Picking path planning method for small and medium-sized planar bins in omni-channel mode

Abstract

The invention relates to a sorting path planning method for small and medium-sized planar warehouses in an omni-channel mode, and the method comprises the steps: selecting a current sorting order and a corresponding target sorting object from a plurality of sorting orders distributed to a plurality of sorting objects, and carrying out the sorting of the current sorting order and the corresponding target sorting object in goods allocation information, detecting a plurality of target sorting channels in which replenishment SKUs corresponding to the sorting SKUs associated with the current sorting order are stored; based on the sorting waiting time required for the target sorting object to enter the target sorting channels for sorting, sequentially determining the order of entering the multiple target sorting channels, and based on the order and a path strategy determined according to the type of the sorting order, generating a target sorting path corresponding to the current sorting order; and the target sorting paths corresponding to the plurality of sorting orders are planned in sequence to obtain a planning result, so that the problems of serious occupation conflict, long time consumption and low sorting efficiency of sorting operation in an omni-channel mode in related technologies are solved, the total sorting operation time is minimized, and the sorting operation efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of smart logistics and warehouse management technology, and in particular to a picking path planning method for small and medium-sized flat warehouses under an omni-channel model. Background Art

[0002] Picking operations are carried out under a certain warehouse layout and a known shelf allocation plan. Pickers walk along the picking aisles and pick the required SKUs from the shelves to be picked. In the omni-channel retail model, picking orders include online orders and offline orders. Among them, online orders are characterized by multiple varieties and small batches. Picking generally requires visiting multiple discrete shelves. Since the picking quantity of each product is small, the operation time at a single shelf is relatively short. Offline orders are characterized by a small variety and large batches. When picking offline orders, only a small number of shelves need to be visited. Since the batch size of goods is large, the operation time at a single shelf is relatively long.

[0003] In related technologies, in order to ensure space utilization, the width of the picking channel usually only allows a single picker to pass through, and the crossing and overtaking actions of multiple pickers are generally completed in the cross channel at the entrance and exit of the picking channel. When multiple pickers complete the picking tasks of online and offline picking orders in the warehouse during the same period, it is inevitable that more than one picker will enter the same picking channel. If they go in opposite directions, congestion will occur at the meeting point. Even if multiple pickers go in the same direction in the same picking channel, due to the different location access requirements of the picking orders, the length of time each picker occupies the channel is inconsistent. When the rear picker has completed the picking needs in the channel and the front picker has not yet completed it, the rear picker cannot catch up due to the limitation of the channel width, and congestion occurs. It can be seen that in a multi-picker omni-channel flat warehouse where online and offline picking tasks are carried out simultaneously, picking congestion is prone to occur. When the picking aisle where a picker's to-be-picked location is located is accessed by other pickers, he or she may be blocked by other pickers moving in the same or opposite direction, and therefore must stop and wait. The waiting caused by aisle congestion is essentially a conflict of occupation of the same picking aisle by different pickers at the same time.

[0004] At present, no effective solution has been proposed to the problems of serious conflicts in picking operations, long picking times and low picking efficiency in the omni-channel model in related technologies. Summary of the Invention

[0005] An embodiment of the present application provides a picking path planning method for small and medium-sized flat warehouses in an omni-channel mode, so as to at least solve the problems in the related art of serious picking operation conflicts, long picking operation time and low picking efficiency in the omni-channel mode.

[0006] The embodiment of the application provides a picking path planning method for a small and medium-sized flat warehouse under a full-channel mode, comprising: after receiving a picking request for picking a plurality of picking orders, obtaining previously planned storage location allocation information, wherein the picking request comprises a picking order type and a picking SKU of the picking order, the storage location allocation information is generated by using a genetic algorithm to plan storage locations for replenishment SKUs considering a goods outbound frequency parameter, a goods correlation parameter and a relocation cost of the replenishment SKUs, and selecting a current picking order and a corresponding target picking object from a plurality of the picking orders allocated to a plurality of picking objects, and detecting a plurality of target picking channels in the storage location allocation information, wherein the target picking channels store the replenishment SKUs corresponding to the picking SKU associated with the current picking order; determining an order of entering the plurality of target picking channels based on a picking waiting time required by the target picking object to enter the target picking channel for picking, and generating a target picking path corresponding to the current picking order based on the order and a path strategy determined according to the picking order type; and planning the target picking paths corresponding to the plurality of picking orders in sequence to obtain a path planning result.

[0007] Compared with the related art, the picking path planning method for a small and medium-sized flat warehouse under a full-channel mode provided by the embodiment of the application selects a current picking order and a corresponding target picking object from a plurality of picking orders allocated to a plurality of picking objects, and detects a plurality of target picking channels in the storage location allocation information, wherein the target picking channels store replenishment SKUs corresponding to a picking SKU associated with the current picking order; determines an order of entering the plurality of target picking channels based on a picking waiting time required by the target picking object to enter the target picking channel for picking, and generates a target picking path corresponding to the current picking order based on the order and a path strategy determined according to the picking order type; and plans the target picking paths corresponding to the plurality of picking orders in sequence to obtain a planning result, thereby solving the problems of serious occupation conflict of picking operations under a full-channel mode, long picking operation time, and low picking efficiency in the related art, relieving the space-time conflict in the picking channel and reducing the waiting time in the picking process under a multi-operation type and multi-picking object operation scenario, minimizing the total picking operation time, and improving the picking operation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0008] Figure 1 FIG. 1 is a hardware structure block diagram of a terminal of the picking path planning method for a small and medium-sized flat warehouse under a full-channel mode according to the embodiment of the application; Figure 2 FIG. 2 is a flowchart of the picking path planning method for a small and medium-sized flat warehouse under a full-channel mode according to the embodiment of the application; Figure 3 FIG. 3 is a schematic diagram of a node network according to the embodiment of the application; Figure 4This is a schematic diagram of a spatiotemporal network according to an embodiment of the present application; Figure 5 An algorithm flow chart for constructing a spatiotemporal network according to a preferred embodiment of the present application; Figure 6 A schematic diagram of the path planning process of a preferred embodiment of the present application; Figure 7 A schematic diagram of a picking path generated after path optimization in an embodiment of the present application; Figure 8 This is a schematic diagram of the spatiotemporal network after path optimization in an embodiment of the present application; Figure 9 This is a schematic diagram of the structure of a double-layer coding entity in an embodiment of the present application; Figure 10 This is a schematic diagram of the genotype structure of a double-layer coded individual in an embodiment of the present application; Figure 11 This is a schematic diagram of the benchmark individuals of the primary population of the embodiment of the present application; Figure 12 This is a schematic diagram of the crossover operation in accordance with an embodiment of the present application; Figure 13 This is a schematic diagram of a variant transport operation according to an embodiment of the present application; Figure 14 A schematic diagram of an order-product relationship matrix according to an embodiment of the present application; Figure 15 A schematic diagram of a commodity association matrix according to an embodiment of the present application; Figure 16 A schematic diagram of a product association network according to an embodiment of the present application; Figure 17 This is a structural block diagram of a picking path planning device for small and medium-sized flat warehouses in an omni-channel mode according to an embodiment of the present application. DETAILED DESCRIPTION

[0009] In order to make the objectives, technical solutions and advantages of this application more clear, this application is described and illustrated below with reference to the accompanying drawings and embodiments.

[0010] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. Taking running on a terminal as an example, Figure 1 This is a hardware structure diagram of the terminal of the picking path planning method for small and medium-sized flat warehouses in the omni-channel mode of the embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0011] The memory 104 can be used to store a computer program, such as the computer program corresponding to the picking path planning method for small and medium-sized flat warehouses in the full-channel mode in the embodiments of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implements the above-mentioned method. The memory 104 can include a high-speed random access memory, a non-volatile memory, and a memory remotely arranged with respect to the processor 102. The transmission device 106 is used to receive or send data via a network. In an example, the transmission device 106 can include a network adapter or a radio frequency module.

[0012] The embodiment provides a picking path planning method for small and medium-sized flat warehouses in the full-channel mode running on the terminal, Figure 2 is a flowchart of the picking path planning method for small and medium-sized flat warehouses in the full-channel mode according to the embodiments of the present application, as Figure 2 shown, the flowchart includes the following steps:

[0013] Step S201, after receiving a picking request for picking a plurality of picking orders, obtaining previously planned storage location allocation information, wherein the picking request includes a picking order type and a picking SKU of the picking order, and the storage location allocation information is generated by using a genetic algorithm to plan storage location allocation of replenishment SKUs considering a goods delivery frequency parameter, a goods correlation parameter, and a relocation cost of the replenishment SKUs of replenishment orders.

[0014] In the embodiment, the picking path planning method is performed to pick SKUs required by the currently received sales order from the flat warehouse and timely deliver goods from the corresponding warehouse to meet the picking demand, improve the warehouse use efficiency, and reduce the picking and storage costs. In the embodiment, the orders to be picked are purchase or demand orders generated at the moment, and the goods (one good corresponds to one SKU) meeting the demand of a plurality of picking orders are stored in the corresponding flat warehouse before the current moment, and the stored goods are replenished and stored according to the replenishment order generated by estimating how many picking orders will be generated in the future (corresponding to the current) and the required goods at the past moment (corresponding to the current moment). That is, after the corresponding replenishment order arrives at the past moment, all goods corresponding to the replenishment order are planned according to the set storage location allocation method (for example, a method of using a genetic algorithm to plan storage location allocation according to the storage location information of the flat warehouse and the goods delivery frequency parameter, the goods correlation parameter, and the relocation cost of the SKU corresponding to the replenishment order) to obtain the storage location allocation information for representing the storage information of the goods in the flat warehouse (for example, the SKU corresponding to one good, the goods aisle where the SKU is located, the storage location number, and the number of the goods on the storage location).

[0015] In this embodiment, after receiving a corresponding picking request, the planning of picking paths corresponding to the multiple received picking orders is initiated. At this time, it is necessary to determine the picking type (online order or offline order) of the corresponding picking order and the items to be picked for each picking order (i.e., the corresponding SKU). In this embodiment, determining the picking type of the corresponding picking order is used to select the path strategy used to pick the items required for the picking order. In this embodiment, the path strategies involved include a return strategy, an S-shaped path strategy, and a combination strategy of the return strategy and the S-shaped path strategy. The return strategy first generates a sequence based on the aisles where the corresponding storage locations of the goods in the picking task order are located, enters each aisle from the picking aisle entrance in a preset order, and returns to the aisle entrance after picking all the required goods in the aisle. The S-shaped path strategy starts from the warehouse entrance and traverses all required storage locations in a serpentine manner according to the aisle arrangement order. In this embodiment, due to the characteristics of offline orders with a small number of varieties and large quantities, the return strategy is used for path planning, while the combination strategy is used for online orders.

[0016] In step S202 , a current picking order and a corresponding target picking object are selected from a plurality of picking orders assigned to a plurality of picking objects, and a plurality of target picking channels storing replenishment SKUs corresponding to the picking SKU associated with the current picking order are detected in the shelf allocation information.

[0017] In this embodiment, the picking of the picking SKUs in the picking list by the picking object is achieved by accessing the storage location where the picking SKU is stored, and the access to the storage location is achieved by accessing the picking channel. Therefore, the picking path planning can be decomposed into two levels: determining the access sequence of the picking channels and determining the access sequence of the storage locations in the channels. At the same time, in this embodiment, it is set that only a single picking object is allowed to enter the same picking channel at any time, and the access sequence of the picking channels of the picking objects is adjusted so that different picking objects at the same time are located in different picking channels to carry out operations, thereby reducing the picking waiting time. In this embodiment, because only a single picking object is allowed to enter the same picking channel at a certain time, the picking object can be placed in a different picking channel at the same time to carry out operations. Each picking object enters the corresponding picking channel, so determining which picking object enters first is also determining the access order; it can be understood that the picking order assigned first and the picking object that receives the picking order must be arranged first, that is, the picking object that is arranged with the picking order first is given priority in picking channel arrangement. In this embodiment, by taking the picking order assigned first as the current picking order and the current picking order as the target, the picking channels and the access order of the picking channels that need to be visited are planned, and then the picking order assigned first (corresponding to the current picking order) and the picking order that are assigned first (corresponding to the current picking order) are determined. SKU的储款成功没到卡里,包括:SKU的储存数以及储款成功时、货行计划进行储行行,包括:SKU的储存数以及储款成功时、货行计划进行储行行,包括:SKU的储存数以及储款成功时、货行计划进行储行行,包括:SKU的储存数等,储款成功时等。 In the information, the replenishment SKUs and the corresponding shelf locations corresponding to all the picking SKUs of the current picking order are detected, and then according to the picking channel where each shelf location is located, multiple target picking channels that store the replenishment SKUs corresponding to the picking SKUs associated with the current picking order are determined. Afterwards, based on the actual picking situation in the current flat warehouse (including which picking channels have corresponding picking objects for picking operations and which picking channels are idle) and the target picking channels to be picked, the order in which the target picking objects visit multiple target picking channels is planned, and the picking path corresponding to the current picking order is planned.

[0018] Step S203, based on the picking waiting time required for the target picking object to enter the target picking channel for picking, the order of entering multiple target picking channels is determined in sequence, and based on the order and the path strategy determined according to the picking order type, the target picking path corresponding to the current picking order is generated.

[0019] In this embodiment, the goal is to reduce the waiting time of the picking object at the entrance and exit of the picking channel. The picking waiting time of the picking object at the entrance and exit of the picking channel is determined according to the number of other picking objects queued before the target picking object and the picking time required for each picking object to complete the corresponding picking. At the same time, the time it takes for the target picking object to walk from the current position to each picking channel (corresponding to the distance between the current position and the entrance and exit of each picking channel) also determines the length of the picking waiting time. In this embodiment, for the determined target picking object and the current picking order, the goal is also to reduce the waiting time. Therefore, the method of determining The picking waiting time of the target picking object entering each target picking channel of all the target picking channels associated with it is used to adjust the order in which the target picking object enters the corresponding target picking channels, that is, the sequence. For example: the channels to be picked by the target picking object currently include channel 1, channel 2, channel 3, channel 4 and channel 5. There are no other picking objects to be picked in all channels. The channel to be entered first is determined according to the distance between the target picking object and the five channels. For example: channel 1, after the target picking object completes picking of channel 1 and exits, the corresponding picking waiting time is recalculated. At this time, if If there is at least one channel among channel 2, channel 3, channel 4 and channel 5 that has no other picking objects to be picked, the channel with the shortest distance to the current position (corresponding to the walking time, that is, the corresponding picking waiting time) will be used as the channel for the next picking of the target picking object. If there are other picking objects in channel 2, channel 3, channel 4 and channel 5, the channel with the shortest picking waiting time will be selected as the channel for the next picking of the target picking object based on the number of picking objects waiting in line under each channel and the picking time required to complete the corresponding picking. At the same time, in this embodiment, when selecting the target to be entered, After picking the channel, picking will be carried out in the target picking channel according to the path strategy corresponding to the picking order type corresponding to the current picking order, that is, entering and exiting the corresponding target picking channel according to the corresponding path strategy. For example: when the current picking order is an offline order, the return strategy is adopted for picking, that is, entering from the channel entrance of the target picking channel close to the warehouse entrance, and then exiting from the channel entrance after completing the picking, forming a partial path in the target picking channel, and then constructing all partial paths of the target picking objects in the sequentially sorted target picking channels into the target picking path corresponding to the current picking order;It is understandable that when picking a certain picking order, there will be other picking objects in multiple picking channels of the flat warehouse to pick other picking orders, and other picking orders may be picking orders assigned before the certain picking order, or they may be picking orders assigned after the certain picking order. In other words, at the same picking moment, several or all of the multiple picking orders will be executed, but for several or all of the picking orders executed at the same picking moment, the picking of these picking orders will be carried out in sequence according to the priority of the picking orders (which represents the order in which the picking orders are assigned, and the picking order assigned first has the highest priority). That is, the first picking aisle that enters first for multiple picking orders is selected sequentially, and the order of subsequent picking aisles after the first picking aisle will change due to the number of picking SKUs in the corresponding picking aisle for each picking order. For example, if there are 5 picking SKUs in the first picking aisle 1 corresponding to picking order 1, and 2 picking SKUs in the first picking aisle 3 corresponding to picking order 2, then the picking object of picking order 2 can enter the next picking aisle 2 before the picking object corresponding to picking order 1, while the picking object corresponding to picking order 1 enters the next picking aisle 3 later.

[0020] In some optional implementations, the following steps are also performed: Step 1. At the current moment, multiple picking orders are assigned to multiple preset picking objects, that is, picking orders are assigned to each picking object; Step 2. After the assignment is completed, the picking object that first receives the picking task (corresponding to the first receiving the picking order) is used as the target picking object, and the corresponding picking order is the current picking order; Step 3. The picking order type of the current picking order is identified, and the corresponding storage locations of all picking SKUs of the current picking order and the target picking channel that needs to be entered to access the corresponding storage locations are retrieved; Step 4. It is determined whether the target picking channel to be entered is has been occupied by other picking objects. If not, proceed to step 5, otherwise proceed to step 6; Step 5, calculate the distance between the location of the target picking object and all unoccupied target picking channels respectively, select the nearest target picking channel as the next target picking channel to be visited, and generate a route according to the path strategy corresponding to the type of picking order; If there is more than one unoccupied picking channel, execute step 7; Step 6, count the number of picking objects waiting to enter each picking channel, select the picking channel with the least number of picking objects waiting as the next picking channel to be visited, and generate a route according to the path strategy corresponding to the type of picking order. If there is more than one picking channel with the least number of people waiting for picking objects, execute step 7; Step 7, select the picking channel with the largest number of picking SKUs to be picked as the next picking channel to be visited, and generate a route according to the path strategy corresponding to the picking order type. If there is more than one picking channel with the largest number of picking SKUs to be picked, select the picking channel closest to the entrance and exit of the flat warehouse; Step 8, remove the picking SKUs that have been picked from the current picking order, and then complete the picking orders of the picking plan in sequence until all picking orders have been planned.

[0021] Step S204 , planning target picking paths corresponding to multiple picking orders in sequence to obtain path planning results.

[0022] Through the above steps S201 to S204, the problems of serious picking operation conflicts, long picking operation time and low picking efficiency in the omni-channel mode in the related technology are solved, and the time and space conflicts in the picking channel are alleviated in the operation scenarios of multiple operation types and multiple picking objects, and the waiting time in the picking process is reduced, so that the total picking operation time is minimized and the picking operation efficiency is improved.

[0023] In some embodiments, a current picking order and a corresponding target picking object are selected from multiple picking orders assigned to multiple picking objects, including the following steps: after assigning multiple picking orders to multiple picking objects, determining the allocation order of picking orders received by each picking object, and selecting the picking order with the earliest allocation order to obtain the current picking order, and using the picking object with the earliest allocation order as the target picking object.

[0024] In some embodiments, detecting, in the shelf allocation information, multiple target picking lanes storing replenishment SKUs corresponding to the picking SKU associated with the current picking order is performed by:

[0025] Step 21, gridding the plane warehouse corresponding to the cargo location allocation information into a node network, wherein the node network includes picking channels, picking channel nodes and cross-channel nodes. A picking channel node is associated with two replenishment SKUs assigned to the same picking channel, and the cross-channel node is used to represent one of the entrance and exit of each picking channel.

[0026] In this embodiment, reference Figure 3 The node network shown in the figure rasterizes the NM-1 planar warehouse model into a node network and discretizes the traversable area into path nodes, including cross-channel nodes and picking channel nodes. The path nodes in the picking channel are all picking channel nodes. In addition to the cross-channel nodes at the warehouse entrance and exit, each picking channel entrance and exit has a cross-channel node. The intersection and overtaking of different pickers (picking objects) are only allowed to occur at the cross-channel nodes. The waiting area is also located at the cross-channel nodes at the entrance and exit of the picking channel. The picking channel index k is converted from the cargo channel index i according to the following formula: , I represents the cargo lane set; in this embodiment, the path node of the warehouse entrance and exit of the flat warehouse is set to o=0, and the remaining nodes are numbered starting from the cross-channel node of the picking channel k=1 entrance, increasing with the cargo location index, and the cross-channel node of the picking channel k=1 exit is also included, and so on, to complete the generation of all path node numbers; construct the two-dimensional coordinates of the path node o (o x ,o y ), horizontal axis o x The value of is the index k of the corresponding picking channel, and the vertical coordinate of the picking channel node is the corresponding shelf index p. In particular, the two-dimensional coordinate of node o=0 is set to (0,0), the vertical coordinate of the cross-channel node at the entrance of each picking channel is set to 0, and the vertical coordinate of the cross-channel node at the exit is set to p+1. In this embodiment, a replenishment SKU is stored in a shelf, that is, a replenishment SKU can be referred to by a shelf index. At the same time, a picking channel node corresponds to a shelf index and is associated with at least one replenishment SKU.

[0027] Step 22 : Retrieve a replenishment SKU whose SKU is one of the picking SKUs associated with the current picking order from the replenishment SKUs associated with the picking channel nodes of all picking channels, and determine the picking channel node associated with the retrieved replenishment SKU.

[0028] In this embodiment, by picking the SKU, a search is performed in the shelf index representing the replenishment SKU. When the shelf index corresponding to the replenishment SKU matching the picking SKU is retrieved, it means that the picking channel node corresponding to the shelf index is storing at least one replenishment SKU matching the picking SKU, and the picking channel where the picking channel node is located is the target picking channel corresponding to the current picking order.

[0029] Step 23: select the picking channel having at least one picking channel node as the target picking channel.

[0030] Through the above steps 21 to 23, the corresponding storage locations and picking channels for the items to be picked for each picking order are determined, providing a data basis for planning the picking paths corresponding to the picking orders.

[0031] In some embodiments, based on the waiting time required for a target picking object to enter a target picking aisle for picking, the order of entering multiple target picking aisles is determined sequentially, and based on the order and a path strategy determined according to the picking order type, a target picking path corresponding to the current picking order is generated, which is achieved through the following steps:

[0032] Step 31, obtain a space-time network generated by sequential picking planning for multiple picking orders, wherein the space-time network includes multiple space-time arcs corresponding to the multiple picking orders, the space-time arc includes at least one space-time sub-arc corresponding to the corresponding space-time state, the space-time sub-arc includes picking space-time sub-arc, waiting space-time sub-arc and walking space-time sub-arc, and the space-time state includes walking, waiting and picking.

[0033] In this embodiment, the spatiotemporal network is generated by preliminary picking planning for multiple picking orders. Specifically, multiple picking orders are assigned to corresponding picking objects (corresponding pickers), and then the priorities of the corresponding picking orders are determined according to the order in which the picking tasks are received. Then, picking planning is performed for multiple picking orders in sequence according to the priority. In the planning process, a congestion handling strategy that allows only a single picking object to enter the same picking channel at any time is used as a hard constraint to plan the picking paths corresponding to multiple picking orders with the goal of not causing congestion. The position and spatiotemporal state (including walking, waiting, and picking) of each picking object during the picking operation of the picking order are characterized, and the time of each spatiotemporal state is quantified, so that the state of the picking object in the plane warehouse at a certain moment is characterized as a set of spatiotemporal arcs, thereby obtaining the corresponding spatiotemporal network. The following is based on Figure 4 The space-time network shown in the figure describes the space-time status of the two pickers during the picking process as follows: the path nodes corresponding to the waiting picking location of picker A are 5, 16, 26, and 27, and the path nodes corresponding to the waiting picking location of picker B are 4 and 12. Before the picking path is adjusted, the access order of the picking channels corresponding to the two picking orders is 0→1→3→4→0 and 0→1→2→0 respectively. Both pickers need to enter channel 1 for operation. Picker A starts from node o=0 at time t=0 and arrives at path node o=1 at time t=1.5. According to the combined path strategy, he returns to node o=1 at t=12.5 after visiting node o=5 to pick the SKU, occupying picking channel 1 during this period, and then he visits the remaining nodes , and returns to the warehouse entrance at t=53; picker B starts from o=0 at time t=1 and arrives at node o=1 at t=2.5. Since channel 1 is occupied by picker A at this time, he can only wait at node o=1 for 10s, wait for picker A to leave and then enter channel 1, then complete the remaining picking tasks, and return to o=0 at t=44; it should be noted that, in this embodiment, the picking path of each picking order in the space-time network is a picking path that considers avoiding channel congestion, but does not consider the waiting time caused by avoiding congestion, that is, there is a waste of waiting time caused by channel congestion, and there is still room for optimization of picking path planning, that is, considering reducing the congestion waiting time to achieve the goal of minimizing the total picking operation time.

[0034] Step 32, based on the space-time network and greedy strategy, respectively determine the target waiting space-time sub-arc generated by the current entry into each target picking channel, and select the target picking channel with the shortest waiting time corresponding to the target waiting space-time sub-arc to obtain the current intended picking channel to be entered, wherein the target waiting space-time sub-arc is used to represent the space-time arc generated by the target picking object walking from the current position to the cross-channel node as the target entrance, and waiting at the cross-channel node. The target entrance includes the corresponding entrance and exit in the intended picking channel, the channel entrance closest to the current position.

[0035] In this embodiment, the target waiting time-space sub-arc includes the walking time-space arc of the target picking object walking from the current position to the target entrance, and also includes the waiting time-space arc waiting at the target entrance. When there are no other picking objects queued at the target entrance of the target picking channel, the corresponding waiting time-space arc is empty, and the target waiting time-space sub-arc includes the walking time-space arc; in this embodiment, the selection of the target entrance is determined according to the corresponding path strategy, and is affected by the exit when exiting the previous picking channel. For example: when the combination strategy and the S-shaped path strategy are used to pass through the current picking channel, the corresponding current position is the exit of the previous picking channel. At this time, the current position is closest to the intended picking channel at the channel entrance at the end of the intended picking channel away from the plane warehouse entrance and exit, that is, the exit corresponding to the intended picking channel.

[0036] Step 33, after generating multiple picking space-time sub-arcs and multiple walking space-time sub-arcs corresponding to picking in the intended picking channel according to the path strategy corresponding to the picking order type, take the end node of the last walking space-time sub-arc corresponding to the current intended picking channel as the current position, and perform repeated operations of selecting the next intended picking channel to be entered and the corresponding target entrance, so as to determine the next target picking channel to be entered in turn, and determine the order of entering multiple target picking channels.

[0037] Step 34, for the target picking channels determined in order, generate multiple picking space-time sub-arcs and multiple walking space-time sub-arcs located in the corresponding target picking channels in turn, and combine the target waiting space-time sub-arcs, multiple picking space-time sub-arcs and multiple walking space-time sub-arcs corresponding to the corresponding target picking channels into the target space-time arc corresponding to the corresponding target picking channels, and synthesize the picking links corresponding to the multiple target space-time arcs into the target picking path corresponding to the current picking order.

[0038] Through the above steps 31 to 34, on the picking path corresponding to the space-time network, the optimization of the picking path planning is considered to reduce the waste of congestion waiting time, so as to achieve the goal of minimizing the total picking operation time, reduce the waiting time of the pickers, and alleviate the picking path conflict problem in the multi-operation type and multi-picker operation scenarios, thereby shortening the total duration of the overall picking process.

[0039] In some embodiments, based on the spatiotemporal network and the greedy strategy, the target waiting spatiotemporal sub-arc generated by the current entry into each target picking channel is determined respectively, which is achieved by the following steps:

[0040] Step 41: Based on the space-time network, determine whether each target picking channel is currently associated with a corresponding space-time arc.

[0041] In this embodiment, whether the target picking channel is currently associated with the corresponding space-time arc is determined to determine whether the target picking channel is occupied by other picking objects.

[0042] Step 42. When it is determined that the corresponding target picking channel is not currently associated with the corresponding space-time arc, determine the first walking space-time sub-arc corresponding to the target picking channel that is not associated with the space-time arc, and use the first walking space-time sub-arc as the target waiting space-time sub-arc, wherein the first walking space-time sub-arc is used to represent that the target picking object is on the first picking link within the first walking time, and the first picking link is used to represent the physical path of the target picking object walking from the current position to the cross-channel node corresponding to the entrance representing the picking channel, and the waiting time corresponding to the target waiting space-time sub-arc includes the first walking time.

[0043] In this embodiment, when it is determined that the target picking channel is not associated with the corresponding space-time arc, it means that the corresponding target picking channel is not occupied, and the target picking object can be allowed to access it. The target picking object will use the corresponding target picking channel as the next target picking channel to be visited, which is determined based on the distance between the current location of the target picking object and each target picking channel, that is, the first walking time corresponding to the corresponding first walking space-time sub-arc. At the same time, the first walking time will also be used as the waiting time corresponding to the target waiting space-time sub-arc.

[0044] Step 43, when it is determined that the corresponding target picking channel is currently associated with the corresponding space-time arc, the first walking space-time sub-arc and the first waiting space-time sub-arc corresponding to the corresponding target picking channel are determined, and the first walking space-time sub-arc and the first waiting space-time sub-arc are used as the target waiting space-time sub-arc, wherein the first waiting space-time sub-arc is used to represent that the target picking object is waiting at the cross-channel node corresponding to the entrance representing the picking channel, and the waiting time corresponding to the target waiting space-time sub-arc includes the corresponding first walking time and the waiting time corresponding to the first waiting space-time sub-arc.

[0045] In this embodiment, when it is determined that the corresponding space-time arc is associated with the target picking channel, it means that the corresponding target picking channel is occupied by other picking objects, and when the target picking object has no other optional target picking channels to visit, the target picking object needs to wait for the picking object currently being picked in multiple target picking channels to complete picking, and then select one of the picking channels that is first allowed to be accessed as the next target picking channel to be visited. At this time, the first walking space-time sub-arc and the first waiting space-time sub-arc of the target picking object walking from the current position to each target picking channel will be determined, and the waiting time corresponding to the first waiting space-time sub-arc includes the time when at least one other picking object that was picked in the corresponding target picking channel before the target picking object completes picking. The picking time selected, the picking time for another picking object to complete picking includes the total picking time calculated based on the picking time (for example: 3 seconds) required to pick the corresponding item (corresponding to a picking SKU) at each picking channel node, the walking time between multiple picking channel nodes, and the walking time from the last picking channel node to the exit of the target picking channel, and the sum of the total picking time and the two walking times is taken as the picking time for a picking object to complete picking; in this embodiment, when determining the waiting time corresponding to the target waiting time-space sub-arc corresponding to the target picking channel, the picking time required for at least one other picking docking and the first walking time of the first walking time-space sub-arc corresponding to the target picking object are used as the corresponding waiting time.

[0046] Through the above steps 41 to 43, the target waiting time-space sub-arc generated by currently entering each target picking channel is determined, and then the picking waiting time required for the target picking object to enter each target picking channel for picking is determined.

[0047] In some embodiments, generating a plurality of picking spacetime sub-arcs and a plurality of walking spacetime sub-arcs located in corresponding target picking channels is achieved by the following steps:

[0048] Step 51 , according to the path strategy corresponding to the picking order type, respectively determine the cross-channel nodes through which the target picking object enters and exits the target picking channel, and obtain a first cross-channel node and a second cross-channel node.

[0049] In this embodiment, the first cross-channel node is used as the entrance to the target picking channel, and the second cross-channel node is used as the exit to exit the target picking channel; it can be understood that according to the corresponding path strategy, the first cross-channel node and the second cross-channel node can be the same channel opening, for example: when the picking order is an offline order, the path strategy is the return strategy, then the first cross-channel node and the second cross-channel node are both channel openings close to the entrance and exit of the flat warehouse; for example: when the picking order is an online order, and the strategy for entering and exiting the target picking channel is the S-shaped path strategy, the first cross-channel node and the second cross-channel node are the channel openings at both ends of the target picking channel respectively.

[0050] In some optional implementations, the picking order types include online picking types and offline picking types. According to the path strategy corresponding to the picking order type, the cross-channel nodes through which the target picking object enters and exits the target picking channel are respectively determined to obtain the first cross-channel node and the second cross-channel node. This is achieved by the following steps:

[0051] Step 511: When the picking order type is offline picking type, the path strategy is determined to be the return path strategy, and the cross-channel nodes corresponding to the entrance of the target picking channel are respectively used as the first cross-channel node and the second cross-channel node.

[0052] Step 512, when the picking order type is an online picking type, determine that the path strategy is a combined path strategy consisting of an S-type path strategy and a return path strategy, and after taking the entrance of the target picking channel as the first cross-channel node, determine the target picking object starting from the last first channel node of the target picking channel, exiting the target picking channel along the alternative exit and walking to the first first channel node of the next target picking channel, and take the alternative exit corresponding to the shortest walking distance as the second cross-channel node, wherein the alternative exit includes one of the entrance and exit of the target picking channel.

[0053] Step 52, among all the picking channel nodes corresponding to the target picking channel, the picking channel nodes storing the picking SKU to be picked are retrieved in turn to obtain the first channel node, and after determining the node paths between the first cross-channel node and the first first channel node, the two adjacent first channel nodes, and the last first channel node and the second cross-channel node, the walking time of the target picking object in the corresponding node path and the corresponding node path are converted into the corresponding walking time-space sub-arc, and the node position of each first channel node and the picking time corresponding to the target picking object are converted into the corresponding picking time-space sub-arc.

[0054] In the embodiment, based on the corresponding node path and time, the spatio-temporal state is described to describe the picking process of the target picking object in the target picking channel.

[0055] In step 53, according to the sequence, it is determined whether the target picking channel is the last picked target picking channel, and in the case where the target picking channel is the last picked target picking channel, the walking path and the corresponding walking time between the second cross-channel node corresponding to the last picked target picking channel and the plane warehouse entrance are converted into the end walking spatio-temporal sub-arc, and the end walking spatio-temporal sub-arc is taken as one of the walking spatio-temporal sub-arc corresponding to the last picked target picking channel.

[0056] In the embodiment, after a picking order is picked, the picking object needs to return to the plane warehouse entrance, and the walking time of the picking object returning to the plane warehouse entrance is also the corresponding picking time, therefore, by adding the walking time corresponding to the end walking spatio-temporal sub-arc to the corresponding picking time, the total time of picking a picking order is quantified, and the accurate time parameter is improved to achieve the goal of minimizing the total picking operation time.

[0057] Through the above steps 51 to 53, the plurality of picking spatio-temporal sub-arc and the plurality of walking spatio-temporal sub-arc located in the corresponding target picking channel are generated.

[0058] In some embodiments, before determining the sequence of entering the plurality of target picking channels, the following steps are also implemented:

[0059] In step 61, the plurality of target picking channels associated with each picking order and the picking channel access sequence are determined respectively, and based on the picking channel access sequence and the path strategy corresponding to the picking order type, the first path of the corresponding picking object in the plurality of target picking channels is determined.

[0060] In step 62, according to the picking sequence determined according to the allocation sequence of allocating the plurality of picking orders to the plurality of picking objects, the order of the plurality of picking objects entering the same target picking channel is determined, and one target picking channel does not allow multiple picking objects to enter at the same time.

[0061] In step 63, according to the order, the congestion waiting time required for the picking object to enter the corresponding target picking channel for picking is determined, and based on the first path and the congestion waiting time corresponding to each target picking channel, the picking time and the walking time required for the picking object to pick in the corresponding target picking channel, the picking position and the spatio-temporal state are associated and mapped, and at least the initial spatio-temporal sub-arc for picking in each target picking channel is generated, wherein the congestion waiting time is determined according to the picking time and the walking time required for the picking object of the previous picking order to pick in the corresponding target picking channel.

[0062] Step 64 : Generate an initial space-time arc corresponding to the picking order based on the initial space-time sub-arcs corresponding to the multiple target picking channels associated with the picking order, and generate a space-time network based on the initial space-time arcs corresponding to the multiple picking orders.

[0063] Through the above steps 61 to 64, the spatiotemporal network corresponding to the multiple picking orders is depicted, providing a data basis for calling the corresponding spatiotemporal network when planning the picking route.

[0064] refer to Figures 3 to 8 The path planning process of the preferred embodiment of the present application is described below:

[0065] In this embodiment, a mathematical model for picking path planning considering channel congestion is established, so that the congestion handling strategy of allowing only a single picker to enter the same picking channel at any time is implemented as a hard constraint, and at the same time, the walking, waiting, and picking time of the picker during the operation is quantified, and the goal of alleviating channel congestion is achieved by constructing an objective function that minimizes the total picking operation time; in this embodiment, the corresponding mathematical model is set as follows: (1) The flat warehouse has only one entrance and exit (I / O), and the picker starts from this entrance and exit when performing the picking task and returns to this entrance and exit after completing the picking task; (2) Each SKU in the picking list can be retrieved in the warehouse, that is, out-of-stock is not allowed; (3) When starting to pick each picking list, the SKU storage location allocation plan is fixed; (4) Picking lists are divided into online and offline types, and each picker can pick any type of picking list. Each picking list can be completed by one picker in one operation, and the picker must complete the picking of all SKUs on the picking list; (5) Path nodes are divided into picking channel nodes and cross-channel nodes, and each storage location can be Access is made through the corresponding picking channel node; (6) Only one picker is allowed at a picking channel node at the same time (crossing is not allowed), and there is no limit on the number of pickers at the cross channel node (crossing is allowed); (7) When any SKU is stored in multiple shelves, the shelf with the smallest shelf number / shelf number (called the shelf to be picked) is preferentially visited when picking; (8) The shelf to be picked of any SKU in each picking order only needs to be visited once, that is, picking any SKU only needs to visit the corresponding picking channel node once; (9) There is a time interval between the arrival times of any two picking orders, that is, there is a time interval between the start of any two picking tasks; (10) The picking time of a SKU in online and offline picking orders is constant; (11) The walking speed of each picker is uniform; in this embodiment, the picker's work process (including walking, waiting and picking) is mapped to a space-time arc in the space-time network, and the time consumption of each link is quantified by the time period information in the space-time arc. The space-time waiting arc is used to characterize the congestion state of the picking channel. Since the completion of the entire picking task depends on the completion time of the slowest picker, in order to shorten the total time of the entire picking process, an objective function is established to minimize the maximum working time of all pickers. The objective function of the picking path planning model is as follows: ; Constraints: (1), It means ensuring that each SKU in the picking list has a confirmed location to be picked and can be accessed through its corresponding path node; (2) ensuring that all SKUs in the picking list (including online and offline picking lists) are picked out. The constraint is: ; ; (3) Ensure that the picker starts from I / O and returns. The constraint is: ; (4) Ensure that the combined path strategy is followed when picking online orders, and the return path strategy is followed when picking offline orders: 、 、 ; (5), constraint type To ensure that the number of pickers entering and leaving each spatiotemporal network vertex is equal, where o∈O\{0} represents excluding the warehouse entrance and exit nodes from the node set; (6) to ensure that the corresponding physical link between adjacent nodes has only one spatiotemporal arc, that is, the coupling constraint between the spatiotemporal arc and the physical link of the operation route, the constraint formula is: ; (7) means that the system only generates one picking order each time, where represents all space-time arcs starting from the warehouse entrance and exit (node ​​0), and the constraint is: ; (8) Ensure that only one picker is allowed to enter the same picking channel at any time, where means excluding all waiting arcs from the space-time arc set, Indicates that the starting point and end point of the space-time arc are both internal nodes of the picking channel k, and the constraint is: ;(9) Dynamic update constraints of space-time arc set: After each picker path is generated, the space-time arc set A will be updated based on the current channel blocking time blocktime k Dynamic update ensures that the newly generated arc does not conflict with the occupied time period. The constraint is: ,A new is the space-time arc set newly generated by the space-time network construction algorithm; Ⅱ () is the indicator function, if the space-time arc With blocktime k If the defined time periods overlap, the space-time arc cannot be selected; Indicates SKU s The path node o corresponding to the cargo location to be picked represents the cargo location allocation plan; the two-dimensional coordinates of the path node o are (ox, oy), and the path node The two-dimensional coordinates of , ), h1 represents the width of the picking channel; h2 represents the width of the cargo aisle; h3 represents the width of the cargo location; M represents the total number of cargo locations in a single cargo aisle; K represents the picking channel set, K={1,2,…,[(N-1) / 2]}; k represents the picking channel index, k=[(i-1) / 2], k∈K, i represents the cargo location index; O is the warehouse path node index set (I / O node number is 0); Qc represents the cross channel node set, Qc∈O; Qa represents the picking channel node set, Q a ∈O;O k,c represents the entry and exit nodes of picking channel k, O k,c ∈Q c ;O k,adenotes the set of nodes in picking aisle k, O k,a ∈ Q a ; denotes a path node, ∈ O; U denotes the set of pickers; u denotes the u-th picker, B denotes the set of picking orders, B online denotes the set of online orders, B online ∈ B; B offine denotes the set of offline orders, B offine ∈ B; b u denotes the picking order of the u-th picker, b u ∈ B; T denotes the set of task times; denotes a task time, ∈ T; denotes the expected occupancy time of picking aisle k; denotes the waiting time of picking aisle k; denotes the picking time of a SKU in an online picking order; denotes the picking time of a SKU in an offline picking order; t u denotes the final time of the u-th picker to complete the current task; A denotes the set of all types of space-time arcs; A V denotes the set of space-time walk arcs, A V ∈ A; A W denotes the set of space-time waiting arcs, A W ∈ A; A H denotes the set of space-time picking arcs, ; denotes the set of online task space-time picking arcs, ; denotes the set of offline task space-time picking arcs, ; a u denotes the space-time arc of the u-th picker, a u ∈ A; denotes the space-time walk arc of the u-th picker, ; denotes the space-time waiting arc of the u-th picker, ; denotes the space-time picking arc of the u-th picker, ; Q denotes the set of vertices in the space-time network; (o, t) denotes a vertex in the space-time network, (o, t) ∈ Q; E denotes the set of physical links between adjacent nodes; denotes a physical link between adjacent nodes, ; denotes the space-time arc formed by the picker being active on the physical link from time t to ; v urepresents the travel speed of the u-th picker; R represents the working route of all pickers; Representation node The distance between them; D1 represents the node under the return strategy The distance between them; D2 represents the distance between them under the combined path strategy the distance between them; represents the blocking time of picking channel k; is the decision variable, when the u-th picker chooses the space-time arc , that is, from time t to In the physical link When doing activities, , otherwise 0; is the decision variable, when the u-th picker selects the physical link hour, , otherwise 0.

[0066] In this embodiment, in order to accurately describe the spatiotemporal state of the picker during the picking process and quantify his walking, waiting and picking time, a spatiotemporal network is introduced to describe the order picking process, and the NM-1 plane warehouse model is gridded into a physical node network. Figure 3 As shown, and taking two selectors as examples, it depicts Figure 4 The space-time network shown in the figure; in order to directly call the space-time network of the order picking process when planning the picking path, refer to Figure 5 The process shown includes: step 1, inputting the path node index set O and the operation time set T; step 2, traversing the operation time t in sequence, and then executing step 3; step 3, traversing the path node o in sequence, and then executing step 4; step 4, adding the vertex (o, t) in the spatiotemporal network to the spatiotemporal vertex set Q, and then executing step 5; step 5, determining whether the path node o belongs to the cross-channel node O c If yes, go to step 6, otherwise go to step 7; Step 6, add space-time waiting arc ( ,t,t+1) to the set A of time-space waiting arcs W , then, execute step 13; step 7, determine whether the path node o belongs to the picking channel node O a If yes, go to step 8, otherwise go to step 9; Step 8, determine whether the picking order of the u-th picker is an online picking order, if yes, go to step 10, otherwise go to step 11; Step 9, add a time-space walking arc ( ) to the spacetime walking arc set A V , then, execute step 13; step 10, add time-space picking arc ( ) to the online task space-time selection arc set , then, execute step 12; step 11, add space-time picking arc ( ) to offline task space-time picking arc set , then, execute step 12; Step 12, collect online task space-time picking arcs and offline task spatiotemporal picking arc set Merge into space-time selection arc set A H , then, execute step 13; Step 13, set the set A of time-space waiting arcs W , space-time selection arc set A H and the spacetime walking arc set A V Merge into a space-time arc set A; Step 14, loop through all path nodes o and all operation times t; Step 15, output the space-time arc set A.

[0067] In this embodiment, reference Figure 6 In order to meet the constraint that only one picker can work in the picking channel at any time, the block time interval table is used in the algorithm. k ) to record the cumulative occupancy time of the picking channel. At the same time, according to the picker's operation type in the channel (online or offline picking orders), the time required for picking SKUs and the walking time under the corresponding path strategy can be determined. From this, the expected occupancy time of the picker for a certain picking channel k can be obtained. , according to blocktime k and The waiting time of the picker in channel k can be obtained , further, with Generate the next picker's job path for the constraint until all job paths in the spatiotemporal network are generated; in this embodiment, through the greedy strategy (select Minimum picking channel) and dynamic update mechanism (blocktime k ) Reduce the waiting time of pickers and alleviate the picking path conflict problem in multi-job type and multi-picker operation scenarios, thereby shortening the total time of the overall picking process; Specifically, the process includes: Step 1, input the picking order set B, operation time set T, time-space arc set A, and storage location allocation plan X, then execute Step 2; Step 2, initialize blocktime k ,R,t,o u ,t u, then, execute step 3; step 3, determine whether the operation time t≤T?, if yes, execute step 4, otherwise, the process ends; step 4, traverse all pickers starting from the first picker, then execute step 5; step 5, determine whether picker u belongs to the picker set U, if yes, execute step 6, otherwise, execute step 20; step 6, process picker u, then execute step 7; step 7, obtain the SKU storage node, then, until step 8; step 8, calculate the channel occupancy and waiting time, then execute step 9; step 9, select the next picking channel k*, then execute step 10; step 10, construct the time-space walking arc , then, execute step 11; step 11, judge If yes, go to step 12, otherwise go to step 13; Step 12, construct the space-time waiting arc , then, execute step 16; step 13, judge b u ∈B online ?, if yes, go to step 14, otherwise go to step 15; Step 14, construct online spatiotemporal picking arc , then, execute step 16; step 15, construct offline space-time picking arc , then, execute step 16; Step 16, merge space-time arcs , then execute step 17, update status: t u , ,blocktime k ,o u After that, execute step 18 to update the space-time network. Then, execute step 19. Step 19: increment the picker index u by one. Step 20: determine if B=∅? If so, execute step 21. Otherwise, execute step 22. Step 21: construct the space-time walking arc back to the entrance. , then execute step 23; step 22, update the global time t=min(t u ), then, execute step 3; step 23, merge path R←a u , then, execute step 24; Step 24, calculate the objective function F(R), then, execute step 25; Step 25, output: R, F(R).

[0068] refer to Figure 7 and Figure 8Taking the picking of two picking orders as an example, the corresponding path planning and spatiotemporal path planning process are explained. Picker A performs the online order picking task, and picker B performs the offline order picking task. The two walk using the combination strategy and the return path strategy respectively. The algorithm recognizes that picking aisle 1 is occupied by picker A. Therefore, when planning a path for picker B, it chooses to avoid picking aisle 1 and adjusts the access order of its picking aisles to 0→2→1→0. When picker A is working in picking aisle 1 from time t=1.5 to t=2.5, picker B is guided to enter the unoccupied picking aisle 2 to carry out his work. When picker B completes the picking task of node o=12 and arrives at node o=1 at the entrance of picking aisle 1 at time t=29.5, picker A has already left aisle 1, so he enters aisle 1 to complete the picking task of node o=4. Picker B's latest picking path avoids the 10 s waiting time at node o=1 before the adjustment, and completes this picking operation and returns to the entrance and exit in advance at time t=44.

[0069] In some embodiments, before obtaining the cargo space allocation information that has been planned before the current time, the following steps are further performed:

[0070] SKU和SKU包括:Stock Keeping Unit、 ... When the goods are delivered, the outbound frequency parameter and the commodity correlation parameter can be used as corresponding known parameters to carry out the cargo space allocation planning; in this embodiment, in the process of triggering the cargo space allocation planning by the arrival of the replenishment batch, the replenishment batch includes the new batch allocation and the old batch reallocation. For the new batch allocation, the items in the new batch order are arranged to the corresponding cargo space, but for the old batch reallocation, the items in the old batch order will be transferred to the warehouse, that is, in the process of reallocation of the old batch, if the items in the old batch are allocated to the new cargo space, it is necessary to transfer the warehouse, and the transfer operation will generate transfer costs, which will further affect the picking cost. Therefore, in this embodiment, it is necessary to consider whether to transfer the items in the old batch based on the transfer cost and the corresponding picking cost, that is, to reallocate the old batch, and the transfer operation is based on the original cargo space information of the old batch items.

[0071] Step 72: Perform double-layer coding and initialization for cargo location allocation based on the target information and the corresponding target SKU to generate multiple double-layer coded individuals.

[0072] In this embodiment, in order to realize the two-level decision of cargo location allocation and warehouse transfer operation, a two-level coding solution algorithm based on genetic algorithm and a genetic operator suitable for two-level coding are adopted; in this embodiment, the symbol coding method is selected to arrange the allocation of cargo locations. At the same time, in order to completely describe the two situations of arranging cargo locations for new batches of goods and reallocating cargo locations for old batches of goods, the storage allocation and warehouse transfer decision are decoupled to avoid the coupling of decision variables, and a two-level coding structure is designed.

[0073] In this embodiment, the first layer of the double-layer encoding uses X to describe the mapping relationship between new and old batches of goods and the cargo locations, which is expressed as a cargo location code. Specifically, the cargo location code X is divided into N×M segments according to the total number of cargo locations, where N is the total number of aisles and M is the number of cargo locations in a single aisle. The arrangement order of the cargo location index follows a double loop traversal, that is, the outer loop aisle number i and the inner loop cargo location number p; each segment represents a gene, and each gene represents the type of goods stored at a cargo location position (i, p); the value of the gene is the index s of the product type, that is, the SKU code; in this embodiment, the SKU code is an integer code form, and the non-existent SKU code is set to 0. Therefore, the cargo location code adopts a matrix symbol coding structure, the row vector corresponds to the cargo aisle number, and the column vector represents the cargo location sequence. Its genotype can be expressed as follows when expanded row by row: ,in, ,when When the product s is stored at the pth position of the aisle i, in this embodiment, the second-level coding is designed to use Y to record the transfer information of the old batch, which is expressed as a transfer code. Specifically, the transfer code Y is divided into N×M segments according to the total number of cargo positions. Each segment represents a gene, and each gene represents the transfer status of the old batch stored at the cargo position; the value of each segment gene is the index of the new cargo position (i, p) to which the old batch stored at the cargo position (i, p) is moved. ), since this layer of coding is used to record the transfer information of old batches of goods, the gene corresponding to the storage location where the old batch of goods is stored is only recorded. In particular, when no transfer occurs, the original storage location index is used as the gene value, and the storage location code adopts conditional coordinate coding. Its genotype can be expressed as: , where y ip =∅ means that the location (i,p) does not store the old batch of goods, Indicates that the old batch of goods is moved from (i,p) to ( ); Therefore, a double-layer coded individual is represented in genetic form as: [ | ], the corresponding structure can be referred to Figure 9 shown.

[0074] In some specific implementations, assume that the flat warehouse has 2 aisles, each aisle has 5 cargo locations, a total of 10 cargo locations; a product has been stored in cargo locations (2,1) and (2,2), and its SKU code is 3, which is called the old batch of SKU3. At this time, if three products need to be put into the warehouse and assigned cargo locations, their SKU codes are: 1, 2, and 3, which are called the new batches of SKU1, SKU2, and SKU3, and correspondingly need to be assigned cargo locations 2, 3, and 1 respectively, refer to Figure 10, a double-layer coded individual is generated by the following method: SKU1 is placed in the cargo locations (1,1) and (1,2); similarly, SKU2 is placed in (1,3), (1,4), and (1,5); and SKU3 is placed in (2,3). X={1,1,2,2,2,3,3,3,0,0} is the genotype of the individual's cargo location code, indicating a possible cargo location allocation plan; Y={ , , , ,,(2,1),(2,2), , ,} is the genotype of the individual's transfer code, indicating that the transfer operation needs to be performed on the old batch of goods. [X|Y]=[{1,1,2,2,2,3,3,3,0,0}|{ , , , ,(2,1),(2,2), , ,}] is the complete genotype of the individual.

[0075] In this embodiment, before generating the first generation population, in order to ensure that the population individuals always meet the historical storage constraints, it is necessary to complete the population initialization work first; since the old batch has been stored in the cargo location, its original storage location is known. After the first generation population is generated, the storage location of the old batch may change after subsequent genetic operations. In order to accurately describe the specific information of the transfer operation in the encoding, it is necessary to generate a reference individual in advance when constructing the first generation population, that is, to assign a value to the gene corresponding to the cargo location where the old batch is stored in the cargo location code X, and take the value as the stored SKU code; similarly, to assign a value to the gene corresponding to the cargo location where the old batch is stored in the transfer code Y, and take the value as the original cargo location index; specifically, taking the above-mentioned cargo location allocation scheme as an example, the old batch of SKU3 has been stored in cargo locations (2,1) and (2,2). At this time, the reference individual of the first generation population is generated: [X|Y]=[{0,0,0,0,0,3,3,0,0,0}|{ , , , , ,(2,1),(2,2), , ,}], where x in the cargo location code X ip =0 means the storage location is empty, and the y in the transfer code Y ip =∅ indicates that the location corresponding to the gene locus does not store old batches, and there is no need to record the transfer information. The reference individuals of the generated primary population are shown in the figure. Figure 11 shown.

[0076] Step 73, using a preset genetic algorithm and target fitness function, performs preset genetic evolution operations and evolution iterations on the cargo location sub-code body of the double-layer coded individual, and based on the cargo location sub-code body that completes the corresponding genetic evolution operation, updates the transfer code of the transfer sub-code body of the corresponding double-layer coded individual, until the target double-layer coded body corresponding to the current replenishment order is obtained, and the cargo location allocation information is decoded from the cargo location sub-code body corresponding to the target double-layer coded body, wherein the target fitness function is constructed based on the commodity outbound frequency parameter and commodity correlation parameter associated with the cargo location code of the corresponding cargo location sub-code body and the transfer cost associated with the corresponding transfer code, the cargo location code is used to represent the cargo location arrangement for a target SKU, and the transfer code is used to represent a change in cargo location arrangement.

[0077] In this embodiment, the use of genetic algorithms to perform corresponding genetic evolution operations and genetic evolution iterations is clear and feasible for those skilled in the art. The difference lies in the genetic evolution operations specifically adopted in the embodiments of the present application: for example, an adaptive crossover operator, an adaptive local mutation strategy. At the same time, the embodiments of the present application also use a corresponding target fitness function based on the frequency of goods leaving the warehouse, the correlation between goods, and the cost of transferring goods, and then selects, crosses, and mutates the genetic evolution operations through the function value of the corresponding target fitness function until the corresponding target two-layer encoding body including the cargo location allocation information is finally generated.

[0078] In this embodiment, during the cargo location allocation planning process, a cargo location allocation mathematical model driven by replenishment events is also constructed to mathematically model the cargo location allocation planning process, and by solving the target value corresponding to the target function according to the set target function based on the constructed cargo location allocation mathematical model, the corresponding cargo location allocation plan is obtained; in this embodiment, the following settings are made for the cargo location allocation mathematical model: (1) The warehouse has only one entrance and exit (I / O); (2) Each cargo location only stores a single SKU, and the same SKU can be stored in multiple cargo locations; (3) The volume of a single product of any SKU does not exceed the maximum capacity of a single cargo location, that is, a cargo location can store at least one product of the same SKU; (4) Only new and old batches of products are distinguished, the SKU of the newly arrived goods to be allocated to the cargo location is the new batch, and the SKU stored in the cargo location is the old batch, and both the new and old batches may include multiple SKUs; (5) The new and old batches of each SKU SKU的存储空间 (SKU的存储空间) SKU的体积小 (提示) SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 (SKU的办法 ( s The correlation between the product All are known parameters; in this embodiment, the objective function of cargo space allocation considering the frequency of goods leaving the warehouse is constructed. : ; Construct an objective function for allocating cargo space considering the relationship between commodities : ;Construct an objective function that considers the cost of warehouse transfer : ; Set the corresponding constraints: 1. The number of storage locations for each SKU product is constrained. The constraint formula is: 2. Each shelf only stores a single SKU. The constraint is: 3. Each old batch location must and can only be moved to one target location. The constraint is: ; 4. Avoidance of conflict in transfer: When the old batch s old From the original location Move to new destination location (not the original location), the target location Must not be occupied by any SKU. The constraint is: 5. Constraint for adjacent allocation within the same aisle: Different locations of the same product are preferentially allocated consecutively within a single aisle. The constraint is: 6. Cross-aisle proximity allocation constraint: When the number of available shelves in a single aisle is insufficient to meet the product's shelf space requirements, continuous allocation is performed across adjacent aisles (⌈⋅⌉ represents rounding up). The constraint is: ; In this embodiment, the cargo location allocation mathematical model is a multi-objective optimization model, and the solution of the multi-objective model needs to consider the frequency of goods leaving the warehouse, the correlation between goods, and the transfer distance, so that each objective function is close to the optimization direction within a certain range to achieve the optimal overall goal; In this embodiment, the three objective functions are given weights, and the multi-objective optimization model of the cargo location allocation model is converted into a single-objective optimization model by using a scalarization method. The objective function expression is as follows: minF=θ1minf1+θ2minf2+θ3minf3,θ1+θ2+θ3=1,θ1,θ2,θ3∈(0,1); In this embodiment, minF is used as the corresponding target fitness function, and the target value of the target fitness function is used as the corresponding fitness value; In this embodiment, I represents the cargo lane index set, I={1,2,…,N}, P represents the cargo location index set, P={1,2,…,M}; N represents the total number of cargo lanes, ; M represents the number of cargo spaces in a single aisle. ;i, Each represents a channel index, i, ∈I; Each represents a location index, p, ∈P; represents the pth cargo location in the i-th cargo aisle, S represents the product SKU code set, S={1,2,…,s}, Each represents a SKU code, s, ;Num s Indicates the total number of cargo spaces required to store one s; S old Represents the set of old batch SKUs, S old ∈S;s old Represents an old batch of s, s old ∈S old ;L old Indicates the original location set occupied by the old batch, L old = ; F s Indicates the frequency of s's shipment; Indicates s and The correlation between ∈[0,1], h represents the height limit of the storage location; h1 represents the width of the picking channel; h2 represents the width of the aisle; h3 represents the width of the storage location; h4 represents the width of the cross channel; Indicates cargo location Distance from warehouse entrances and exits; Indicates cargo location and cargo space the distance between them; Indicates cargo location and cargo space The distance between them is 1 (through the near-end cross passage of the entrance and exit); Indicates cargo location and cargo space The distance between them is 2 (through the cross passage at the far end of the entrance and exit); is a decision variable, indicating that when s is stored in the storage location On time, , otherwise 0; Is a decision variable, indicating that when the old batch s old From the original location Move to target location On time, , otherwise 0.

[0079] Through the above steps 71 to 73, the planning of the cargo location allocation information based on the arrival of the replenishment batch is realized, so that the picking efficiency during the picking operation is improved through the planned cargo location allocation information, and the total picking operation time is minimized.

[0080] In some embodiments, after generating a plurality of double-layer coded individuals, the following steps are further performed:

[0081] Step 81 : Perform a tournament selection operation on a plurality of double-layer coded individuals to obtain a plurality of candidate double-layer coded individuals.

[0082] In this embodiment, a tournament selection operator is used to select two-layer coded individuals with higher fitness (calculated according to the corresponding target fitness function) as parents through random competition. K two-layer coded individuals are randomly selected from the population, their fitnesses are compared, and the two-layer coded individual with the highest fitness is selected as the parent. This process is repeated until a sufficient number of parent individuals are selected to obtain multiple candidate two-layer coded individuals. In this embodiment, the selection pressure can be flexibly controlled by adjusting the tournament size k: when k is large, the selection pressure is greater and the probability of excellent individuals being selected is higher; when k is small, the selection pressure is smaller and the population diversity is better.

[0083] Step 82: Randomly select any two alternative cargo location sub-coding bodies from the cargo location sub-coding bodies of multiple alternative double-layer coding individuals as the intended cargo location sub-coding body group, and after selecting at least two cargo location codes corresponding to the target SKU from all the cargo location codes corresponding to the two alternative cargo location sub-coding bodies of the intended cargo location sub-coding body group, cross and perform cargo location conflict correction on the cargo location codes selected from the corresponding two alternative cargo location sub-coding bodies to generate two corresponding first cargo location sub-coding bodies, and update the transfer code corresponding to the cargo location code that has moved the cargo location during the cargo location conflict correction.

[0084] In some of these optional embodiments, reference Figure 12 The cargo location codes selected from the corresponding two candidate cargo location sub-code bodies are crossed and the cargo location conflicts are corrected to generate two corresponding first cargo location sub-code bodies. This is achieved through the following steps:

[0085] Step 821 , respectively determine the cargo location codes selected from the two candidate cargo location sub-code bodies to obtain a first cross cargo location code and a second cross cargo location code, and determine the coding position without a cargo location code in each candidate cargo location sub-code body.

[0086] Step 822: Determine the cargo location code with the same coding position as the second cross cargo location code among all the alternative cargo location sub-coding bodies corresponding to the first cross cargo location code, replace all the second cross cargo location codes with the determined cargo location code, and correct the second cross cargo location code that is not in the coding position corresponding to the first cross cargo location code and the coding position without the cargo location code to the coding position without the cargo location code, and obtain the first cargo location sub-coding body corresponding to the alternative cargo location sub-coding body corresponding to the first cross cargo location code.

[0087] Step 823: Determine the cargo location code with the same coding position as the first cross cargo location code among all the cargo location codes of the alternative cargo location sub-coding body corresponding to the second cross cargo location code, replace all the first cross cargo location codes with the determined cargo location code, and correct the first cross cargo location code that is not in the coding position corresponding to the second cross cargo location code and the coding position without the cargo location code to the coding position without the cargo location code, so as to obtain the first cargo location sub-coding body corresponding to the alternative cargo location sub-coding body corresponding to the second cross cargo location code.

[0088] In this embodiment, an adaptive crossover operator based on SKU is adopted, with SKU as the basic crossover unit, to ensure that the number of shelves occupied by each SKU remains unchanged, and adaptability is achieved by dynamically adjusting the crossover ratio parameter c. In this embodiment, M alternative double-layer coded individuals are randomly divided into M / 2 groups, and the alternative double-layer coded individuals in the group are paired with each other, that is, a group of intended shelf sub-coding bodies is formed. For each group of alternative double-layer coded individuals, a part of SKUs is randomly selected for crossover according to the crossover ratio parameter c, and the parameter c is dynamically adjusted to adapt to different scales. In this embodiment, in order to ensure that the constraint condition of "only a single SKU is stored in each shelf" is always met, a conflict detection and repair mechanism is added to the crossover operator: when the crossover leads to shelf conflict, the idle shelf reallocation strategy is executed, and the SKU on the conflicting shelf is randomly mapped to an idle shelf. After generating a new shelf code X, the transfer code Y is updated to record the transfer information of the old batch of goods.

[0089] In a specific example, assume that the flat warehouse has two aisles, each aisle has 5 cargo locations, a total of ten cargo locations, SKU3 in the old batch of goods occupies two cargo locations (2,1) and (2,2), and SKU1, SKU2, and SKU3 in the new batch of goods need to be allocated 2, 3, and 1 cargo locations respectively. After the population is initialized, two alternative double-layer coded individuals are selected from the initial double-layer coded individuals and are denoted as Parent1: [X|Y]=[{1,1,2,2,2,3,3,3,0,0}|{ , , , , ,(2,1),(2,2), , ,}] and Parent2: [X|Y]=[{3,0,0,1,1,3,3,2,2,2}|{,,,,,(2,1),(2,2),,,}], set the cross ratio parameter c=1.0 to perform global crossover, that is, for each SKU in the location code X of the two bodies, randomly select some location locations to exchange, refer to Figure 12The specific crossover process is as follows: Step 1: Select the crossover location in the two candidate double-layer coded individual storage locations. For Parent1, select the storage location (1,1) for SKU1, (1,3) and (1,4) for SKU2, and (2,2) for SKU3. For Parent2, select (1,4) for SKU1, (2,3) and (2,4) for SKU2, and (1,1) for SKU3. Step 2: Perform a global crossover based on the SKU, pairing Parent1 and Parent2. Using the SKU as the basic crossover unit, swap the selected storage location for each SKU. In this embodiment, when crossing, Parent1 is first used as the individual that needs to be crossed, and Parent2 is used as the individual that provides the crossing unit. The SKUs on the shelves (1,1), (1,4), (2,3), and (2,4) of Parent2 are first exchanged to the shelves (1,1), (1,4), (2,3), and (2,4) of Parent1. Before the exchange, the empty shelves on Parent1 are (1,1), (1,3), (1,4), (2,2) (2,4) (2,5). After the exchange, because shelf (2,3) is not an empty shelf, SKU2 on shelf (2,3) of Parent2 conflicts with SKU3 on shelf (2,3) of Parent1. At this time, SKU2 provided by Parent2 is corrected from shelf (2,3) of Parent1 to shelf (2,5). When Parent2 is the individual that needs cross processing and Parent1 is the individual that provides the cross unit, the SKUs on the shelves (1,1), (1,3), (1,4), and (2,2) of Parent1 are first exchanged to the shelves (1,1), (1,3), (1,4), and (2,2) of Parent2. Before the exchange, the empty shelves on Parent2 are (1,1), (1,2), (1,3), (1,4) (2,3) (2,4). After the exchange, because the shelf (2,2) is not an empty shelf, the SKU3 on the shelf (2,2) of Parent1 conflicts with the SKU2 on the shelf (2,2) of Parent2. At this time, the SKU3 provided by Parent1 is corrected from the shelf (2,2) of Parent2 to the shelf (2,4), thus obtaining two new shelf sub-code bodies.Step 3, updating the transfer code Y: the original location (2, 2) of the SKU3 old batch in Y:Parent1 no longer stores SKU3, searching for the nearest location (2, 4) storing SKU3, updating the value of the locus corresponding to the original location in the Child1 transfer code to (2, 4), and finally obtaining two corresponding double-layer encoding individuals: Child1: [X|Y]=[{3,1,0,1,2,3,0,3,2,2}|{,,,,,(2,1),(2,4),,,};Child2: [X|Y]=[{1,0,2,2,1,3,3,0,3,2}|{,,,,,(2,1),(2,2),,,}].

[0090] Step 83, randomly selecting a set number of location codes from all location codes of all first location sub-codes, and performing crossover mutation on each group of location codes to generate a second location sub-code corresponding to each first location sub-code, and updating the transfer code corresponding to the location code that moves in the crossover mutation to obtain the location sub-code of the double-layer encoding individual that completes the genetic evolution iteration.

[0091] In this embodiment, referring to Figure 13 , an adaptive local mutation strategy is adopted to focus on local search and diversity maintenance. First, a slight mutation is performed with an initial probability p mo =0.1 to randomly exchange the values of two loci in the individual location code X, and then update the transfer code Y.

[0092] To obtain the target parameters corresponding to the replenishment SKU, in some embodiments, the commodity out-of-warehouse frequency parameters and commodity correlation parameters corresponding to the first SKU and the second SKU are determined by a predetermined statistical method, including the following steps:

[0093] Step 91, using one-hot encoding to construct a matrix of a plurality of historical order data and a third SKU required by the historical order data, generating an order commodity correlation matrix with historical order data as rows and SKUs as columns.

[0094] In this embodiment, one-hot encoding is used to represent the demand for commodity SKUs in different historical orders, and an order commodity correlation matrix is constructed with order names and commodity SKUs in historical orders, as shown in Figure 14 .

[0095] Step 92, calculating the out-of-warehouse frequency vector corresponding to the order commodity correlation matrix, and detecting the out-of-warehouse frequency vector value of the third SKU corresponding to the first SKU and the second SKU in the out-of-warehouse frequency vector to obtain the corresponding commodity out-of-warehouse frequency parameter.

[0096] In this embodiment, the frequency vector of goods shipped out is F=(f i ) 1×K , determine the corresponding delivery frequency, where K is the total number of product SKUs, and element f i is the frequency of product i being shipped out of all customer orders. The larger the value, the higher the frequency of shipping out. i The value of is realized by calculating the relationship matrix R between orders and products, and its calculation formula is: , r i is the product vector, i.e. the column vector r in the order-product relationship matrix R i , indicating which orders the product is required by; unit vector 1 D is a D×1 vector of all ones.

[0097] Step 93: Generate a product co-occurrence matrix based on the order-product association matrix, and use the product co-occurrence matrix to generate a corresponding product association network. The product co-occurrence matrix includes multiple inter-product associations, and the inter-product associations are used to represent the co-occurrence frequency of the corresponding two products in all historical order data. A network graph node of the product association network is used to represent a corresponding SKU, and a node path of the product association network is used to represent the association relationship between two corresponding SKUs.

[0098] In this example, to analyze and quantify the relationships between products, we construct a product co-occurrence matrix by counting the number of times each product co-occurs across all customer orders. This co-occurrence matrix then generates a product association network. The product association network is an undirected, weighted network. Each node in the network represents a single product SKU. The total number of product SKUs determines the number of network nodes. The edges between nodes represent the relationships between products. The edge weights, representing the number of times each product co-occurs across all orders, indicate the strength of the association.

[0099] In this embodiment, based on Figure 14 The order and product relationship matrix R shown in the figure further constructs the product association matrix C=(c ij ) K×K , where K is the total number of product SKUs, and element c ij is the number of times product i and product j appear together in all customer orders, reflecting the degree of association between the two products. The larger the value, the stronger the association. ij The value of is obtained by calculating the relationship matrix R between orders and products, and its calculation formula is: , r i and r j It is not an order vector, but a product vector, that is, the column vector r in the order-product relationship matrix R i and column vector r j, both indicate which orders the product is required by; the constructed product association matrix C refers to Figure 15 In this embodiment, based on the commodity association matrix, a corresponding commodity association network can be constructed. Figure 16 As shown, nodes 1-6 of the product association network correspond to products s1-s6, and edges correspond to the association relationships between different products. The weights on the edges directly refer to the c of the product association matrix C. ij In this embodiment, in order to avoid the difficulty of convergence of the objective function in the subsequent cargo location allocation mathematical model after the introduction of the correlation matrix (co-occurrence matrix) due to the large difference in the order of magnitude of the co-occurrence times, it is also necessary to normalize the co-occurrence times so that they fall within the interval [0,1] to obtain the correlation between the commodities. The calculation formula is: , where the number of co-occurrences Directly corresponds to c of the product association matrix C ij value, max is the number of co-occurrences, and is the largest element value in the matrix C.

[0100] Step 94 : Detect the node paths corresponding to all first SKUs and all second SKUs in the product association network to obtain product association parameters corresponding to the first SKUs and the second SKUs.

[0101] This embodiment also provides a picking path planning device for small and medium-sized flat warehouses in an omni-channel mode. The device is used to implement the above embodiments and preferred implementation methods, and those that have been explained will not be repeated here.

[0102] Figure 17 This is a structural block diagram of a picking path planning device for small and medium-sized flat warehouses in an omni-channel mode according to an embodiment of the present application. Figure 17 As shown, the device includes an acquisition module 171, a selection module 172, a planning module 173 and a processing module 174, wherein:

[0103] The acquisition module 171 is used to obtain the previously planned shelf allocation information after receiving a picking request for picking multiple picking orders, wherein the picking request includes the picking order type and the picking SKU of the picking order, and the shelf allocation information is generated by considering the product outbound frequency parameters, product correlation parameters and transfer costs corresponding to the replenishment SKU of the replenishment order, and using a genetic algorithm to perform storage shelf allocation planning for the replenishment SKU; the selection module 172 is coupled to the acquisition module 171 and is used to select the current picking order and the corresponding target picking object from multiple picking orders assigned to multiple picking objects, and In the shelf allocation information, multiple target picking channels for replenishment SKUs corresponding to the picking SKUs associated with the current picking order are detected and stored; the planning module 173 is coupled to the selection module 172 and is used to determine the order of entering the multiple target picking channels based on the picking waiting time required for the target picking object to enter the target picking channel for picking, and generate the target picking path corresponding to the current picking order based on the order and the path strategy determined according to the picking order type; the processing module 174 is coupled to the planning module 173 and is used to plan the target picking paths corresponding to the multiple picking orders in turn to obtain the path planning results.

[0104] This embodiment further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0105] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0106] Optionally, in this embodiment, the above-mentioned processor can be configured to perform the following steps through a computer program: S1, select the current picking order and the corresponding target picking object from multiple picking orders assigned to multiple picking objects, and detect multiple target picking channels that store the replenishment SKUs corresponding to the picking SKU associated with the current picking order in the shelf allocation information; S2, select the current picking order and the corresponding target picking object from multiple picking orders assigned to multiple picking objects, and detect multiple target picking channels that store the replenishment SKUs corresponding to the picking SKU associated with the current picking order in the shelf allocation information; S3, based on the picking waiting time required for the target picking object to enter the target picking channel for picking, determine the order of entering the multiple target picking channels in sequence, and generate the target picking path corresponding to the current picking order based on the order and the path strategy determined according to the picking order type; S4, plan the target picking paths corresponding to the multiple picking orders in sequence to obtain the path planning results.

[0107] In addition, in conjunction with the above-mentioned embodiments of the picking path planning method for small and medium-sized flat warehouses in an omni-channel model, embodiments of the present application may provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, it implements any of the above-mentioned embodiments of the picking path planning method for small and medium-sized flat warehouses in an omni-channel model.

[0108] The above embodiments only express several implementation methods of the present application. The descriptions thereof are relatively specific and detailed, but they should not be understood as limiting the scope of the invention patent.

Claims

1. A picking path planning method for small and medium-sized flat warehouses in an omni-channel model, characterized in that: include: After receiving a picking request for picking multiple picking orders, obtaining previously planned shelf allocation information, wherein the picking request includes the picking order type and the picking SKU of the picking order, and the shelf allocation information is generated by considering the product shipment frequency parameter, product relevance parameter, and transfer cost corresponding to the replenishment SKU of the replenishment order, and performing storage shelf allocation planning for the replenishment SKU using a genetic algorithm; Selecting a current picking order and a corresponding target picking object from the plurality of picking orders assigned to the plurality of picking objects, and detecting, in the shelf allocation information, a plurality of target picking lanes storing the replenishment SKU corresponding to the picking SKU associated with the current picking order; Based on the picking waiting time required for the target picking object to enter the target picking channel for picking, the order of entering the plurality of target picking channels is determined in sequence, and based on the order and the path strategy determined according to the picking order type, a target picking path corresponding to the current picking order is generated; The target picking paths corresponding to the plurality of picking orders are planned in sequence to obtain a path planning result.

2. The method according to claim 1, characterized in that Selecting a current picking order and a corresponding target picking object from a plurality of the picking orders assigned to a plurality of picking objects, including: after assigning the plurality of the picking orders to the plurality of the picking objects, determining the allocation order in which each of the picking objects receives the picking orders, and selecting the picking order with the earliest allocation order to obtain the current picking order, and using the picking object with the earliest allocation order as the target picking object.

3. The method according to claim 1, characterized in that Detecting, in the shelf allocation information, a plurality of target picking lanes storing the replenishment SKU corresponding to the picking SKU associated with the current picking order, including: Gridding the plane warehouse corresponding to the cargo location allocation information into a node network, wherein the node network includes picking aisles, picking aisle nodes, and cross-aisle nodes, wherein one picking aisle node is associated with two replenishment SKUs assigned to the same picking aisle, and the cross-aisle node is used to represent one of the entrance and exit of each picking aisle; Retrieve the replenishment SKU whose SKU is one of the picking SKUs associated with the current picking order from among the replenishment SKUs associated with the picking channel nodes of all the picking channels, and determine the picking channel node associated with the retrieved replenishment SKU; The picking channel having at least one picking channel node is used as the target picking channel.

4. The method according to claim 3, characterized in that Based on the picking waiting time required for the target picking object to enter the target picking channel for picking, the order of entering the plurality of target picking channels is determined in sequence, and based on the order and the path strategy determined according to the picking order type, a target picking path corresponding to the current picking order is generated, including: Obtaining a space-time network generated by sequential picking planning for the plurality of picking orders, wherein the space-time network includes a plurality of space-time arcs corresponding to the plurality of picking orders, the space-time arcs including at least one space-time sub-arc corresponding to a corresponding space-time state, the space-time sub-arcs including a picking space-time sub-arc, a waiting space-time sub-arc, and a walking space-time sub-arc, and the space-time states including walking, waiting, and picking; Based on the spatiotemporal network and the greedy strategy, the target waiting spatiotemporal sub-arcs generated by the current entry into each target picking channel are respectively determined, and the target picking channel corresponding to the target waiting spatiotemporal sub-arc with the shortest waiting time is selected to obtain the currently intended picking channel to be entered, wherein the target waiting spatiotemporal sub-arc is used to represent the spatiotemporal arc generated when the target picking object walks from the current position to the cross-channel node as the target entrance and waits at the cross-channel node, and the target entrance includes the channel entrance closest to the current position among the corresponding entrance and exit in the intended picking channel; After generating a plurality of the picking space-time sub-arcs and a plurality of the walking space-time sub-arcs corresponding to picking in the intended picking channel according to the path strategy corresponding to the picking order type, taking the end node of the last walking space-time sub-arc corresponding to the current intended picking channel as the current position, performing a repeated operation of selecting the next intended picking channel to be entered and the corresponding target entrance, so as to sequentially determine the next target picking channel to be entered and determine the order of entering the plurality of target picking channels; For the target picking channel determined in the order, multiple picking space-time sub-arcs and multiple walking space-time sub-arcs are generated in turn and located in the corresponding target picking channel, and the target waiting space-time sub-arc, multiple picking space-time sub-arcs and multiple walking space-time sub-arcs corresponding to the corresponding target picking channel are combined into the target space-time arc corresponding to the corresponding target picking channel, and the picking links corresponding to multiple target space-time arcs are synthesized into the target picking path corresponding to the current picking order.

5. The method according to claim 4, characterized in that Based on the spatiotemporal network and the greedy strategy, the target waiting spatiotemporal sub-arcs generated by currently entering each target picking channel are determined respectively, including: Based on the space-time network, determining whether each of the target picking channels is currently associated with the corresponding space-time arc; In the case of determining that the corresponding target picking channel is not currently associated with the corresponding space-time arc, determine the first walking space-time sub-arc corresponding to the target picking channel that is not associated with the space-time arc, and use the first walking space-time sub-arc as the target waiting space-time sub-arc, wherein the first walking space-time sub-arc is used to represent that the target picking object is on the first picking link within the first walking time, and the first picking link is used to represent the physical path of the target picking object walking from the current position to the cross-channel node corresponding to the entrance representing the picking channel, and the waiting time corresponding to the target waiting space-time sub-arc includes the first walking time; When it is determined that the corresponding target picking channel is currently associated with the corresponding space-time arc, the first walking space-time sub-arc and the first waiting space-time sub-arc corresponding to the corresponding target picking channel are determined, and the first walking space-time sub-arc and the first waiting space-time sub-arc are used as the target waiting space-time sub-arc, wherein the first waiting space-time sub-arc is used to represent that the target picking object is waiting at the cross-channel node corresponding to the entrance representing the picking channel, and the waiting time corresponding to the target waiting space-time sub-arc includes the corresponding first walking time and the waiting time corresponding to the first waiting space-time sub-arc.

6. The method according to claim 4, characterized in that Generating a plurality of the picking space-time sub-arcs and a plurality of the walking space-time sub-arcs located in the corresponding target picking channel, comprising: According to the path strategy corresponding to the picking order type, respectively determining the cross-channel nodes through which the target picking object enters and exits the target picking channel, to obtain a first cross-channel node and a second cross-channel node; Among all the picking channel nodes corresponding to the target picking channel, the picking channel nodes storing the picking SKU to be picked are retrieved in sequence to obtain the first channel node, and after determining the node paths between the first cross-channel node and the first first channel node, two adjacent first channel nodes, and the last first channel node and the second cross-channel node, the walking time of the target picking object in the corresponding node path and the corresponding node path are converted into the corresponding walking time-space sub-arc, and the node position of each first channel node and the picking time corresponding to the target picking object are converted into the corresponding picking time-space sub-arc; According to the order, determine whether the target picking channel is the last picked target picking channel, and if it is determined that the target picking channel is the last picked target picking channel, convert the walking path and corresponding walking time between the second cross-channel node corresponding to the last picked target picking channel and the plane warehouse entrance into a terminal walking space-time sub-arc, and use the terminal walking space-time sub-arc as one of the walking space-time sub-arcs corresponding to the last picked target picking channel.

7. The method according to claim 6, characterized in that The picking order type includes an online picking type and an offline picking type. According to the path strategy corresponding to the picking order type, the cross-channel nodes through which the target picking object enters and exits the target picking channel are respectively determined to obtain a first cross-channel node and a second cross-channel node, including: When the picking order type is an offline picking type, determining that the path strategy is a return path strategy, and using the cross-channel nodes corresponding to the entrances of the target picking channels as the first cross-channel node and the second cross-channel node respectively; When the picking order type is an online picking type, the path strategy is determined to be a combined path strategy consisting of an S-type path strategy and a return path strategy, and after taking the entrance of the target picking channel as the first cross-channel node, the target picking object is determined to start from the last first channel node of the target picking channel, exit the target picking channel along the alternative exit, and walk to the first first channel node of the next target picking channel, and the alternative exit corresponding to the shortest walking distance is taken as the second cross-channel node, wherein the alternative exit includes one of the entrance and exit of the target picking channel.

8. The method according to claim 4, characterized in that Before determining the order of entering the plurality of target picking channels, the method further comprises: Determining the plurality of target picking channels and a picking channel access order associated with each picking order, and determining a first path corresponding to picking of the corresponding picking object within the plurality of target picking channels based on the picking channel access order and the path strategy corresponding to the picking order type; According to a picking order determined according to an allocation order of allocating the multiple picking orders to the multiple picking objects, the order in which the picking objects corresponding to the multiple picking orders enter the same target picking channel is determined, wherein multiple picking objects are not allowed to enter the same target picking channel at the same time; According to the sequence, the congestion waiting time required for the picking objects to enter the corresponding target picking channels for picking is determined, and based on the first path and the congestion waiting time corresponding to each target picking channel, the picking time and walking time required for the picking objects to pick in the corresponding target picking channels, a mapping is performed between the picking position and the spatiotemporal state association, and at least an initial spatiotemporal sub-arc for picking in each target picking channel is generated, wherein the congestion waiting time is determined based on the picking time and walking time required for the picking objects of the previous picking order to pick in the corresponding target picking channels; According to the initial space-time sub-arcs corresponding to the multiple target picking channels associated with the picking order, an initial space-time arc corresponding to the picking order is generated, and according to the initial space-time arcs corresponding to the multiple picking orders, the space-time network is generated.

9. The method according to claim 1, characterized in that Before obtaining the cargo space allocation information that has been planned before the current time, the method further includes: Obtaining multiple first SKUs associated with a first replenishment order and multiple second SKUs associated with a second replenishment order, and determining target information corresponding to each of the first SKUs and the second SKUs, wherein the target information includes at least a product shipment frequency parameter and a product correlation parameter determined according to a preset statistical method, and the target information for the second SKU also includes original shelf location information corresponding to the second SKU; Perform double-layer coding and initialization of cargo location allocation based on the target information and the corresponding target SKU to generate multiple double-layer coded individuals; Using a preset genetic algorithm and a target fitness function, the preset genetic evolution operation and evolution iteration are performed on the cargo location sub-code body of the double-layer coding individual, and based on the cargo location sub-code body that has completed the corresponding genetic evolution operation, the transfer code of the transfer sub-code body of the corresponding double-layer coding individual is updated until the target double-layer coding body corresponding to the current replenishment order is obtained, and the cargo location allocation information is decoded from the cargo location sub-code body corresponding to the target double-layer coding body, wherein the target fitness function is constructed based on the commodity outbound frequency parameter and the commodity correlation parameter associated with the cargo location code of the corresponding cargo location sub-code body and the transfer cost associated with the corresponding transfer code, the cargo location code is used to represent the cargo location arrangement for a target SKU, and the transfer code is used to represent a change in cargo location arrangement.

10. The method according to claim 9, characterized in that After generating a plurality of double-layer coded individuals, the method includes: performing a tournament selection operation on the plurality of double-layer coded individuals to obtain a plurality of candidate double-layer coded individuals; Randomly select any two alternative storage location sub-code bodies from the plurality of the alternative double-layer coded individual storage location sub-code bodies as the intended storage location sub-code body group, and after selecting the storage location codes corresponding to at least two target SKUs from all the storage location codes corresponding to the two alternative storage location sub-code bodies in the intended storage location sub-code body group, perform intersection and storage location conflict correction on the storage location codes selected from the corresponding two alternative storage location sub-code bodies to generate two corresponding first storage location sub-code bodies, and update the transfer code corresponding to the storage location code that has been moved during the storage location conflict correction; From all the cargo location codes of all the first cargo location sub-coding bodies, a set number of groups of the cargo location codes are randomly selected, and each group of the cargo location codes is exchanged and mutated to generate a second cargo location sub-coding body corresponding to each of the first cargo location sub-coding bodies, and the transfer code corresponding to the cargo location code that has moved the cargo location during the exchange mutation is updated to obtain the cargo location sub-coding body of the double-layer coding individual that has completed the current genetic evolution iteration, wherein the set number of groups is calculated based on the dynamic mutation probability corresponding to the adaptive local mutation strategy.

11. The method according to claim 10, characterized in that Crossing the cargo location codes selected from the corresponding two candidate cargo location sub-codes and performing cargo location conflict correction to generate two corresponding first cargo location sub-codes, including: Determine the cargo location codes selected from the two candidate cargo location sub-code bodies respectively to obtain a first cross cargo location code and a second cross cargo location code, and determine the coding position in each candidate cargo location sub-code body that does not have the cargo location code; Determine the cargo location code having the same coding position as the second cross cargo location code among all the cargo location codes of the alternative cargo location sub-coding body corresponding to the first cross cargo location code, replace all the second cross cargo location codes with the determined cargo location code, and correct the second cross cargo location codes that are not in the coding position corresponding to the first cross cargo location code and the coding position without the cargo location code to the coding position without the cargo location code, thereby obtaining the first cargo location sub-coding body corresponding to the alternative cargo location sub-coding body corresponding to the first cross cargo location code; Among all the cargo location codes of the alternative cargo location sub-coding body corresponding to the second cross cargo location code, determine the cargo location code with the same coding position as the first cross cargo location code, and replace all the first cross cargo location codes with the determined cargo location code, and correct the first cross cargo location code that is not in the coding position corresponding to the second cross cargo location code and the coding position without the cargo location code to the coding position without the cargo location code, so as to obtain the first cargo location sub-coding body corresponding to the alternative cargo location sub-coding body corresponding to the second cross cargo location code.

12. The method according to claim 9, characterized in that Determining the product delivery frequency parameters and product correlation parameters corresponding to the first SKU and the second SKU according to a preset statistical method includes: Use one-hot encoding to construct a matrix of multiple historical order data and the third SKU required by the historical order data, and generate an order product association matrix with the historical order data as rows and the SKU as columns; Calculating the delivery frequency vector corresponding to the order product association matrix, and detecting the delivery frequency vector value of the third SKU corresponding to the first SKU and the second SKU in the delivery frequency vector to obtain the corresponding delivery frequency parameter of the product; Based on the order-item association matrix, a product co-occurrence matrix is ​​generated, and a corresponding product association network is generated using the product co-occurrence matrix, wherein the product co-occurrence matrix includes multiple inter-item association degrees, and the inter-item association degrees are used to represent the co-occurrence frequency of two corresponding products in all historical order data. A network graph node of the product association network is used to represent a corresponding SKU, and a node path of the product association network is used to represent the association relationship between two corresponding SKUs; The node paths corresponding to all the first SKUs and all the second SKUs are detected in the product association network to obtain product association parameters corresponding to the first SKUs and the second SKUs.

Images