Study On Vehicle Routing Problem With Backhauls Considering Uncertain Factors

Urban cargo distribution and collection is an important process in logistics, which is aprocess of cargo flow between logistic nodes and clients. The foundation of an efficient andeffective visual dispatch system for cargo distribution/collection vehicles is set up, byconstructing a multi-cost target function for evaluation the dispatch solution, simulatingsubjection curve of uncertain information, designing efficient reconnoiter search algorithmand conjugate optimization algorithm based on imprecise distance matrix. The issues exist incurrent research are focused to solve, which include inaccurate cost model, irrationalhypothesis of uncertain information, low efficiency of dispatch method as well as dispatchresult difficult to execute.In this thesis, all related information in dispatch process is described mathematically atfirst, then correspondent collection method is designed, algorithm to solve distance matrixbased on hierarchical structure is put forwarded so as to fast figure out distance matrixsolution in complex road map model. Multi-cost target function covered mathematic model ofVRP with backhauls is set up. The function uses integrate judge method based on subjectionfunction to calculate carriage arrange cost and consider fuel cost, vehicle depreciation cost aswell as driver cost together. Afterwards, traffic data of typical roads in Huhhot and realbusiness data in logistic enterprise are used as basis, subjection function of uncertain vehiclespeed, cargo weight and volume are simulated, and subjective judge indicators of dispatcherare introduced to describe the restrictions for mathematic model.VRP with uncertain vehicle speed and uncertain cargo weight/volume is described. Fastand efficient reconnoiter search algorithm is designed, which included strategy to fast createinitial executive solution, derivative method to dynamically design correspondent candidatesolution, strategy to dynamically select taboo length and to construct amnesty rule. Theeffectiveness of reconnoiter search algorithm is validated by calculation example.In case of new client suddenly occurs during the execution process of dispatch solution,concept of “hypothesis client” is introduced to transfer this dynamic problem to static problem,meanwhile, conjugate optimization algorithm is put forwards to optimize dispatch solutionbased on imprecise distance matrix. Furthermore, method to amend vehicle location isdesigned to eliminate the influence of optimization result, which caused by vehicle locationexcursion resulted from uncertainty of execution timing.Visual vehicle dispatch system is designed and developed for cargo distribution and collection in Huhhot. The property, network framework, main function, GIS data format,human-machine interface and dispatch solution release of the system are demonstrated indetail. After trial operation lasted one month, all the performance of the dispatch system suchas stability and real-time response meet the design requirements. The application of thisdispatch system obviously improves the economic efficiency of the logistic enterprise.