| The integration of international market has promoted the rapid development of containertransportation.The rapid growth of container throughput and the large-oriented of container vessels,which make the requirement of terminal handling efficiency improved.Beacause of the advantages of handling operation stability and efficiency,and low labor cost,etc,the automated container terminal has become an important direction for the development and construction of China's ports(such as Qingdao Port and Shanghai Yangshan Port).As the most flexible horizontal transport equipment in automated container terminal,AGV is responsible for the transportation of containers between the wharf apron and the yard.Thus reasonable L-AGV real-time dynamic scheduling can reduce the waiting time of quay crane,improve ship handling efficiency,and reduce terminal operating cost.First of all,based on the scheduling optimization of L-AGV for mixed yard operation system in automated container terminal,the comparative analysis of the typical automated container terminal handling operation system around the world was made,and the operating characteristics of the mixed yard system with cantilever crane and non-cantilever crane at Yangshan Port were described,so that the necessity of studying this operating system could be explained easily.Based on the analysis of the AGV operation system of container terminal,the research content of this paper was clearly defined,and the mix-cross scheduling mode of the container terminals and the group-based independent shipment mode were determined.Then a AGV dynamic scheduling model which under the group-based independent loading mode for mixed container yard operation system was proposed to minimize the makespan of all the tasks for quay crane.Considering the complexity of solving the L-AGV scheduling problem as well as the characteristics of the research problem,a L-AGV scheduling IM-?-Q algorithm was designed.Also,according to the uncertainty of container logistics system,a method of representation based on multidimensional state space of L-AGV and container task was carried out,and two actions were combined in one action strategy,which could show the systematic real-time state and achieve the dynamic scheduling of L-AGV.Then the accuracy and convergence speed of the algorithm were guaranteed through the improved ?-greedy exploration strategy.The determination rules for the learning factor and the discount factor of the algorithm were also proposed.Besides,the solving process was simplified with the help of discrete eventsimulation mechanism,which serve as reference for Q algorithm in the follow-up application of container terminal logistics system.Finally,the optimize performance of the IM-?-Q algorithm are verified by experimental analysis.Experiments show that: 1)The optimize performance of IM-?-Q algorithm is better.the quay crane and L-AGV waiting time reduce by more than 9% and 5% respectively than other 4 scheduling strategies,and gains 8.0% improvment compared to the accuracy of ordinary Q algorithm.2)The IM-?-Q algorithm can meet the dynamic scheduling requirement.which reduces the handling time by about 5% and increase the L-AGV utilization by more than 9% in contrast with the greedy dispatching method.3)With the increase of L-AGV,the quay crane operating time and ARMG waiting time decrease first and then slow down,the utilization of AGV shows a accelerating downward trend. |
PDF Full Text Request