Data Structure Design and Order-Picking Optimization for Irregular Warehouse

In order to calculate the distance between every two storage points, this paper proposed a data structure to store irregular warehouse layout diagram. A warehouse unit can be defined as four kinds, which are storage point, passage, node and block respectively. Storage point is a unit to store goods. Passage is an aisle by which the operator can access goods stored on storage point. Node is the intersection of a vertical passage and a horizontal passage. The number of nodes is evidently smaller than the number of storage points. So, we can compute out the shortest distance and route between every two nodes and save these data in disk. Then, when we need to calculate the shortest distance and route between two storage points, we can turn to these two storage points’ adjacent nodes. Because of the limit of the order-picking vehicle’s capacity, an order usually needs to be divided into several order-picking operations. In this paper, a hybrid genetic algorithm is proposed to optimizing order-picking. A chromosome consists of storage point’s order in an order and the order-picking beginning point. A chromosome determines the number of order-picking operations and the total distance. The chromosomes evolve by special designed crossover and mutation operators. The experimental results revealed this new hybrid genetic algorithm’s better performance. KeywordsHybrid Genetic Algorithm; Irregular Warehouse; Order-Picking; Data Structure; Warehouse layout diagram I. DESCRIPTION OF THE PROBLEM As we know many large-scale distribution centers or manufacturing enterprises have already owned automated storage and retrieval system (ASRS) which brings sharp decline in order-picking costs. While, ASRS is not suitable for some industries such as small manufacturing businesses, small warehouses, small distribution centers, supermarkets, and book stores. One reason is mini orders always occur in these industries. The other reason is that multi-shelves in these warehouses usually have different specifications and have no uniform arrangement in direction. The research of optimizing order-picking of irregular warehouse will benefit these industries. The aim of optimization is to find a shortest order-picking route for an order. II. DATA STRUCTURE DESIGNED FOR IRREGULAR WAREHOUSE LAYOUT DIAGRAM A warehouse has three important elements which are passages, SPs and nodes. Passage is an aisle by which the operator can access goods stored on storage point (SP). In order to simplify the problem, we assume that there be only two kinds of straight passage, one is vertical and the other is horizontal. Node is the intersection of a vertical passage and a horizontal passage. SP is a unit to store goods. Each SP has a unique number. In order to save these elements’ information, the warehouse is divided into m×n cells. A cell can be a passage unit, a SP, and a block unit. A matrix with m lines and n columns is used to save the information of these cells. The value of the matrix’s element can be 1, 2, 3, 4, 5, and 6. The value 1, 2, 3, and 4 means that the cell is a SP. The value 1 means this cell’s access point is in the up direction, i.e., the up direction in the layout diagram of this cell is a passage unit by which to access the goods stored in this SP is most convenient. The value 2 means the down direction, 3 denoting left direction, and 4 denoting right direction. The value 5 means that the cell is a passage unit, and 6 denoting a block unit. The matrix saved all the warehouse cells’ information. So, the information of all the passages and nodes can be obtained by virtue of this matrix. Other matrixes are used to save passage’s and node’s information. Two adjacent nodes are double nodes that they are located in a same passage and there is no other node located between them. Since the information of all the passages and nodes is obtained. All the adjacent nodes can be computed. A matrix with m lines and m columns is used to save the distance between every two nodes, where m is the total number of the nodes. Firstly, only the distance between every two adjacent nodes can be computed and saved in this matrix. Secondly, Floyd algorithm can be used to calculate the shortest distance between every two nodes. A SP’s adjacent nodes refer to its access point’s adjacent nodes. A SP’s access point is always a passage unit which has no more than two adjacent nodes. The access point is located between its two adjacent nodes and they are in the same passage. The distance between a SP and its adjacent nodes can be easily calculated. Since the shortest distance between every two nodes has been figured out, the shortest distance between any two SPs can be worked out by virtue of their adjacent 978-1-4244-4639-1/09/$25.00 ©2009 IEEE nodes. In this way, the distance of an order-picking route represented by SP’s number can be figured out. III. A HYBRID ALGORITHM FOR OPTIMIZATION OF ORDER-