Strategy Optimization Research Of Zone Automated Order Picking System

The orders become more and more diversified and the demand of customer service becomes higher and higher in the modern logistics distribution center. All of above need a faster order fulfillment time. Picking is an important operation in the order handling process and the working efficiency of the picking operation is the key factor which affects the total working effieicncy of the distribution center. According to statistics, the picking time accounts for about35%of the total order fulfillment time. Therefore, more and more distribution centers become to apply automated order picking system. Compared to manual order picking system, the automated order picking system is faster and less prone to mistakes, so it is suitable for the order in which there are many kinds of SKU and the total number of SKUs is small.The dispenser system is widely used in the distribution center among all kinds of automated order picking system since it is faster and needs less space. Usually, the dispensers are divided into several zones, so different zones can pick one order simutaniously and the order picking time will be reduced. Therefore, the research on the zone automated order picking system will have a great significance.However, most researchers focus on the manual order picking system as so far, the literatures about the automated order picking system are much fewer. Worse than that, most literatures about the automated order picking system focus on the mechanical improvement and the equipment selection, little refers to the optimization of picking strategy. Because of all above reasons, this paper concludes three picking strategies which have most important impact on the working efficiency of the zone automated order picking system:item assignment strategy, merging sequence assignment strategy and the order picking sequence assignment strategy. Based on this, the tabu search algorithm, dynamic cluster algorithm, greedy heuristic algorithm, dynamic programming method and genetic algorithm are used to optimize the strategies to minimize the order fulfillment time. The main content and the achievement of this paper are as follows. (1) Analyze the working process of the zone automated order picking system, builds the mathematical model of the system based on parallel picking and serial merging method.In the model of the zone automated order picking system, the order picking time is divided into two parts:merging time and delay time. The merging time is decided by parameters of the equipment and the number of SKUs in the current order. If the system can run steady, the merging time is a constant; whereas the delay time is a variable since it is decided by the order structure of the current order and prior order.The impact factors which will influence the order picking time are given based on the model of the system.(2) Considering the sub problem of item assignment, converts the optimizing object from minimizing order picking time to minimizing the summation of the delay time, and then design two algorithms to solve the problem.First, the delay factor is proposed to represent the delay time under a special situation. Under the special situation, the delay times of the following zones in the prior order and prior zones in the current order are zero. The delay factor is proven to have the same vary trend with the delay time for each zone, so it can be used to solve the model of the sub problem to simply the complexity.The tabu search algorithm based on item exchange and the dynamic cluster algorithm based on item shift are proposed respectively aimed at whether the number of kind of SKUs is fixed in each zone. The simulation verified the validity of the two algorithms.(3) Considering the sub problem of merging sequence optimization, the necessary condition of changing the merging sequence is proposed firstly, and then the greedy heuristic algorithm based on the necessary condition is proposed to solve the problem.First, we analyze the impact of different merging sequences on the order picking time, and then build the system model regarding the merging sequences of zones as variables. The model is abstracted as a general system model and expressed as 1(group)|rη=c(r-1)j+tη|Cmax. This model is proven to be a NP-hard model with3-partition theory.The necessary condition of changing the merging sequence of a zone is proposed, and then a greedy heuristics algorithm combined with the dynamic programming method is designed to solve the model. The simulation result shows the order picking time can be reduced and picking efficiency is improved by the algorithm.(4) Regarding sub problem of the order picking sequence optimization, we improve the self-adaptive genetic algorithm to solve it.First, we analyze the impact of the order picking sequence on the order picking time, and then build the mathematical model regarding the order picking sequence as variable. This model can be simpled to be TSP problem model. At last, the improved self-adative genetic algorithm is proposed to solve the model. There are two differences between the proposed algorithm and the traditional self-adaptive genetic algorithm:First, the proposed algorithm brings HD distance into it. The HD distance of two genes represents the degree of difference between them. We choose the genes which have longer HD distance with other genes to form the initial solutions group. This will help avoiding local best solution. Second, the proposed algorithm improves the crossing probability and the variation probality. When the evolution process is stopped, increasing the crossing probability and the variation probability of the better genes will help avoiding local best solution. The simulation result shows the validity of the proposed algorithm.