CN115456537B - Warehouse picking path planning method and system - Google Patents
Warehouse picking path planning method and system Download PDFInfo
- Publication number
- CN115456537B CN115456537B CN202211144852.9A CN202211144852A CN115456537B CN 115456537 B CN115456537 B CN 115456537B CN 202211144852 A CN202211144852 A CN 202211144852A CN 115456537 B CN115456537 B CN 115456537B
- Authority
- CN
- China
- Prior art keywords
- cluster
- points
- point
- clustering
- path
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0835—Relationships between shipper or supplier and carriers
- G06Q10/08355—Routing methods
Abstract
The invention provides a warehouse picking path planning method, which comprises the steps of regarding each order number as a point, and converting a one-dimensional picking path planning problem into a two-dimensional point clustering problem; clustering all points for the first time by using a fast clustering algorithm; clustering the points which cannot be clustered after the first clustering by using a minimum rack clustering algorithm for the second time; performing multiple point exchange operations in the cluster by using a global optimization algorithm until all the clusters are traversed, so as to obtain a new cluster; the new cluster is output as a pick slip while the path of each cluster and the total path of the pick slip are calculated. The invention can sort a large number of orders received by the warehouse into a plurality of sub orders, so that the repeated paths among the sub orders are smaller, thereby completing the picking of the plurality of orders at one time, greatly reducing the total picking path and greatly improving the working efficiency.
Description
Technical Field
The invention relates to the technical field of warehouse logistics, in particular to a warehouse picking path planning method and system.
Background
Storage picking is the most complex link in the storage system and plays an important role in the whole supply chain system. With the development of the e-commerce market, the storage throughput is larger and larger, and the storage picking demand is increased rapidly in certain special time periods such as e-commerce festival. Rapid sorting and shorter pick paths are the focus of warehouse management, and therefore, reasonably efficient warehouse pick path planning methods are becoming increasingly important.
At present, large-scale warehouse goods are picked by seeding type goods picking, the system receives a certain number of orders and packages the orders to be sent to a goods picking person, and the goods picking person picks the goods according to the order from small goods shelf codes to large goods shelf codes. But large-scale storage order volume is huge, and a plurality of pickers' pick route can be a large amount of repetition, and this distance that can increase the transport when picking, not only can increase the cost of transportation, work efficiency also can reduce because of the collision problem simultaneously.
Therefore, there is an urgent need to provide a warehouse picking path planning method to solve the above-mentioned problems of the conventional method during picking.
Disclosure of Invention
Therefore, the technical problem to be solved by the invention is to overcome the problems existing in the prior art, and a warehouse picking path planning method and system are provided, which greatly reduce the total picking path and greatly improve the working efficiency.
In order to solve the technical problems, the invention provides a warehouse picking path planning method, which comprises the following steps:
s1, regarding each order number as a point based on one-dimensional arrangement characteristics of a goods shelf, and converting a one-dimensional goods picking path planning problem into a two-dimensional point clustering problem;
s2, performing first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
s3, clustering the points which cannot be clustered after the first clustering for the second time by using a minimum goods shelf clustering algorithm to obtain a second cluster;
s4, merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
s5, outputting the optimized cluster as a pick list, and simultaneously calculating the path of each cluster and the total path of the pick list.
In one embodiment of the present invention, a method for first clustering all points using a fast clustering algorithm includes:
s21, defining a picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, y represents the maximum shelf number of the order number, and obtaining the path length of any point;
s22, arranging all points in descending order according to the path length of the links to obtain a descending order table of the points, and taking the first point in the descending order table as an initial point C of the first cluster i,j ＝C 1,1 Wherein C i,j A j-th point representing an i-th cluster;
s23, point C 1,1 The corresponding path section is taken as a limiting section, and the path sections in the descending order table are not included in the limiting sectionThe points are removed, a new descending list is obtained,
s24, sequentially taking the points and the points C from a new descending list according to the number of defects in the cluster 1,1 Clustering to obtain a first cluster, wherein one cluster is a pick sub-sheet;
s25, storing the clustered clusters, and repeating S22-S24 on the rest points until the clusters cannot be obtained.
In one embodiment of the invention, when clustering all points for the first time, the points with the larger path length are preferably clustered so that the points with the larger path length are all in the same pick order.
In one embodiment of the present invention, the method for clustering the points which cannot be clustered after the first clustering for the second time by using the minimum rack clustering algorithm comprises the following steps:
s31, taking the remaining non-clustered points, and carrying out ascending arrangement on all the points according to the x values of the points to obtain an ascending list;
s32, sequentially taking points from the ascending list to form clusters according to the number of orders required in each order picking sub-list, and obtaining a second cluster.
In one embodiment of the present invention, before the global optimization algorithm is used to perform multiple point exchange operations in the cluster, the cluster is screened, whether the percentage of the number of repeated points contained in the cluster to the total number of points is greater than or equal to a preset value is determined, if yes, the cluster is removed from the cluster, the point exchange operation is performed on the remaining clusters in the cluster, and if no, the point exchange operation is performed on the clusters in the cluster.
In one embodiment of the present invention, a method for performing a plurality of point exchange operations within a cluster using the global optimization algorithm includes:
s41, starting from a cluster i, searching all adjacent clusters, namely a cluster A, if the cluster A does not contain any cluster, i+1, and returning to screening operation, wherein the initial value of i is 1;
s42, taking the mth point C in the cluster i i,m As the exchanged point, the initial value of m is 1;
s43, taking the nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1;
s44, if C j,n If the cluster j is the exchange point, n+1, if n is the maximum value, j+1 returns to S43;
s45, calculating initial paths and Z of the cluster i and the cluster j 1 Exchange C within cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m ；
S46, calculating new paths and Z of the cluster i and the cluster j 2 If Z 2 <Z 1 The point exchange is carried out in the two clusters, otherwise, the point exchange is canceled, and if n is the maximum value, j+1 is returned to S43;
s47, if j is the maximum value, m+1 is returned to S42;
s48, if m is the maximum value, i+1 is returned to the screening operation;
s49, if i is the maximum value, the global optimization is completed.
In addition, the invention also provides a warehouse picking path planning system, which comprises:
the path planning conversion module is used for regarding each order number as a point based on one-dimensional arrangement characteristics of the goods shelves and converting a one-dimensional order picking path planning problem into a two-dimensional point clustering problem;
the first clustering module is used for carrying out first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
the second clustering module is used for carrying out second clustering on the points which cannot be clustered after the first clustering by using a minimum goods shelf clustering algorithm to obtain a second cluster;
the global optimization module is used for merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
and the order picking list output module is used for outputting the optimized cluster group into an order picking list and simultaneously calculating the path of each cluster and the total path of the order picking list.
In one embodiment of the present invention, the method for the first clustering module to perform the first clustering on all points by using a fast clustering algorithm includes:
s21, defining a picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, y represents the maximum shelf number of the order number, and obtaining the path length of any point;
s22, arranging all points in descending order according to the linked path length values to obtain a descending order table of the points, and taking the first point in the descending order table as an initial point C of the first cluster i,j ＝C 1,1 Wherein C i,j A j-th point representing an i-th cluster;
s23, point C 1,1 The corresponding path interval is taken as a limiting interval, points, which are not included in the limiting interval, of the path interval in the descending list are removed, a new descending list is obtained,
s24, sequentially taking the points and the points C from a new descending list according to the number of defects in the cluster 1,1 Clustering to obtain a first cluster, wherein one cluster is a pick sub-sheet;
s25, storing the clustered clusters, and repeating S22-S24 on the rest points until the clusters cannot be obtained.
In one embodiment of the present invention, the method for the second clustering module to cluster the points which cannot be clustered after the first clustering using the minimum rack clustering algorithm comprises:
s31, taking the remaining non-clustered points, and carrying out ascending arrangement on all the points according to the x values of the points to obtain an ascending list;
s32, sequentially taking points from the ascending list to form clusters according to the number of orders required in each order picking sub-list, and obtaining a second cluster.
In one embodiment of the present invention, the method for performing multiple point exchange operations within a cluster by using the global optimization algorithm by the global optimization module includes:
s41, starting from a cluster i, searching all adjacent clusters, namely a cluster A, if the cluster A does not contain any cluster, i+1, and returning to screening operation, wherein the initial value of i is 1;
s42, taking the mth point C in the cluster i i,m As the exchanged point, the initial value of m is 1;
s43, taking the nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1;
s44, if C j,n If the cluster j is the exchange point, n+1, if n is the maximum value, j+1 returns to S43;
s45, calculating initial paths and Z of the cluster i and the cluster j 1 Exchange C within cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m ；
S46, calculating new paths and Z of the cluster i and the cluster j 2 If Z 2 <Z 1 The point exchange is carried out in the two clusters, otherwise, the point exchange is canceled, and if n is the maximum value, j+1 is returned to S43;
s47, if j is the maximum value, m+1 is returned to S42;
s48, if m is the maximum value, i+1 is returned to the screening operation;
s49, if i is the maximum value, the global optimization is completed.
Compared with the prior art, the technical scheme of the invention has the following advantages:
the method and the system for planning the picking path of the warehouse can classify a large number of orders received by the warehouse into a plurality of sub orders, so that repeated paths among the sub orders are smaller, the picking of the plurality of orders is finished at one time, the total picking path is greatly reduced, and the working efficiency is greatly improved.
Drawings
In order that the invention may be more readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof that are illustrated in the appended drawings.
Fig. 1 is a flow chart of a warehouse picking path planning method according to an embodiment of the present invention.
Detailed Description
The present invention will be further described with reference to the accompanying drawings and specific examples, which are not intended to be limiting, so that those skilled in the art will better understand the invention and practice it.
Referring to fig. 1, an embodiment of the present invention provides a warehouse picking path planning method, including:
s1, regarding each order number as a point based on one-dimensional arrangement characteristics of shelves, converting a one-dimensional order picking path planning problem into a two-dimensional point clustering problem, and defining an order picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, and y represents the maximum shelf number of the order number;
s2, performing first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
s3, clustering the points which cannot be clustered after the first clustering for the second time by using a minimum goods shelf clustering algorithm to obtain a second cluster;
s4, merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
s5, outputting the optimized cluster as a pick list, and simultaneously calculating the path of each cluster and the total path of the pick list.
In S1, each order number includes a plurality of goods, that is, one order number requires a picker to pick up the goods to a plurality of shelf numbers. Based on the one-dimensional arrangement of the goods shelves, the pickers pick goods from the minimum goods shelf number of the order, sequentially pass the goods shelf number where the goods are located, and end from the maximum goods shelf number. Based on this feature, the single order number pick path interval and path length are independent of the number of orders contained, depending only on the minimum and maximum container numbers. At this time, the data dimensions of all order numbers are the same, i.e. only two shelf numbers are reserved, and one order can be treated as one point. The problem of order sorting is thus translated into a clustering problem of geometrically clustered points, which is logically easier to handle.
The path length of each order and the corresponding point are represented as follows:
Z＝max(X)-min(X)
P＝(min(X),max(X))
S＝[min(X),max(X)]
wherein x= { X 1 ,x 2 ,…,x m M represents the number of orders contained in the same order number, x m And representing the shelf number of the goods with the serial number of m, wherein X is the shelf number set contained in the order number. Z is the path length of the order. P is the point coordinate, where the x value is the minimum shelf number and the y value is the maximum shelf number. S is the path interval of the point.
Preferably, the software can be run on a windows platform, an appearance l form containing the bill is placed under the directory of the software, and the bill name is input to import the order, wherein the bill contains 2000 orders and 5406 cargos. Each order contains one to a plurality of goods, each corresponding to a shelf number. The distribution center needs to combine every 20 orders into one pick sub-order for a total of 100. Each order retains only a minimum shelf number and a maximum shelf number, which are equal if there is only one item for an order. An order is regarded as a point, the minimum shelf number of the order is regarded as the x value of the point, the maximum shelf number of the order is regarded as the y value of the point, so that 2000 orders are converted into 2000 points, namely 20 orders in the original problem are clustered into 20 points, and a one-dimensional order picking problem is changed into a two-dimensional clustering problem.
Wherein, in S2, the method for first clustering all points using the fast clustering algorithm includes:
s21, each order corresponds to 2 values, a minimum shelf number and a maximum shelf number, the path length of all points is calculated for each point, and the path length is the maximum shelf number minus the minimum shelf number.
S22, arranging all the points in descending order according to the path length to obtain a descending order table, wherein the first point in the descending order table is the point with the longest path length, and the first point in the table is taken as the initial point C of the first cluster 1,1 。
S23, taking point C 1,1 Corresponding path zoneThe interval is defined as a limiting interval, points, which are not included in the limiting interval, of the path interval in the descending list are removed, a new descending list is obtained, after the updated descending list is obtained, the points in the list are reduced, and the points are still arranged in descending order, at the moment, 19 points are taken from the descending list, namely C 1,2 -C 1,20 And finishing the first cluster.
S24, repeating the step S23 until a certain cluster cannot be clustered. Failure to cluster means that after the cluster has reselected the initial point, there are no 19 points in the updated descending list, i.e., a myriad of sufficient points. At this time, after the points in the descending list and the initial point of the cluster are marked, the quick clustering algorithm is not participated. It should be noted that the inability of the cluster to cluster does not mean that the fast clustering algorithm is unable to continue clustering, e.g., cluster C 78 Not clustered, but cluster C 79 ，C 80 ，C 81 Clusters can still be formed. This is because the points that cannot be clustered are excluded, the remaining points continue to participate in step S22, and the limited interval of the initial points is changed after the initial points are selected, so that the points in the descending list are also changed, and the points in the list can still be clustered when the points are equal to or greater than 20.
S25, when no point participates in the rapid clustering algorithm, the algorithm is ended. About 90% of the dots have been clustered at this time, resulting in a first cluster, which now contains 92 clusters.
Based on the characteristics of the single paths of the pickers, when the fast clustering algorithm is used for clustering, the initial point in each cluster is the point with the largest path length in the pickers, and the path intervals of the rest points are all contained in the limited interval of the initial point. When the clusters are formed, the path of the whole pick bill can be determined by selecting an initial point, and the path properties of the clusters are not affected when other points are added into the clusters. This may allow clusters with a large initial point path to accommodate as many points as possible with other paths that are larger, thereby reducing the pick path as a whole. It also prevents the point of larger path from becoming confused within the cluster of smaller path, such that the cluster path increases in multiples.
Based on the arrangement characteristics of the storage shelves, the sorted goods picking sub-orders are classified by a rapid clustering algorithm, and the total path of each sub-order is determined by the order with the largest path length. After the data of all the points is obtained, the points with larger path length are preferably selected to be clustered, so that the points with larger path length are all in the same pick sub-list. The reason for this is to avoid that a certain order with a large path length gets mixed into a pick sub-list with a smaller path, resulting in a suddenly longer pick sub-list path for a certain short path. When the first clustering is performed on all the points, the points with larger path length values are clustered preferably, so that the points with larger path length values are all in the same pick sub-list.
After the first clustering, the fast clustering algorithm is able to sort about 90% of the total order into pickers for system pickers based on the defined interval characteristics described in the fast clustering algorithm. The rest orders cannot be classified by the fast clustering algorithm, but the shelf numbers of the orders are generally close, and the clustering algorithm has little influence on the final total path in the actual test. In view of the need to deal with the situation that warehouse throughput is fluctuated in certain time periods, a clustering algorithm with higher speed and smaller path is adopted for processing, so that the invention adopts a minimum rack clustering algorithm to cluster the non-clustered points for the second time.
Wherein in S3, the method for clustering the points which cannot be clustered after the first clustering for the second time by using the minimum rack clustering algorithm includes: and taking the remaining non-clustered points, and carrying out ascending arrangement on all the points according to the x value of the points to obtain an ascending list, namely carrying out ascending arrangement on the remaining orders according to the minimum shelf number of each order. And sequentially taking points from the ascending list to form clusters according to the number of orders required in each order picking sub-list, namely the number of points required to be contained in the clusters, so as to obtain a second cluster, wherein the second cluster comprises 8 clusters.
Because the greedy characteristics of the fast clustering algorithm and the minimum goods shelf clustering algorithm can only obtain a suboptimal solution close to an optimal solution, the total path of the clusters still has a further reduced space, and therefore the invention provides a global optimization algorithm based on the point exchange idea. The idea of point switching is: in the clustered clusters, a point in any cluster is denoted as C i,m Finding a point C in other clusters j,n If crossing in two clustersAfter two points are changed, the sum of the two clusters of paths becomes short, and the two clusters of paths are recorded as one effective point exchange.
Before the global optimization algorithm is used for carrying out multiple point exchange operations in the cluster, the cluster is screened, whether the percentage of the repeated points contained in the cluster to the total points is larger than or equal to a preset value is judged, if yes, the cluster is removed from the cluster, the rest clusters in the cluster are subjected to point exchange operations, and if no, the clusters in the cluster are subjected to point exchange operations. Preferably, the preset value of this embodiment is 25%.
Based on this, in S4, the method for performing a plurality of point exchange operations within a cluster using the global optimization algorithm includes:
s41, starting from a cluster i, searching all adjacent clusters, namely a cluster A, if the cluster A does not contain any cluster, i+1, and returning to screening operation, wherein the initial value of i is 1;
s42, taking the mth point C in the cluster i i,m As the exchanged point, the initial value of m is 1;
s43, taking the nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1;
s44, if C j,n If the cluster j is the exchange point, n+1, if n is the maximum value, j+1 returns to S43;
s45, calculating initial paths and Z of the cluster i and the cluster j 1 Exchange C within cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m ；
S46, calculating new paths and Z of the cluster i and the cluster j 2 If Z 2 <Z 1 The point exchange is carried out in the two clusters, otherwise, the point exchange is canceled, and if n is the maximum value, j+1 is returned to S43;
s47, if j is the maximum value, m+1 is returned to S42;
s48, if m is the maximum value, i+1 is returned to the screening operation;
s49, if i is the maximum value, the global optimization is completed.
The idea of the global optimization algorithm is that effective exchange points may exist in the search between two clusters, which is essentially that the algorithm consumes a long time by performing traversal search on the points, so that clusters need to be screened, and some clusters with low possibility of existence of the exchange points are removed, so that the time of searching the algorithm is reduced. Because the quick pick algorithm will preferentially sort points that are close to each other into a cluster, and because many orders contain the same goods, i.e., the same shelf number, multiple duplicate points in a cluster will occur. The more repeat points means that the cluster is more compact in real physical space, with a lower likelihood of having an exchange point. So that a small fraction of the switching points can be properly discarded for faster operation speeds. True data testing also verifies this guess, and typically the switching points are contained within clusters where there are no duplicate points or few. The 25% screening conditions set by the algorithm were derived from multiple experiments. After screening, the algorithm trades off for nearly 3-fold improvement in operating speed at very low performance cost compared to no screening. The screening condition can be reset according to actual conditions, for example, the computing power of a hardware platform is sufficient, and non-screening can be set.
Two clusters are adjacent, where there is an intersection between the path intervals defined as two clusters. And the switching point can only exist in two intersecting clusters. This is because if the two clusters have no intersection in the path space, the two clusters are not physically adjacent to each other, and switching between any two points in the two clusters causes the respective path sections to become longer, resulting in a longer total path, and therefore, there is necessarily no switching point in the non-adjacent clusters. After one cluster is selected, the operation of finding the exchange point is only performed in the adjacent clusters, so that the search range can be greatly reduced, and the operation speed of the algorithm is increased. The path interval of a cluster has the same meaning as the path interval of a point, i.e., s= [ min (X), max (X) ], where X is the set of all shelf numbers contained in the cluster.
In general, most point switching operations have no effect on path changes for both clusters, or result in an increase in the overall path, with the effective point switching operation being a small fraction of the overall switching operation. Frequent switching points inside clusters increases the running time of the algorithm and also increases the redundancy of programming. Thus, in programming, the actual point operations and calculations should be performed on a partial copy of the cluster, without affecting the original cluster, and only when a valid point exchange operation occurs, will this operation be passed to the cluster execution. From the whole program, the optimization of the cluster is performed in real time along with the running of the program. The path length of a cluster has the same meaning as the path length of a point, i.e., z=max (X) -min (X), where X is the set of all shelf numbers contained within the cluster.
A global optimization algorithm approximately executes point operation with 1.5 times of clusters, and the point operation is further close to a theoretical optimal solution on the basis of greedy thought. The execution of one optimization operation takes about 5 times of the time of the rapid picking path algorithm, and the specific optimized path is related to the number of orders, the number of clusters and the number of points in the clusters. It is quite worthwhile to trade more effort and part of the time for a shorter total path from the real scene, so that further optimization of the cluster is necessary. In practice, global optimization may be performed multiple times. This is because each point swap operation performed is not necessarily the best swap operation to shorten the path, but is unpredictable by the algorithm. Meanwhile, the operation of the algorithm on the cluster is essentially traversing points, and the points still have the potential of being exchanged again, which cannot be achieved by one traversing. The total path of the cluster after one global optimization has room for optimization. However, in practical testing, if multiple optimizations are performed, the path that can be shortened from the second optimization is exponentially reduced, and the run time is greatly extended, which is not compatible with the rapid pick requirements of large warehouse. Thus, global optimization is performed only once by default in the present invention.
For example, taking 92 clusters contained in the first cluster and 8 clusters contained in the second cluster as an example, the first cluster and the second cluster are combined to obtain a total cluster, and global optimization is performed on the total cluster. In this data, since 100 clusters should be divided into 20 points per cluster, the maximum value of i, j is 100 and the maximum value of m, n is 20 in the following steps. The method comprises the following specific steps:
screening the total clusters, and if the number of the repeated points contained in a certain cluster is more than 5, eliminating the cluster from the clusters. At this point 76 remain in the cluster.
And secondly, searching all adjacent clusters from the cluster i, and marking the clusters as a cluster A. The initial value of i is 1.
Step three, if A does not contain any cluster, i+1, returning to step one
Step four, taking the m-th point C in the cluster i i,m As the point to be exchanged, m has an initial value of 1.
Fifthly, taking an nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1.
Step six, if C j,n In cluster j is the switching point, then n+1; if n=20, j+1, return to step five.
Step seven, calculating initial paths and Z of the cluster i and the cluster j 1 。
Step eight, exchanging C in cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m . The path length of the original cluster may be changed after the switching point.
Step nine, calculating new paths and Z of the cluster i and the cluster j 2 。
Step ten, if Z 2 <Z 1 And (3) performing the point exchange in the two clusters, otherwise, canceling the point exchange, and returning to the step five if n=20 and j+1.
Step eleven, if j=100, then m+1, return to step four.
Step twelve, if m=100, i+1, return to step one.
Step thirteen, if i=100, global optimization is completed, and a total of 133 point exchanges are performed.
Without loss of generality, cluster C is selected 26 Demonstrating the operation of finding the exchange points:
cluster C 26 The path interval of (a) is [213,1033 ]]Cluster C 28 The path interval of (a) is [263,1034 ]]Intersection of two clusters is [263,1033 ]]Then adjacent.
Further, cluster C 26 Is 820, cluster C 28 Is 771. The current two clusters have a total path length of 1591.
Further, cluster C is taken 26 First point C in 26,1 As exchanged points, in turn to cluster C 28 The 26 points in the first exchange point C perform exchange operation 28,1 。
Further, the positions of the two points in the original cluster are exchanged. Cluster C at this time 26 The path interval of (a) is [234,1034 ]]The path length is 800; cluster C 28 The path interval of (a) is [213,1034 ]],821. The new path length is 1621, which is increased compared to the original path length and is therefore not an efficient point exchange.
Further, point C 26,1 And point C 28,2 The switching position, the path length is found not to be reduced after calculation, and the next point C is continued 28,3 Exchange until the last point C of the cluster 28,26 No effective switching point has yet been found.
Further, cluster C is taken 26 The second point C in 26,2 As exchanged point, re-cluster C 28 And (5) internal retrieval of the exchange points which may exist. If not, then cluster C is taken 26 Third point C in 26,3 As exchanged points until a valid exchange point is retrieved.
Further, when taking point C 26,17 As exchanged point, get Point C 28,2 As a switching point. Cluster C 26 The path interval of (a) is [213,1034 ]]The path length is 821; cluster C 28 The path interval of (a) is [243,1033 ]],790. The new path length is 1611, which is reduced by 10 compared to the original path length and is therefore an effective point exchange, this operation being performed in the original cluster.
The above examples are merely for further explanation of the point exchange operation. An effective point exchange minimizes the path drop distance to 1 and to 164. Notably, multiple point exchanges within two clusters also occur frequently, which can significantly reduce the path and drop of the two clusters. For this document data, a total of 133 point exchanges occur, eventually reducing the total path by 2501, which is equivalent to the total path sum of 3 to 4 general clusters, while the total running time of the global optimization algorithm is about 4s, which is an effective performance improvement.
Wherein in S5 the total cluster contains 100 clusters, each cluster containing 20 points. The total cluster and the order are essentially the same, each order contains 100 order sheets, one for each cluster; a pick sub-order contains 20 points, with a unique order corresponding to one point. In contrast, a unique order may contain multiple repeated orders, corresponding to multiple shelf numbers that are hidden during conversion to points, and the warehouse needs to display these hidden shelf numbers on the pick sub-order against the original bill when outputting the pick bill. Because each unique order contains a different number of repeat orders, the number of shelf numbers contained in each pick sub-order may vary widely, but all are on the same path as their corresponding cluster paths.
Further, the distribution center may calculate the total path distance. The total path distance is summed up from the path distances of all pick sub-sheets. The path distance of each pick sub-order is equal to the largest shelf number minus the smallest shelf number of all shelf numbers within the order. The individual pick sub-unit path distances generated by the system tend to be large to small, and the situation that the individual pick sub-unit paths are extremely large can occur, so that the total path distance is reduced. Through real data test, the invention has effectiveness.
Further, the distribution center may perform pick-and-dispatch. The system generates the order picking sheets, and the repeated paths among the order picking sheets are small, so that the existing order picking equipment or personnel can be reasonably utilized, and blockage, wrong picking and the like caused by path overlapping are prevented.
Further, parameters of the system can be changed to adapt to different warehouse types and different time periods, and portability is good.
The method and the system for planning the picking path of the warehouse can classify a large number of orders received by the warehouse into a plurality of sub orders, so that repeated paths among the sub orders are smaller, the picking of the plurality of orders is finished at one time, the total picking path is greatly reduced, and the working efficiency is greatly improved.
The following describes a warehouse picking path planning system disclosed in the embodiments of the present invention, and the warehouse picking path planning system described below and the warehouse picking path planning method described above may be referred to correspondingly.
The embodiment of the invention also provides a warehouse picking path planning system, which comprises:
the path planning conversion module is used for regarding each order number as a point based on one-dimensional arrangement characteristics of a shelf, converting a one-dimensional order picking path planning problem into a two-dimensional point clustering problem, and defining an order picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, and y represents the maximum shelf number of the order number;
the first clustering module is used for carrying out first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
the second clustering module is used for carrying out second clustering on the points which cannot be clustered after the first clustering by using a minimum goods shelf clustering algorithm to obtain a second cluster;
the global optimization module is used for merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
and the order picking list output module is used for outputting the optimized cluster group into an order picking list and simultaneously calculating the path of each cluster and the total path of the order picking list.
The warehouse picking path planning system of this embodiment is used to implement the warehouse picking path planning method described above, so that the detailed description of the system can be found in the example section of the warehouse picking path planning method described above, and therefore, the detailed description of the system can refer to the description of the corresponding examples of the various sections, and will not be described herein.
In addition, since the warehouse picking path planning system of the present embodiment is used to implement the warehouse picking path planning method described above, the functions thereof correspond to those of the method described above, and will not be described herein.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations and modifications of the present invention will be apparent to those of ordinary skill in the art in light of the foregoing description. It is not necessary here nor is it exhaustive of all embodiments. And obvious variations or modifications thereof are contemplated as falling within the scope of the present invention.
Claims (4)
1. A warehouse picking path planning method, comprising:
s1, regarding each order number as a point based on one-dimensional arrangement characteristics of a goods shelf, and converting a one-dimensional goods picking path planning problem into a two-dimensional point clustering problem;
s2, performing first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
s3, clustering the points which cannot be clustered after the first clustering for the second time by using a minimum goods shelf clustering algorithm to obtain a second cluster;
s4, merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
s5, outputting the optimized cluster as a picking bill, and calculating the path of each cluster and the total path of the picking bill;
the method for carrying out first clustering on all points by using the fast clustering algorithm comprises the following steps:
s21, defining a picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, y represents the maximum shelf number of the order number, and obtaining the path length of any point;
s22, arranging all points in descending order according to the path length of the links to obtain a descending order table of the points, and taking the first point in the descending order table as an initial point C of the first cluster i,j ＝C 1,1 Wherein C i,j A j-th point representing an i-th cluster;
s23, point C 1,1 The corresponding path interval is defined asDetermining the section, removing points in the descending list, which are not included in the limited section, to obtain a new descending list,
s24, sequentially taking the points and the points C from a new descending list according to the number of defects in the cluster 1,1 Clustering to obtain a first cluster, wherein one cluster is a pick sub-sheet;
s25, storing the clustered clusters, and repeating S22-S24 on the rest points until the clusters cannot be obtained;
the method for clustering the points which cannot be clustered after the first clustering for the second time by using the minimum goods shelf clustering algorithm comprises the following steps:
s31, taking the remaining non-clustered points, and carrying out ascending arrangement on all the points according to the x values of the points to obtain an ascending list;
s32, sequentially taking points from the ascending list to form clusters according to the number of orders required in each order picking sub-list, so as to obtain a second cluster;
the method for performing multiple point exchange operations in the total cluster by using the global optimization algorithm comprises the following steps:
s41, starting from a cluster i, searching all adjacent clusters, namely a cluster A, if the cluster A does not contain any cluster, i+1, and returning to screening operation, wherein the initial value of i is 1;
s42, taking the mth point C in the cluster i i,m As the exchanged point, the initial value of m is 1;
s43, taking the nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1;
s44, if C j,n If the cluster j is the exchange point, n+1, if n is the maximum value, j+1 returns to S43;
s45, calculating initial paths and Z of the cluster i and the cluster j 1 Exchange C within cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m ；
S46, calculating new paths and Z of the cluster i and the cluster j 2 If Z 2 <Z 1 The point exchange is carried out in the two clusters, otherwise, the point exchange is canceled, and if n is the maximum value, j+1 is returned to S43;
s47, if j is the maximum value, m+1 is returned to S42;
s48, if m is the maximum value, i+1 is returned to the screening operation;
s49, if i is the maximum value, the global optimization is completed.
2. A warehouse picking path planning method as claimed in claim 1, wherein: when clustering all the points for the first time, the points with larger path length are clustered preferentially, so that the points with larger path length are all in the same pick sub-list.
3. The warehouse picking path planning method according to claim 1, characterized in that before the global optimization algorithm is used for carrying out a plurality of point exchange operations in a total cluster, the cluster is screened, whether the percentage of the repeated points contained in the cluster to the total points is larger than or equal to a preset value is judged, if yes, the cluster is removed from the cluster, the rest of the clusters in the cluster are subjected to the point exchange operation, and if no, the clusters in the cluster are subjected to the point exchange operation.
4. A warehouse picking path planning system, comprising:
the path planning conversion module is used for regarding each order number as a point based on one-dimensional arrangement characteristics of the goods shelves and converting a one-dimensional order picking path planning problem into a two-dimensional point clustering problem;
the first clustering module is used for carrying out first clustering on all points by using a fast clustering algorithm to obtain a first cluster;
the second clustering module is used for carrying out second clustering on the points which cannot be clustered after the first clustering by using a minimum goods shelf clustering algorithm to obtain a second cluster;
the global optimization module is used for merging the first cluster and the second cluster into a total cluster, and performing multiple point exchange operations in the total cluster by using a global optimization algorithm until all the total clusters are traversed, so as to obtain an optimized cluster;
the order picking list output module is used for outputting the optimized cluster group into an order picking list and simultaneously calculating the path of each cluster and the total path of the order picking list;
the method for the first clustering module to perform first clustering on all points by using a fast clustering algorithm comprises the following steps:
s21, defining a picking path interval of each order number as [ x, y ], wherein x represents the minimum shelf number of the order number, y represents the maximum shelf number of the order number, and obtaining the path length of any point;
s22, arranging all points in descending order according to the path length of the links to obtain a descending order table of the points, and taking the first point in the descending order table as an initial point C of the first cluster i,j ＝C 1,1 Wherein C i,j A j-th point representing an i-th cluster;
s23, point C 1,1 The corresponding path interval is taken as a limiting interval, points, which are not included in the limiting interval, of the path interval in the descending list are removed, a new descending list is obtained,
s24, sequentially taking the points and the points C from a new descending list according to the number of defects in the cluster 1,1 Clustering to obtain a first cluster, wherein one cluster is a pick sub-sheet;
s25, storing the clustered clusters, and repeating S22-S24 on the rest points until the clusters cannot be obtained;
the method for the second clustering module to cluster the points which cannot be clustered after the first clustering by using the minimum rack clustering algorithm comprises the following steps:
s31, taking the remaining non-clustered points, and carrying out ascending arrangement on all the points according to the x values of the points to obtain an ascending list;
s32, sequentially taking points from the ascending list to form clusters according to the number of orders required in each order picking sub-list, so as to obtain a second cluster;
the method for performing multiple point exchange operations in the total cluster by the global optimization module by using the global optimization algorithm comprises the following steps:
s41, starting from a cluster i, searching all adjacent clusters, namely a cluster A, if the cluster A does not contain any cluster, i+1, and returning to screening operation, wherein the initial value of i is 1;
s42, taking the mth point C in the cluster i i,m As the exchanged point, the initial value of m is 1;
s43, taking the nth point C in the cluster j from the cluster A j,n As the switching point, the initial values of j and n are 1;
s44, if C j,n If the cluster j is the exchange point, n+1, if n is the maximum value, j+1 returns to S43;
s45, calculating initial paths and Z of the cluster i and the cluster j 1 Exchange C within cluster i and cluster j i,m And C j,n Is the position of (C) i,n And point C j,m ；
S46, calculating new paths and Z of the cluster i and the cluster j 2 If Z 2 <Z 1 The point exchange is carried out in the two clusters, otherwise, the point exchange is canceled, and if n is the maximum value, j+1 is returned to S43;
s47, if j is the maximum value, m+1 is returned to S42;
s48, if m is the maximum value, i+1 is returned to the screening operation;
s49, if i is the maximum value, the global optimization is completed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211144852.9A CN115456537B (en) | 2022-09-20 | 2022-09-20 | Warehouse picking path planning method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211144852.9A CN115456537B (en) | 2022-09-20 | 2022-09-20 | Warehouse picking path planning method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115456537A CN115456537A (en) | 2022-12-09 |
| CN115456537B true CN115456537B (en) | 2023-04-25 |
Family
ID=84305014
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211144852.9A Active CN115456537B (en) | 2022-09-20 | 2022-09-20 | Warehouse picking path planning method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115456537B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447344A (en) * | 2018-10-26 | 2019-03-08 | 国网天津市电力公司 | Based on the repairing stationary point of Distribution Network Failure big data and method for optimizing route and system |
| CN110807559A (en) * | 2019-11-07 | 2020-02-18 | 陕西科技大学 | Order batching and picking path combined optimization method |
| CN111582582A (en) * | 2020-05-08 | 2020-08-25 | 西安建筑科技大学 | Warehouse picking path optimization method based on improved GA-PAC |
| CN111860957A (en) * | 2020-06-18 | 2020-10-30 | 浙江工业大学 | Multi-vehicle type vehicle path planning method considering secondary distribution and balance time |
| CN113673922A (en) * | 2021-07-09 | 2021-11-19 | 合肥工业大学 | Fishbone type warehouse layout-based multi-vehicle picking path problem optimization method and system |
| CN113848936A (en) * | 2021-10-13 | 2021-12-28 | 京东科技信息技术有限公司 | Path planning method and device, electronic equipment and computer readable storage medium |
-
2022
- 2022-09-20 CN CN202211144852.9A patent/CN115456537B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447344A (en) * | 2018-10-26 | 2019-03-08 | 国网天津市电力公司 | Based on the repairing stationary point of Distribution Network Failure big data and method for optimizing route and system |
| CN110807559A (en) * | 2019-11-07 | 2020-02-18 | 陕西科技大学 | Order batching and picking path combined optimization method |
| CN111582582A (en) * | 2020-05-08 | 2020-08-25 | 西安建筑科技大学 | Warehouse picking path optimization method based on improved GA-PAC |
| CN111860957A (en) * | 2020-06-18 | 2020-10-30 | 浙江工业大学 | Multi-vehicle type vehicle path planning method considering secondary distribution and balance time |
| CN113673922A (en) * | 2021-07-09 | 2021-11-19 | 合肥工业大学 | Fishbone type warehouse layout-based multi-vehicle picking path problem optimization method and system |
| CN113848936A (en) * | 2021-10-13 | 2021-12-28 | 京东科技信息技术有限公司 | Path planning method and device, electronic equipment and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115456537A (en) | 2022-12-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109447317B (en) | Warehouse shortest path optimization method based on simultaneous picking of multiple persons | |
| CN103914506B (en) | Data searcher, date storage method and data retrieval method | |
| Burnwal et al. | Scheduling optimization of flexible manufacturing system using cuckoo search-based approach | |
| Gharehgozli et al. | A decision-tree stacking heuristic minimising the expected number of reshuffles at a container terminal | |
| US7627604B2 (en) | Method for handling tree-type data structure, information processing device, and program | |
| US10642257B2 (en) | Tree search-based scheduling method and electronic apparatus using the same | |
| CN108792387B (en) | Shelf hit method, apparatus, server and medium | |
| CN105929690B (en) | A kind of Flexible Workshop Robust Scheduling method based on decomposition multi-objective Evolutionary Algorithm | |
| JPH0410165A (en) | Optimum plan generating system | |
| CN107193896A (en) | A kind of diagram data division methods based on cluster | |
| CN111754176B (en) | Two-stage intelligent order sorting method for multiple mobile shelves | |
| CN111027902B (en) | Unmanned warehouse goods transportation matching method and device | |
| CN112561225B (en) | Flexible job shop scheduling method based on marker post co-evolution algorithm | |
| CN107209768A (en) | Method and apparatus for the expansible sequence of data set | |
| CN103765381A (en) | Parallel operation on B+ trees | |
| CN116341765B (en) | Automatic order source searching and splitting method and system | |
| US7403944B2 (en) | Reduced comparison coordinate-value sorting process | |
| US9495654B2 (en) | Stack handling operation method, system, and computer program | |
| CN112785132A (en) | Task allocation method for multi-robot mobile shelf for intelligent warehouse | |
| CN106802787B (en) | MapReduce optimization method based on GPU sequence | |
| CN115456537B (en) | Warehouse picking path planning method and system | |
| JPWO2012043217A1 (en) | Method, program and apparatus for grouping a plurality of elements | |
| Lin et al. | A TABU SEARCH ALGORITHM TO MINIMIZE TOTAL WEIGHTED TARDINESS FOR THE JOB SHOP SCHEDULING PROBLEM. | |
| US8417652B2 (en) | System and method for effecting optimization of a sequential arrangement of items | |
| CN116011723A (en) | Intelligent dispatching method and application of coking and coking mixed flow shop based on Harris eagle algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |