The Study On Inventory Transportation Integrated Optimization And Visualization In Agricultural Products Logistics Supported By TGIS

The infrastructure of Agricultural products logistics is increasingly perfect,while software system is relatively weak in China.In order to improve the asymmetric development and use fully the potential of the infrastructure,the Inventory Transportation Integrated Optimization(ITIO)problem around the links of inventory and transportation was proposed,which was used to heighten the level of comprehensive decision.And an object-oriented spatio-temporal data model for logistics was designed,which was used for access and expression effectively of spatio-temporal data produced by implementing the optimization scheme.The content is described as follows.The many-to-one and one-to-many ITIO(Inventory Transportation Integrated Optimization)were abstracted and decomposed from the direct link between farmers and supermarket centered on the distribution center.In the basis of periodical vehicle routing problem(PVRP),the inventory variable was introduced into the PVRP to make an overall optimization for inventory and transportation simultaneously.The hybrid heuristics algorithm based on two dimension matrix was designed.In the algorithm,the Tabu Search process was defined by neighbourhood solution and Tabu object,and the branch and bound method was embedded into the process,so as to optimize the product amountand vehicle routes.The analysis of examples shows that,the inventory and transportation cost remains the trade-off relationship;Compared with HGA,the algorithm is suitable for the condition of a smaller number of service phases or objects.In order to refine further the agricultural products distribution requirement,the hybrid heuristic algorithm was modified for the multi-product distribution problem.The two dimension matrix was extended with the dimension representing product type.The roulette wheel method was adopted for store selection.And the tabu objects were compared with two-level mechanism.The example result indicates that,a type of products in the same store is seldom delivered separately;And multi-product distribution can be seen as single-product distribution problem.The algorithm used for multi-vehicle distribution problem is concentrated on the improved C-W method.The stores were scanned once,then multiple routes were constructed simultaneously using this method.What's more,routes merging was designed to heighten the vehicle load rate.The example result shows that,vehicle routes are still influenced by the inventory cost;and the vehicle with higher full load can save more logistics cost without considering empty-load rate.In order to store and access effectively the spatio-temporal data during the process of logistics,based on object oriented analysis,the logistics subjects were abstracted into two types of logistics-object class,inventory-object and transportobject class.The storage management structure was designed,which is composed of three levels,including logistics objects,object state and feature state.And the spatio-temporal conceptual model was constructed according to the relationship among objects,events and event rules.Then version-based event-driven management mechanism was proposed.All the events and objects were linked through the version in the mechanism.In accordance to the storage structure,the method of getting spatio-temporal data based on the time point and time interval was proposed.To enhance the visuality and usability of the spatio-temporal data,using the third-party map service,the key method of coordinate system transformation,coordinate points filtering and layers overlaying was realized from data layer,process layer and user layer respectively.The method was used to solve the map visualization problem of different coordinate system,redundant and abnormal points,lost feature points.Based on LBS technology,the function of real-time location monitoring,historical track replaying and planning routes demonstration was primarily realized.The visualization effect of data reaches expectations.In order to verify the validation of data filtering and feature points extraction,taking the vehicle track in reality for example,the filtering rate reaches 30% without affecting the visualization effect.