Research On Decision Making Of The Container Stowage Multi-port Master Planning Under Forecast Bias

Container Stowage Multi-port Master Planning(referred to as “CSMMP”)is a kind of multi-stage dynamic loading problem in the time dimension based on the space,The packing problem which considers ship or the Bay of ship as a container or backpack and assigns the storage units of the ship to the port on the whole route is the basis for studying on CSMMP,and the goal is to maximize the storage unit utilization or shipping benefits of the full liner.CSMMP(known as pre-allocation plan)is based on booking situation of the current node and forecast information of the follow-up node to achieve the preallocation of the current node liner and reservation-stack for follow-up node,so predicting the information of containers is the basis for making the stowage plan.For the actual number of loading container does not match the pre-planning problem which is caused by the predicted error and the late arrival container or temporary container and other small probability random events,the current stowage plan needs to be adjusted to improve the robustness of the full-line stowage plan if the original preallocation plan is required and the ship's safety performance requirements are met.(1)Prediction of container information based on moving average Markov methodFor uncertain container information in liner route,in order to reduce the impact of stowage information that be caused by small probability random event and improve the prediction accuracy,the moving average Markov prediction model is constructed,and the cases made by the investigation data of Nanjing port show the average error of forecast is 4.43%.(2)Decision research on the whole route pre-allocation based on the forecastIn order to improve the utilization ratio of ship storage units,the k-partition strategy of ship's shell structure is proposed in this paper,and the "k-partitioning strategy" storage model is constructed,its goal is to minimize the average number of occupied storage units of ship on the whole route.At the same time,different empirical cases are constructed based on the forecast results,and the solution is analyzed by CPLEX solver.The results show that: ?the model has a certain feasibility and effectiveness;?the smaller granularity of the partition is,and the higher utilization is for the storage units of the ship.The more detailed division of the particle size is more suitable for cases which the occupancy rate of the liner is larger.(3)Research on Rolling Loading Decision of All Routes under Predictive DeviationThe actual loading amount is different from the pre-allocation plan loading situation which is caused by prediction errors and random events.In this paper,we propose to improve the solution of pre-allocation on the whole route without trying to change the original distribution planning and meet the safety of the ship.Simultaneously,based on the whole route forecasting results,the empirical cases under different deviations are constructed and are solved by CPLEX solver.The results show that this method can be feasible for different deviations.