Order-driven mixed loading tray stereoscopic warehouse sorting optimization method

Abstract

The invention discloses an order-driven mixed loading tray stereoscopic warehouse sorting optimization method, which is based on a sorting system comprising a stereoscopic warehouse, a secondary stock area, an RGV conveying system and a disassembly sorting robot, and comprises the following steps: acquiring order and inventory data through a warehouse management system; a hierarchical optimization mathematical model with the maximum order completion number as a first priority and the minimum tray call-in number as a second priority is constructed, an optimal tray combination is solved by adopting a two-stage improved greedy heuristic algorithm, tray call-in and sorting operation is executed, invalid SKU merging and clearance are performed on the remaining trays after sorting, and finally, the optimal tray combination is obtained. After the cache space is released, loop iteration optimization is carried out until all orders are completed; according to the method, the warehouse-in and warehouse-out frequency of the stereoscopic warehouse is remarkably reduced, the cache space utilization rate and the order fulfillment efficiency are improved, the algorithm solving speed is high, the real-time scheduling requirement is met, and the method is suitable for zero-removal sorting scheduling of the large-scale automatic stereoscopic warehouse.

Description

Technical Field

[0001] This invention relates to the field of automated warehouse scheduling technology, and in particular to an order-driven method for optimizing picking in mixed-pallet automated warehouses. Background Technology

[0002] In automated storage and retrieval systems (AS / RS), to improve storage space utilization, a common storage mode is to mix two or more stock keeping units (SKUs) of goods on a single pallet. For the need for picking in smaller quantities, the existing mainstream operating modes are divided into the following two categories: The direct dispatch mode involves directly retrieving the target pallet from the automated warehouse to the picking station, where picking personnel or robots complete the order picking and then return the pallet to the automated warehouse. In this mode, the conveying distance from the automated warehouse inbound / outbound end to the picking station is relatively long, and the round-trip dispatch time for a single pallet is generally over 100 seconds. For multi-batch split orders with dispersed SKUs, it is necessary to frequently return to the automated warehouse to retrieve different pallets, resulting in frequent start-stop and empty runs of the conveyor equipment, with an effective equipment utilization rate of less than 30% and serious energy waste. The pre-caching mode involves setting up a temporary buffer area next to the picking station and retrieving some of the expected demand pallets from the automated warehouse in advance according to the order sequence to the buffer area, thereby reducing the frequency of warehouse inbound and outbound operations. However, this model has obvious technical flaws: The existing pre-caching mode only randomly retrieves pallets based on the order arrival time or schedules them according to simple manual experience rules. It lacks a mathematical model to quantitatively optimize the pallet combination in the cache space. Under the limited capacity constraints, it cannot achieve the optimal match between pallet combination and order demand, resulting in a cache space utilization rate of less than 45%. A large number of cache slots are occupied by invalid SKUs that are not needed by the orders. This not only fails to effectively reduce the number of inbound and outbound operations in the automated warehouse, but also increases the frequency of pallet return and rescheduling due to invalid pallets occupying cache slots, forming an invalid cycle of "retrieval-return to warehouse-retrieval again". In addition, in the existing pre-caching mode, after an order is picked, the pallet usually remains in the cache area until it is called up again by the next order or is manually removed. Due to the lack of an automatic sorting mechanism for the remaining inventory after picking, the pallet carrying invalid SKUs occupies the cache space for a long time, which makes the cache space unable to be reused. For multiple batches of orders in a row on the same day, it is impossible to form an iterative optimization capability, and the overall order fulfillment cycle is extended by more than 40%. The utilization efficiency of the cache space continues to deteriorate as the operation progresses. The optimal combination scheduling problem for mixed pallets is essentially an NP-hard set covering problem. It requires selecting a subset of pallets from tens of thousands of pallets that can cover all order SKU requirements while also satisfying cache capacity constraints. Existing exact solution algorithms (such as branch and bound and integer programming algorithms) take more than 30 minutes per solution when the order size exceeds 100 orders and the pallet size exceeds 10,000 pallets. This cannot meet the 15-minute real-time scheduling cycle requirement of the warehousing system. While traditional heuristic algorithms are fast, they have poor optimization effects, are prone to getting trapped in local optima, and cannot guarantee order fulfillment rates. Currently, there is no efficient solution algorithm that can achieve a balance between solution efficiency and optimization effect and is suitable for engineering implementation. Summary of the Invention

[0003] In view of this, the present invention provides an order-driven picking optimization method for mixed pallet automated warehouses, which solves the problems of frequent warehouse inbound and outbound operations, low cache space utilization, poor order fulfillment efficiency, and inability to meet real-time scheduling requirements in existing mixed pallet picking scenarios. The present invention provides an order-driven pallet scheduling optimization method that can achieve efficient reuse of cache space, reduce the frequency of warehouse inbound and outbound operations, and meet real-time scheduling requirements.

[0004] To achieve the above objectives, the present invention provides an order-driven method for optimizing picking in a mixed-pallet automated warehouse, comprising the following steps: S1. Construct an automated storage and retrieval system (AS / RS) for picking, including the AS / RS warehouse, secondary stocking area, RGV (Remote Guided Vehicle) conveyor system, splitting and picking robots, warehouse management system, and scheduling control system. S2. Collect basic data and perform preprocessing; The warehouse management system collects the set of valid orders to be completed on the same day, real-time inventory data of available mixed pallets in the automated warehouse, real-time number of occupied pallets in the secondary preparation area, number of available buffer bits, and real-time operating status data of the RGV conveyor system and the picking robot on the circular track guide trolley according to a fixed scheduling cycle. The collected basic data is then preprocessed in a standardized manner. S3. Solve for the optimal pallet combination; The scheduling and control system constructs a hierarchical optimization mathematical model based on standardized preprocessed basic data, and obtains the optimal pallet combination that satisfies the constraints through a two-stage improved greedy heuristic algorithm. S4. Pallet receiving and picking operations are executed. The RGV conveying system, which guides the circular track, transfers the pallets corresponding to the optimal pallet combination from the automated warehouse to the secondary preparation area. Then, the pallets are sequentially sent to the order picking robot workstation to complete the order picking. After the picking is completed, the pallets are sent back to the secondary preparation area. S5. Optimize the management of remaining pallet inventory; After the current batch of orders is picked, the remaining pallets in the secondary stocking area are counted and matched with SKU demand. Invalid SKUs that are not needed for the day's orders are filtered out. The pallets carrying invalid SKUs are merged and sent back to the automated warehouse to release the buffer space in the secondary stocking area. S6. Based on the available capacity of the secondary stocking area after release, the remaining uncompleted orders and newly added valid orders for the day, repeat steps S3 to S5 until all orders for the day are completed.

[0005] Preferably, the hierarchical optimization mathematical model prioritizes maximizing the number of orders completed on the same day as the first priority objective, minimizes the number of pallets transferred into the automated warehouse as the second priority objective, and takes the maximum buffer capacity of the secondary stocking area as the core constraint. The hierarchical optimization model defines an order set. ,in Indicates the number of valid orders. Indicates order Inventory quantity unit SKU set, Indicates order For inventory units The quantity required; Define the pallet collection available for automated warehouses ,common One mixed pallet; For tray SKU collection, For tray Internal inventory unit The real-time inventory quantity is a non-negative integer; This represents the maximum buffer capacity of the secondary stock preparation area, and is a positive integer. Define decision variables: tray-loaded decision variables , Indicates the tray Transferred from the automated warehouse to the secondary preparation area. Indicates tray Not transferred to the secondary stocking area; Order completion decision variables , Indicates order All SKU requirements for inventory units were fully met. Indicates order SKU demand was not met; Define the picking quantity decision variable And it is an integer. Indicates from the tray Picking is used to fulfill orders. Inventory unit Quantity; Construct a hierarchical objective function, including a first-priority core objective and a second-priority secondary objective. The first-priority core objective is to maximize the number of orders completed on the day, expressed as: The second priority secondary objective is to minimize the number of pallets transferred into the automated warehouse, expressed as: Preferably, the constraints of the hierarchical optimization mathematical model include order fulfillment constraints, pallet inventory constraints, cache capacity hard constraints, and variable validity constraints. The order fully satisfies the constraint expression as follows: ; The pallet inventory constraint expression is: ; The hard constraint expression for the cache capacity is: ; The expression for the variable validity constraint is: , And it is an integer.

[0006] Preferably, the two-stage improved greedy heuristic algorithm specifically includes: Phase 1: Initial pallet selection prioritizing order coverage. Iteratively calculate the marginal order coverage of each unselected pallet. Each time, select the pallet with the highest coverage and add it to the selected set. Continue until any of the following conditions are met: all orders are satisfied, the secondary stocking area capacity is exhausted, or the remaining pallets cannot be added to satisfy any orders. Output the initial set of pallets. Phase 2: Redundant Pallet Removal and Local Optimization. Perform reverse verification on each pallet in the initial pallet set, remove redundant pallets that do not affect the number of orders fulfilled, and output the final optimal pallet combination, the set of orders that should be fulfilled, and the remaining available capacity.

[0007] Preferably, the order coverage expression is: in, Indicates tray The number of orders that can be fulfilled after being added to the selected set. Indicates tray The number of SKU types that are not required in the order. This represents the weighting coefficient.

[0008] Preferably, the pallet receiving and picking operation specifically includes: S401, the dispatch control system issues instructions to the RGV conveyor system, and according to the order fulfillment priority, the system will combine the best pallets. The corresponding pallets are transferred from the automated warehouse to the secondary preparation area, and the WMS inventory ledger and the status data of the secondary preparation area are updated. S402. According to order priority, the RGV sequentially delivers pallets from the secondary stocking area to the corresponding picking robot workstations. The robot completes the picking according to the order picking list. After picking, the RGV returns the pallet to the corresponding buffer position in the secondary stocking area. S403. After each order is picked, update the WMS order status, pallet inventory ledger, and real-time database of the scheduling control system.

[0009] Preferably, the optimization of remaining pallet inventory management includes: S501. Calculate the real-time inventory of SKUs on the remaining pallets in the secondary stocking area, match them with the SKU demand of the remaining unfulfilled orders for the day, and filter out invalid SKUs with no matching demand. S502. Pallets containing only invalid SKUs are marked as pallets to be returned to the warehouse; for pallets containing both valid and invalid SKUs, invalid SKUs are grouped into the same pallet by a split-pick robot and marked as pallets to be returned to the warehouse. S503. The pallets brought back to the warehouse are sent back to the automated warehouse via the RGV conveyor system guided by the circular track, updating the real-time available capacity of the secondary stocking area.

[0010] Preferably, the optimization method further includes anomaly handling: when an abnormal pallet inventory, equipment failure, or order cancellation occurs, the abnormal pallet is immediately marked as unavailable, the basic data is updated, and the model solution and scheduling in step S3 are re-executed.

[0011] Compared with the prior art, the beneficial effects of the present invention are: This invention constructs a hierarchical optimization model that prioritizes maximizing the number of completed orders and minimizing the number of pallets transferred in, and combines it with a two-stage improved greedy heuristic algorithm centered on marginal order coverage. Under the constraint of limited buffer capacity in the secondary stock preparation area, this invention achieves the goal of covering the most orders with the fewest pallets, thus solving the technical problems of frequent inbound and outbound operations and low equipment utilization in the existing direct transfer mode. This invention uses a hard constraint on cache capacity to fit physical space limitations, introduces an invalid SKU penalty term into the marginal order coverage calculation formula, and actively suppresses the selection of pallets containing a large number of invalid SKUs during pallet screening. At the same time, through the redundant pallet removal mechanism in stage 2, invalid occupancy in the selected set is further removed, and cache space is used efficiently, effectively solving the technical defect of invalid SKUs occupying cache slots for a long time. This invention establishes an optimized process for managing remaining inventory after picking, merging and clearing invalid SKUs in the secondary stocking area to promptly release occupied cache space. Combined with an order-driven iterative optimization mechanism, the optimal pallet combination can be re-solved based on the latest inventory status and order requirements in each scheduling cycle, solving the technical problems of traditional methods such as lack of a closed-loop management mechanism, inability to reuse cache space, and long fulfillment cycles. Attached Figure Description

[0012] Figure 1 This is a flowchart of the picking optimization method of the present invention; Figure 2 This is a schematic diagram of the layout of the automated three-dimensional warehouse picking system of the present invention; Figure 3 This is a flowchart of the two-stage improved greedy heuristic algorithm of the present invention; Figure 4 This is a logic block diagram of the hierarchical optimization mathematical model of the present invention. Detailed Implementation

[0013] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below. Example 1

[0014] This embodiment provides an order-driven method for optimizing picking in a mixed-pack pallet automated warehouse, based on an automated warehouse picking system. The automated warehouse picking system includes a high-rise rack warehouse, a secondary preparation area, a rail-guided vehicle (RGV) conveyor system, a six-axis splitting and picking robot, a warehouse management system (WMS), and a scheduling and control system. The warehouse stores standardized pallets containing two or more SKUs mixed together. The secondary preparation area is adjacent to the picking robot's workstation and has a fixed number of pallet buffer positions, with a maximum buffer capacity of [missing information]. , The value is a positive integer; the RGV conveyor system is used for closed-loop transport of pallets between the outbound end of the automated warehouse, the secondary preparation area, and the picking robot workstation; the scheduling and control system has a built-in optimization model and solution algorithm to execute full-process scheduling instructions; The preferred method provided in this embodiment specifically includes the following steps: S1. Collect basic data and perform preprocessing; The Warehouse Management System (WMS) collects the set of valid orders to be completed on the same day, real-time inventory data of available mixed pallets in the automated warehouse, real-time number of occupied pallets in the secondary preparation area, number of available buffer bits, and real-time operating status data of the RGV conveyor system and the picking robot, with a scheduling cycle of 15 minutes. The collected basic data is then standardized and preprocessed. Remove cancelled and completed orders from the set of valid orders to be completed on the same day. Extract the SKU set, single SKU required quantity, and fulfillment deadline for each order. Extract the SKU set, single SKU real-time inventory quantity, and storage location address for each pallet from the real-time inventory data of available mixed pallets in the automated warehouse. Unify the SKU coding rules for basic data collection, remove invalid SKUs with zero inventory and invalid order data with zero demand quantity, and store the standardized format data in the real-time database of the scheduling and control system.

[0015] S2. Solving for the optimal pallet combination: The scheduling and control system constructs a hierarchical optimization mathematical model based on the standardized preprocessed basic data, and obtains the optimal pallet combination that satisfies the constraints through a two-stage improved greedy heuristic algorithm. The hierarchical optimization mathematical model prioritizes maximizing the number of orders completed on the same day as the first priority objective, minimizes the number of pallets transferred into the automated warehouse as the second priority objective, and uses the maximum buffer capacity of the secondary stocking area as the core constraint. The hierarchical optimization model defines an order set. ,in Indicates the number of valid orders. Indicates order Inventory quantity unit SKU set, Indicates order For inventory units The required quantity is a positive integer. Define the pallet collection available for automated warehouses ,common One mixed pallet; For tray SKU collection, For tray Internal inventory unit The real-time inventory quantity is a non-negative integer; This represents the maximum buffer capacity of the secondary stock preparation area, and is a positive integer. Define decision variables: tray-loaded decision variables , Indicates the tray Transferred from the automated warehouse to the secondary preparation area. Indicates tray Not transferred to the secondary stocking area; Order completion decision variables , Indicates order All SKU requirements for inventory units were fully met. Indicates order SKU demand was not met; Define the picking quantity decision variable And it is an integer. Indicates from the tray Picking is used to fulfill orders. Inventory unit Quantity; Construct a hierarchical objective function, including a first-priority core objective and a second-priority secondary objective. The first-priority core objective is to maximize the number of orders completed on the day, expressed as: The second priority secondary objective is to minimize the number of pallets transferred into the automated warehouse, expressed as: The constraints of the hierarchical optimization mathematical model include order fulfillment constraints, pallet inventory constraints, hard constraints on cache capacity, and variable validity constraints. The order fully satisfies the constraint expression as follows: An order is considered complete only when all SKU requirements are fully covered. The pallet inventory constraint expression is: Only pallets transferred to the secondary stock preparation area can be used for picking, and the picking quantity cannot exceed the actual inventory of the pallets. The hard constraint expression for cache capacity is: The total number of pallets transferred in shall not exceed the maximum capacity of the secondary stock preparation area. The variable validity constraint expression is: , And it is an integer; The core logic of the two-stage improved greedy heuristic algorithm is as follows: The first stage uses marginal order coverage capacity as the criterion to select the initial combination of pallets that meets the capacity constraint and maximizes the number of orders covered; the second stage removes redundant pallets through reverse verification, minimizes the number of pallets loaded, and releases cache space, specifically including: Phase 1: Initial Pallet Selection Prioritizing Order Coverage (1) Initialization, input order set Tray collection Cache capacity Initialize the selected tray set The order set has been satisfied. Remaining available capacity ; (2) Iteratively calculate the marginal order coverage of each unselected pallet. The expression is: in, Indicates tray The number of orders that can be fulfilled after being added to the selected set. Indicates tray The number of SKU types that are not required in the order. This represents the weighting coefficient. In this embodiment, the weighting coefficient... The range of values ​​is ; (3) Select the tray with the highest coverage each time. Add to the selected set, if ,Will join in ,renew Synchronous updates ; (4) Repeat steps (2) to (3) until any of the following occurs: all orders are fulfilled, the secondary stocking area is exhausted, or no more pallets can be added to fulfill any order. Output the initial set of pallets. ; Phase 2: Redundant Pallet Removal and Local Optimization Initial set of pallets Each tray inside Perform reverse verification: temporarily remove After j, if the remaining trays can still fully meet the needs. All orders within, determine For redundant pallets; Redundant pallets are sorted from low to high based on the percentage of valid SKUs, and redundant pallets that do not affect the quantity of orders to be fulfilled are removed in turn. Output the final optimal tray combination The order set should have been satisfied. Remaining available capacity .

[0016] S3. Pallet receiving and picking operations are executed. S301, the dispatch control system sends instructions to the RGV conveyor system to combine the best pallets according to order fulfillment priority. The corresponding pallets are transferred from the automated warehouse to the secondary preparation area, and the WMS inventory ledger and the status data of the secondary preparation area are updated. S302. According to order priority, the RGV sequentially delivers pallets from the secondary stocking area to the corresponding picking robot workstations. The robot completes the picking according to the order picking list. After picking, the RGV returns the pallet to the corresponding buffer position in the secondary stocking area. S303. After each order is picked, update the WMS order status, pallet inventory ledger, and real-time database of the scheduling control system.

[0017] S4. Optimize the management of remaining pallet inventory; After the current batch of orders is picked, the remaining pallets in the secondary stocking area are counted and matched with SKU demand. Invalid SKUs that are not needed for the day's orders are filtered out. The pallets carrying invalid SKUs are merged and sent back to the automated warehouse to release the buffer space in the secondary stocking area. S401. Calculate the real-time inventory of SKUs on the remaining pallets in the secondary stocking area, match them with the SKU demand of the remaining unfulfilled orders for the day, and filter out invalid SKUs with no matching demand. S402. Pallets containing only invalid SKUs are marked as pallets to be returned to the warehouse; for pallets containing both valid and invalid SKUs, invalid SKUs are grouped into the same pallet by a split-pick robot and marked as pallets to be returned to the warehouse. S403. The pallet brought back to the warehouse is sent back to the automated warehouse via the RGV conveyor system of the circular track guide trolley, the new WMS inventory ledger and the real-time available capacity of the secondary stocking area are updated, and the buffer space is released.

[0018] S5. Based on the available capacity of the secondary stocking area after release, the remaining uncompleted orders and newly added valid orders for the day, repeat steps S2 to S4 to complete a new round of pallet combination solution, picking operation and inventory sorting until all orders for the day are completed.

[0019] S6. Exception handling: When pallet inventory is abnormal, equipment malfunctions, or orders are canceled, the abnormal pallet is immediately marked as unavailable, the basic data is updated, and the model solution and scheduling in step S2 are re-executed to ensure the continuity of order fulfillment.

[0020] This embodiment addresses the issues of frequent inbound and outbound operations and low equipment utilization rates in existing direct-access automated warehouses. By employing a limited secondary stocking area buffer capacity constraint, it achieves the goal of covering the maximum number of orders with the minimum number of pallets, thereby minimizing the frequency of pallet inbound and outbound operations in the automated warehouse. Existing technologies suffer from the problems of lacking a quantitative optimization model and low cache space utilization in pre-caching modes. This embodiment addresses these issues by constructing a hierarchical optimization mathematical model that fits the scenario of picking individual mixed pallets, clarifying the constraints and optimization objectives, and providing a quantitative basis for pallet scheduling. This embodiment establishes a sorting and clearing process for remaining inventory after picking and an order-driven iterative optimization mechanism to achieve efficient cyclic reuse of cache space and solve the problem of existing technologies lacking a closed-loop sorting and iterative optimization mechanism; The two-stage improved greedy heuristic algorithm constructed in this invention achieves millisecond-level solution while ensuring optimization effect, meeting the real-time scheduling requirements of warehouses. Example 2

[0021] This embodiment is based on an e-commerce automated warehouse for splitting and picking. The system hardware configuration is as follows: the high-rise racks of the automated warehouse store 12,000 mixed pallets, with each pallet carrying 28 SKUs and 550 items per SKU; the secondary preparation area is adjacent to the picking station, with a maximum buffer capacity of C=20 pallets; the circular track RGV conveyor system is equipped with 2 trolleys, with an average round-trip time of 120 seconds per pallet; 4 six-axis splitting and picking robots, with an average picking time of 15 seconds per order; the WMS and scheduling control system are deployed on an industrial server with an Intel Xeon E-2378G CPU and 32G of memory.

[0022] In this embodiment, there are 80 orders to be completed on the same day, covering 120 SKUs. The specific implementation process is as follows: S1. Data Acquisition and Preprocessing: The WMS collects the SKU requirements of 80 orders, the real-time inventory of 12,000 pallets, and the available capacity of 20 pallets in the secondary stocking area according to a 15-minute scheduling cycle. After data cleaning and standardization, the data is transmitted to the scheduling control system. S2. Model Construction and Solution: A hierarchical optimization model is constructed for the scheduling and control system, considering order size. Pallet size Cache capacity Weighting coefficient The algorithm is set to 0.2. The solution is obtained through a two-stage improved greedy heuristic algorithm. After the first stage iteration, an initial set of 20 pallets is obtained, which can satisfy 58 orders. After the second stage reverse verification, 3 redundant pallets are removed. The final optimal pallet combination is 17 pallets, which can fully satisfy 58 orders and release 3 cache slots. The total algorithm time is 398ms. S3. Pallet Relocation and Picking Execution: The RGV retrieves 17 pallets from the automated warehouse into the secondary preparation area according to the instructions. The pallets are then delivered to the workstations of 4 picking robots in order of order priority to complete the split picking of 58 orders. After the picking is completed, the pallets are returned to the secondary preparation area. S4. Remaining Inventory Management and Optimization: The inventory of 17 remaining pallets was analyzed, and the SKU requirements of the remaining 22 incomplete orders were matched. 12 types of invalid SKUs were identified. Among them, 4 pallets contained only invalid SKUs and were directly marked as pallets awaiting return to warehouse. The remaining 13 pallets had their invalid SKUs merged into 2 pallets by a robot and marked as pallets awaiting return to warehouse. The RGV sent 6 pallets awaiting return to the automated warehouse. The secondary preparation area had 11 valid pallets remaining, releasing 9 cache slots, updating the total available cache slots to 12. S5. Iterative Optimization: Based on 12 available cache slots, 22 remaining unfinished orders, and 10 new orders, combined with the 11 existing valid pallets in the secondary stocking area, the optimal pallet combination is recalculated. Only 8 pallets need to be transferred from the automated warehouse to meet all the needs of the remaining 32 orders. After picking is completed, inventory reorganization is performed again, and the process is repeated until all orders for the day are completed. S6. Exception Handling: If an inventory anomaly occurs on one pallet during the operation, the scheduling system will immediately mark it as unavailable, update the basic data, resolve the model, and add another pallet to ensure that order picking is not interrupted. In this embodiment, only 25 pallets are retrieved from the automated warehouse throughout the entire process, compared to 42 pallets in the traditional random scheduling mode. This reduces the number of inbound and outbound operations by 40.5%, and the space utilization rate of the secondary preparation area remains stable at over 92%. All orders are fulfilled on the same day, and the average response time per order is shortened by 60% compared to the traditional mode.

[0023] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An order-driven method for optimizing picking in a mixed-pallet automated warehouse, characterized in that, Includes the following steps: S1. Construct an automated storage and retrieval system (AS / RS) for picking, including the AS / RS warehouse, secondary stocking area, RGV (Remote Guided Vehicle) conveyor system, splitting and picking robots, warehouse management system, and scheduling control system. S2. Collect basic data and perform preprocessing; The warehouse management system collects the set of valid orders to be completed on the same day, real-time inventory data of available mixed pallets in the automated warehouse, real-time number of occupied pallets in the secondary preparation area, number of available buffer bits, and real-time operating status data of the RGV conveyor system and the picking robot on the circular track guide trolley according to a fixed scheduling cycle. The collected basic data is then preprocessed in a standardized manner. S3. Solve for the optimal pallet combination; The scheduling and control system constructs a hierarchical optimization mathematical model based on standardized preprocessed basic data, and obtains the optimal pallet combination that satisfies the constraints through a two-stage improved greedy heuristic algorithm. S4. Pallet receiving and picking operations are executed. The RGV conveying system, which guides the circular track, transfers the pallets corresponding to the optimal pallet combination from the automated warehouse to the secondary preparation area. Then, the pallets are sequentially sent to the order picking robot workstation to complete the order picking. After the picking is completed, the pallets are sent back to the secondary preparation area. S5. Optimize the management of remaining pallet inventory; After the current batch of orders is picked, the remaining pallets in the secondary stocking area are counted and matched with SKU demand. Invalid SKUs that are not needed for the day's orders are filtered out. The pallets carrying invalid SKUs are merged and sent back to the automated warehouse to release the buffer space in the secondary stocking area. S6. Based on the available capacity of the secondary stocking area after release, the remaining uncompleted orders and newly added valid orders for the day, repeat steps S3 to S5 until all orders for the day are completed.

2. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 1, characterized in that, The hierarchical optimization mathematical model prioritizes maximizing the number of orders completed on the same day as its first priority objective, minimizes the number of pallets transferred into the automated warehouse as its second priority objective, and takes the maximum buffer capacity of the secondary stock preparation area as its core constraint. The hierarchical optimization model defines an order set. ,in Indicates the number of valid orders. Indicates order The set of SKUs for inventory quantity units, Indicates order For inventory units The quantity required; Define the collection of pallets available for an automated warehouse. ,common One mixed pallet; For tray SKU collection, For tray Internal inventory unit The real-time inventory quantity is a non-negative integer; This represents the maximum buffer capacity of the secondary stock preparation area, and is a positive integer. Define decision variables: tray-loaded decision variables , Indicates the tray Transferred from the automated warehouse to the secondary preparation area. Indicates tray Not transferred to the secondary stocking area; Order completion decision variables , Indicates order All SKU requirements for inventory units were fully met. Indicates order SKU demand was not met; Define the picking quantity decision variable And it is an integer. Indicates from the tray Picking is used to fulfill orders. Inventory unit Quantity; Construct a hierarchical objective function, including a first-priority core objective and a second-priority secondary objective. The first-priority core objective is to maximize the number of orders completed on the day, expressed as: The second priority secondary objective is to minimize the number of pallets transferred into the automated warehouse, expressed as: 。 3. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 2, characterized in that, The constraints of the hierarchical optimization mathematical model include order fulfillment constraints, pallet inventory constraints, cache capacity hard constraints, and variable validity constraints. The order fully satisfies the constraint expression as follows: ; The pallet inventory constraint expression is: ; The hard constraint expression for the cache capacity is: ; The expression for the variable validity constraint is: , And it is an integer.

4. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 1, characterized in that, The two-stage improved greedy heuristic algorithm specifically includes: Phase 1: Initial pallet selection prioritizing order coverage. Iteratively calculate the marginal order coverage of each unselected pallet. Each time, select the pallet with the highest coverage and add it to the selected set. Continue until any of the following conditions are met: all orders are satisfied, the secondary stocking area capacity is exhausted, or the remaining pallets cannot be added to satisfy any orders. Output the initial set of pallets. Phase 2: Redundant Pallet Removal and Local Optimization. Perform reverse verification on each pallet in the initial pallet set, remove redundant pallets that do not affect the number of orders fulfilled, and output the final optimal pallet combination, the set of orders that should be fulfilled, and the remaining available capacity.

5. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 4, characterized in that, The order coverage expression is: in, Indicates tray The number of orders that can be fulfilled after being added to the selected set. Indicates tray The number of SKU types that are not required in the order. This represents the weighting coefficient.

6. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 1, characterized in that, The pallet receiving and picking operation specifically includes: S401, the dispatch control system issues instructions to the RGV conveyor system, and according to the order fulfillment priority, the system will combine the best pallets. The corresponding pallets are transferred from the automated warehouse to the secondary preparation area, and the WMS inventory ledger and the status data of the secondary preparation area are updated. S402. According to order priority, the RGV sequentially delivers pallets from the secondary stocking area to the corresponding picking robot workstations. The robot completes the picking according to the order picking list. After picking, the RGV returns the pallet to the corresponding buffer position in the secondary stocking area. S403. After each order is picked, update the WMS order status, pallet inventory ledger, and real-time database of the scheduling control system.

7. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 1, characterized in that, Remaining pallet inventory management and optimization includes: S501. Calculate the real-time inventory of SKUs on the remaining pallets in the secondary stocking area, match them with the SKU demand of the remaining unfulfilled orders for the day, and filter out invalid SKUs with no matching demand. S502. Pallets containing only invalid SKUs are marked as pallets to be returned to the warehouse; for pallets containing both valid and invalid SKUs, invalid SKUs are grouped into the same pallet by a split-pick robot and marked as pallets to be returned to the warehouse. S503. The pallets brought back to the warehouse are sent back to the automated warehouse via the RGV conveyor system guided by the circular track, updating the real-time available capacity of the secondary stocking area.

8. The order-driven, mixed-pallet automated warehouse picking optimization method according to claim 1, characterized in that, The optimization method also includes exception handling: when there is an abnormal pallet inventory, equipment failure or order cancellation, the abnormal pallet is immediately marked as unavailable, the basic data is updated, and the model solution and scheduling in step S3 are re-executed.

Images