Truck Arrival Time Prediction Method Research Of Container Terminals

With increasingly fierce competition between ports, container terminals put increasingly attention to improve service level of their ports. Concentrated arrival brings convenience to layout of the terminal handling scheduling, but causes inconvenience in owner and truck fleet arranging transportation and leads to congestion of container terminal port gates and excessive transportation pressure of port area. Some ports take the random arrival strategies of export containers to improve their competitiveness and provide facilitation services. But the arrival time is unknown under this mode which will brings new challenge to Container terminal operations. Traditionally, terminal planning "berth planâ†'stowage planâ†'Container Yard planâ†'Container into the territory plan", which will not be applicable any more. Because port cannot determine the volume and arrival time of daily export containers of particular voyages, export containers can stack dispersedly based on their actual arrival time and the container yard availability, which challenges ccontainer yard allocation greatly. Therefore, the arrival law of external trucks and the prediction of arrival time are beneficial to improve the operation efficiency and service level.In this article, both the static and dynamic arrival time prediction model are established. The static model use v-SVM based on history data to predict the arrival time and transport routes. The dynamic model is established to use the dynamic running data to renew the prediction result which is based on SVM prediction and Kalman. Research results show that the Kalman filter algorithm can effectively predict truck arrival time, compared with the method of average travel time and SVM algorithm used alone. Significantly improved prediction accuracy and relative prediction error with the distance of time increases and decreases gradually stabilized is found.