| In the process of automatic container terminal operation,horizontal transportation with AGV(Automatic Guided Vehicle)is the key that links quay crane and yard crane.Its operational efficiency affects the efficiency of quay crane operation,which in turn determines the berthing time of container ships.Therefore,in order to improve the operation efficiency of AGV in automatic container terminal,a multi-AGV dynamic dispatching model based on Qlearning algorithm is proposed to shorten the total operation time of quay cranes and the invalid operation time of AGV,which takes the battery energy of AGV as a constraint.Firstly,the existing research of multi-AGV scheduling problems shows that the existing multi-AGV scheduling solutions mostly used static scheduling,such as genetic algorithm and heuristic algorithm,which lacks consideration of the real-time system state of the terminal and does not consider or resolve the impact of unexpected situations in the process of operation on multi-AGV scheduling.The existing research also neglected the actual situation that electric drive AGV needs charging,which meant they did not take AGV battery level constraint into account.Secondly,in order to realize the rapid response of multi AGV scheduling in container terminal to the dynamic change of system environment,a multi-AGV dynamic scheduling model based on Q-learning algorithm is proposed,in which each AGV can dynamically select the next transportation task according to the real-time status of the terminal operation.However,the complex environment of automatic container terminal makes the dimension of state space in the model a large number,leading to the slowly converging of the Q-learning algorithm.Thus,the system state is mapped to BP neural network to solve the dimension catastrophe and improve the convergence speed of the algorithm.In the end,a Q-learning scheduling model combining BP neural network is established,with the aims of the shortest total operation time of quay crane and invalid operation time of AGV.With the above research,aiming at the charging problem of electric drive AGV,the influence of AGV charging process and opportunity charging strategy on AGV scheduling is analyzed.Based on the actual situations that AGV’s lithium battery discharge curves and nonlinearity battery charging curves under heavy load and no load conditions are different,the unloading operation model of automatic container terminal is built.With the objective to minimize the total task time and charging time,the model is built on the premise of taking the opportunity charging strategy.By adjusting the width and position of the opportunity charging strategy interval,the control experiment is carried out to prove the influence of the charging process of AGV and find the best opportunity charging interval.Finally,combined with the optimal interval of the opportunity charging strategy of AGV,a simulation experiment is performed on the AGV dynamic scheduling model based on QL-BP.The method is compared with the traditional genetic algorithm and the scheduling method based on the longest queue principle by solving the same example.The result shows that the total operation time of the method was shortened by 9.32%、0.75% and the ineffective operation time of AGV by 24.9% and 22.68% comparing to the other two methods,which verifies the effectiveness of the method,which verifies the effectiveness of the method in improving the terminal and AGV operation efficiency.Based on the above research,proposing to take the battery capacity constraints of the AGV into consideration,this article mainly implements the dynamic scheduling of horizontal transportation of the AGV in the automated container terminal and proves the effect of AGV charging process on terminal operation efficiency,which makes it more in line with the actual operation of the automated terminal.The paper can provide some ideas for the subsequent research on the automation of container terminals and trigger more discussion and research. |
PDF Full Text Request