Research On Prediction Method Of Truck Arrival Time In Transportation System

In the container dispatching system of a port,it not only helps to improve the service level of the container terminal and the operational efficiency of the truck scheduling,but also dynamically optimizes the order of loading and unloading containers during operaton,and reduces the number of reshuffles by predicting the arrival time of the truck.It is also very important for developing the container yard plan,improving the utilization rate of the machinery and equipment,and further increasing the overall efficiency of the terminal.In this thesis,the multi-source data in the container dispatching system is exploited to get the model of predicting the arrival time of the truck.The obtained model is successfully applied in a real container dispatching system.Firstly,for the GPS data of the trucks,a model of predicting the arrival time of the truck is developed based on an improved algorithm of K-nearest neighbor regression by consider:ing the possible misalignment of traffic conditions.Secondly,for entry and exit date of yard or terminal,a model of predicting the arrival time of the truck is proposed based on Kalman filter model.The proposed model is improved in the sense of prediction accuracy by reducing the jump between data us:ing linear interpolation.Finally,the two prediction models based on the different data sources are merged to get an adaptive weighted fusion prediction model,which reduces the dependence on a single data source and the maximal absolute error,and has strong robustness and high accuracy and improves the robustness and accuracy of the prediction results.The obtained results are well illustrated by successfully applying in a real container dispatching system of a famous international port.