| Currently domestic demands for AGV are booming, in addition to the traditional AGV usage in tobacco and other industries, the usage in other industries is also increasing. Because of the particularity of some industries, several enterprises need to adopt AGV for works related to carry, in the actual application process, considering the factor of cost, small and medium-sized domestic enterprises mostly use this kind of AGV, which has low independent control ability and depend on centralized control system. In this process a series of problems happened because of improper scheduling and vehicle collision, so how to realize efficient scheduling of the AGV has become an urgent problem.At present, there is rare research on dispatch of AGV in the automated storage and retrieval system which only use AGV for works related to carry. In view of this situation, this article puts forward the off-line solution of ex ante forecasting and adjusting before happening. First, each task (from loading to unloading) is considered as a task. When there is only one AGV, the problem comes down to TSP problem, then we calculate the execution order of tasks, and at the same time the shortest path is calculated. For the scheduling problem of multiple AGVs, we use the means of solving occupied multiple-vehicle routing problem and the short board effect thought. When a number of tasks are given, the time of completing all tasks depends on the AGV which takes the longest time, therefore the shortest time to complete all tasks as the goal, through the optimization of the task assignment, by making the longest time which AGV cost as little as possible, the whole time completing all tasks is reduced accordingly. Finally the goal of completing all batches of tasks in the shortest time and the multiple-AGV route planning are achieved in the case of certain AGVs in the AS/RS. Considering the timeliness of scheduling, we choose the particle swarm algorithm, which converges relatively fast, then use the discrete strategies appropriately to make it suitable for the solution of our problems. Secondly, to resolve the conflicts appeared on the multi-AGV path in the scheme we have got, we put forward the logical thinking which is similar to the Gantt chart, aiming at solving the problem of conflicts and road-cross section. Finally, we put the final scheduling scheme to the traffic management module by means of the path list for storage, and realize the task of prediction. After prior forecasting, we start appointing AGV to perform a task according to the task list and the list of paths. Prediction costs a little time, which can well meet the actual needs of enterprises.To verify the validity of the method, this paper makes some simulation experiments. We test our solution in these experiments while using single AGV and multiple AGVs, and the results show that in single-AGV case, this method can achieve the optimization of task execution path and the task order sorting on condition of the path cost as small as possible. For multi-AGV test, the result shows that the method used in this paper can effectively achieve the completion target of all the tasks in the shortest time and the task distribution for AGV system more balanced.Finally, this paper discussed the method to solve the AGV scheduling problem, and gives the direction for further study. |
PDF Full Text Request