their own upper and lower bounds. Under all premises above, we research single point to single point and single point to multi-point 4PL routing optimization problems.After reviewing 4PL, stochastic programming and Harmony Search algorithm (HS), we make our study as follows:Firstly, we start with single point to single point 4PL routing optimization problem with stochastic transportation time and establish expected value model, chance-constrained programming model and approximative equivalent deterministic model of chance-constrained programming based on multi-graph.Moreover, design enumeration algorithm, heuristic algorithm with prejudgement of infeasible graph, HS and HS with constrained Dijkstra. Results come from comparison and analysis of these four algorithms:enumeration algorithm can get the accurate solution, yet time-consumming; the heuristic algorithm shorts this time and can get the same result, but the time is still unsatisfactory; HS can gain relatively good solution in grately shorter time, but the flaws are the low stability and high iterations; aside from guaranteeing solution, HS with constrained Dijkstra can improve stability and can lower down times of iteration sharply.In addition to the points above, we choose HS with constrained Dijkstra to solve three models, and from the results we can conclude that although expected value model consume shorter time in calculation, it's robustness is far less than that in demand; chance-constrained programming model cost a lot of time in calculation; approximative equivalent deterministic model of chance-constrained programming can keep balance between robustness and calculating time.Finally, we expand research to single point to multi-point 4PL routing optimization problem with stochastic time. Except for establishing mathematic models, we simulate different examples and analyze the data and results.