Intelligent storage integrated management system

Abstract

According to the intelligent storage integrated management system provided by the invention, warehouse-in, warehouse-out and inventory data are uniformly acquired through the data processing module, decision-making deviation caused by data splitting of each algorithm unit is avoided, each unit in the algorithm decision-making module works cooperatively, and particularly, the inventory distribution unit calls predicted data of the shelf popularity analysis unit to correct a sorting scheme, so that the sorting efficiency is improved. The method does not depend on orders and real-time inventory any more, high-heat short-distance goods shelves can be preferentially selected, and carrying time consumption is reduced; the algorithm units are linked around the whole process, for example, warehouse position recommendation provides reasonable layout for order group wave and inventory distribution, and local optimization loss is avoided. And the instruction execution module controls the equipment according to the cooperative instruction, so that action conflicts are avoided, and the defect that a warehouse management system in the prior art cannot realize a full-process intelligent cooperative effect is overcome.

Description

Technical Field

[0001] This invention relates to the field of warehouse management technology, and in particular to an intelligent warehouse integrated management system. Background Technology

[0002] In the fields of logistics and intelligent manufacturing, warehouse management, as a key link in the supply chain, is facing increasingly higher requirements for operational efficiency, space utilization, and intelligence. Intelligent warehousing systems, through the combination of algorithmic decision-making and automated equipment, have become an important means to improve the efficiency of warehouse management.

[0003] While some existing intelligent warehousing systems have achieved basic automation, they still suffer from issues such as imprecise algorithmic decision-making and insufficient adaptability to dynamic business data when dealing with complex and ever-changing warehousing environments. For example, when faced with sudden increases or decreases in goods or temporary adjustments to storage locations, they struggle to quickly and accurately replan warehousing operations, impacting the smoothness of operations and reducing overall efficiency. Furthermore, existing intelligent warehousing systems have limitations in optimizing space utilization, failing to fully utilize factors such as goods characteristics and inbound / outbound frequencies for intelligent space allocation, easily leading to wasted storage space. Additionally, in terms of intelligence, existing intelligent warehousing systems lack sufficient integration with other related systems, resulting in delays and errors in information exchange, preventing truly intelligent collaborative operations across the entire process. Summary of the Invention

[0004] This invention provides an intelligent warehouse integrated management system to address the shortcomings of existing warehouse management systems that cannot achieve intelligent collaborative operation throughout the entire process.

[0005] On one hand, the present invention provides an intelligent warehouse integrated management system, comprising: The data processing module is used to acquire warehousing business data, which includes inbound information, outbound order information, and inventory information. An algorithm decision-making module, communicatively connected to the data processing module, is used to run a set of collaborative algorithm units to process the warehousing business data and generate decision instructions for optimizing warehousing operations. The algorithm units include at least: a storage location recommendation unit for generating storage locations based on the inbound information and the inventory information; an order grouping unit for clustering and reorganizing the outbound order information; an inventory allocation unit for matching picking schemes based on order demand and the inventory information; and a shelf popularity analysis unit for analyzing and predicting shelf popularity based on the outbound order information and generating shelf location adjustment strategies. The inventory allocation unit is further used to: call the shelf popularity prediction data output by the shelf popularity analysis unit to correct the picking scheme. The instruction execution module is communicatively connected to the algorithm decision module and is used to control the warehouse execution equipment to execute the decision instructions.

[0006] Optionally, the berth recommendation unit is used for: A warehouse location evaluation model is established, which is based on quantitative scoring of handling distance cost and the matching degree between warehouse space and cargo volume; The inventory quantity, volume, weight, and historical outbound frequency of goods entering the warehouse are collected and input into the warehouse evaluation model, and the scoring results are output. Storage locations are dynamically allocated to the incoming goods based on the scoring results; wherein, the incoming goods with a historical outbound frequency higher than a first threshold are allocated to storage locations within a first preset range from the picking station, and the incoming goods with a volume greater than a second threshold are allocated to storage locations that meet the load-bearing and space requirements.

[0007] Optionally, the order group wave unit is used for: The outbound order information is clustered based on the correlation between warehouse location and inventory level; The outbound order information is decomposed and reorganized using a dynamic programming algorithm to generate wave tasks with optimized picking paths and balanced task loads in each wave. The wave task is assigned to multiple automated guided vehicles or picking groups.

[0008] Optionally, the inventory allocation unit is used for: Collect order demand data and inventory status data. The order demand data includes the quantity of goods, delivery priority, and delivery time. The inventory status data includes the inventory quantity, batch number, and expiration date of each warehouse. Establish an optimization model aimed at improving order fulfillment rate, inventory turnover rate, and the first-in-first-out principle; Input the order demand data and the inventory status data into the optimization model, and output the inventory allocation strategy.

[0009] Optionally, the shelf heat analysis unit is used for: Collect historical outbound data; Input the historical outbound data into the Long Short-Term Memory network model, and output the predicted outbound volume; Based on the predicted outbound volume, the shelves are divided into several popularity levels; Based on the heat level, the automated guided vehicle is scheduled to dynamically adjust the shelf position within a preset time period; wherein, the higher the heat level, the closer the shelf is to the picking workstation.

[0010] Optionally, the inventory allocation unit is further configured to: Receive the predicted shelf popularity output by the shelf popularity analysis unit; Construct and solve a multi-objective optimization function to dynamically balance inventory turnover objectives and picking path efficiency objectives, and output the inventory allocation strategy.

[0011] Optionally, the algorithm decision module further includes a model adaptive learning unit; The model adaptive learning unit is used for: Collect historical execution feedback data from the instruction execution module; the historical execution feedback data includes the execution efficiency of decision instructions, the actual time consumed in warehousing operations, and resource consumption; The operating parameters of the warehouse recommendation unit, the order grouping unit, the inventory allocation unit, and the shelf popularity analysis unit are corrected based on the historical execution feedback data.

[0012] Optionally, the warehouse recommendation unit is used to: receive the historical order clustering features output by the order grouping unit, and use the historical order clustering features as weighting parameters of the warehouse evaluation model, so that frequently co-occurring goods are allocated to adjacent warehouses; The order grouping unit is used to: call the warehouse allocation data of the warehouse recommendation unit to aggregate goods orders in the same area into the same wave.

[0013] Optionally, it further includes: a digital twin optimization module, the digital twin optimization module comprising: Digital mirroring units are used to build virtual warehousing systems based on warehousing execution equipment and business rules; A simulation engine is used in the virtual warehouse system to load historical business data and run virtual algorithm units that are consistent with the algorithm unit logic in the algorithm decision module. The parameter optimization unit is used to adjust the running parameters of the virtual algorithm unit and perform iterative simulation through the simulation engine to search for and correct parameter combinations with system-level key performance indicators as the optimization target. The parameter synchronization interface is used to send the modified parameter combination to the algorithm decision module to update the running parameters of the corresponding algorithm unit.

[0014] Optionally, the parameter optimization unit is further configured to: In the virtual warehousing system, historical business data from the past N days is used as simulation input; Multiple sets of different candidate parameter combinations are set for the virtual algorithm unit; For each set of candidate parameter combinations, run the simulation engine to simulate warehouse operations over the next M days and record the simulation results; The simulation results of all candidate parameter combinations are sorted in descending order to select the correct parameter combination based on the sorting results.

[0015] The intelligent warehouse management system provided by this invention obtains inbound, outbound, and inventory data uniformly through a data processing module, avoiding decision-making biases caused by data fragmentation among different algorithm units. Within the algorithm decision-making module, each unit works collaboratively; in particular, the inventory allocation unit uses predictive data from the shelf popularity analysis unit to correct picking plans, no longer relying solely on orders and real-time inventory, and prioritizing picking from high-population, nearby shelves to reduce handling time. Furthermore, the algorithm units work in tandem throughout the entire process; for example, warehouse location recommendations provide a reasonable layout for order grouping and inventory allocation, avoiding losses from localized optimization. The instruction execution module controls equipment according to collaborative instructions, avoiding action conflicts, thus addressing the shortcomings of existing warehouse management systems that cannot achieve intelligent collaborative operation throughout the entire process. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is one of the schematic diagrams of the intelligent warehouse integrated management system provided in the embodiments of the present invention; Figure 2 This is the second schematic diagram of the intelligent warehouse integrated management system provided in the embodiment of the present invention; Figure 3 This is a schematic diagram of the algorithm decision module structure provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0019] Figure 1 This is one of the schematic diagrams of the intelligent warehouse integrated management system provided in the embodiments of the present invention.

[0020] like Figure 1 As shown, the intelligent warehouse integrated management system provided in this embodiment of the invention includes: The data processing module 110 is used to acquire warehousing business data, which includes inbound information, outbound order information, and inventory information.

[0021] For inbound information, the data processing module 110 connects to the barcode scanner and RFID reader at the warehouse entrance to obtain real-time information on the item's inbound time, name, specifications, quantity, supplier information, and storage location. When an item enters the warehouse, the barcode scanner or RFID reader automatically reads the barcode or electronic tag information on the item and transmits the information to the data processing module 110.

[0022] For outbound order information, the data processing module 110 interfaces with the enterprise's order management system. On one hand, it receives outbound request orders initiated by the sales department or other requesting departments, obtaining the order number, customer name, required item name, specifications, quantity, and required outbound time. On the other hand, during the outbound operation, it connects with the outbound operation terminal equipment in the warehouse to obtain real-time information on the actual outbound items, outbound time, and operator information, ensuring the consistency and accuracy of outbound information with order information.

[0023] For inventory information, the data processing module 110 employs a combination of periodic inventory checks and updates. During periodic inventory checks, in conjunction with handheld inventory terminals, staff meticulously count each item in the warehouse, recording the actual quantity and location of each item into the terminal and uploading it to the data processing module 110. Simultaneously, during daily inbound and outbound operations, the data processing module 110 automatically updates inventory quantities based on real-time inbound and outbound data, and records the history of inventory changes for inventory analysis and querying.

[0024] The algorithm decision module 120 is communicatively connected to the data processing module 110 and is used to run a group of collaborative algorithm units to process warehousing business data and generate decision instructions for optimizing warehousing operations.

[0025] The generated decision-making instructions cover multiple aspects. For example, based on the remaining quantity of inventory and demand forecasts, reasonable replenishment instructions are generated to ensure that inventory levels are maintained at an optimal state and to avoid inventory backlog or shortages. Based on the frequency and importance of items entering and leaving the warehouse, optimized instructions for item storage locations are generated, placing frequently accessed and important items in areas that facilitate quick access and improve warehousing efficiency. Combined with the urgency of orders and the actual operational capacity of the warehouse, scheduling instructions for outbound operations are generated to rationally arrange the outbound sequence and personnel, ensuring that outbound tasks are completed on time and accurately. These decision-making instructions are communicated to the corresponding execution units in a timely manner through communication connections with warehousing operation execution equipment, driving the efficient and orderly conduct of warehousing operations.

[0026] The algorithm units include, but are not limited to, inventory optimization algorithms, outbound route planning algorithms, and replenishment forecasting algorithms. The inventory optimization algorithm calculates the optimal inventory level based on historical inbound and outbound data and current inventory status to reduce inventory backlog and stockouts. The outbound route planning algorithm combines warehouse layout and outbound order information to plan the shortest or most economical outbound route, improving outbound efficiency. The replenishment forecasting algorithm analyzes sales data and inventory change trends to predict replenishment demand in the future, allowing for advance replenishment preparation. The algorithm decision module 120 generates specific decision instructions based on the processing results of these algorithm units, such as adjusting inventory strategies, optimizing outbound processes, and arranging replenishment plans.

[0027] Specifically, the algorithm unit includes at least: a storage location recommendation unit 1201 for generating storage locations based on inbound and inventory information; an order grouping unit 1202 for clustering and reorganizing outbound order information; an inventory allocation unit 1203 for matching picking plans based on order demand and inventory information; and a shelf popularity analysis unit 1204 for analyzing and predicting shelf popularity based on outbound order information and generating shelf location adjustment strategies. The inventory allocation unit 1203 is also used to: call the shelf popularity prediction data output by the shelf popularity analysis unit 1204 and correct the picking plan.

[0028] The warehouse recommendation unit 1201 takes the inbound and inventory information obtained by the data processing module 110 as input, and generates a storage warehouse plan by quantifying the adaptability of goods storage, such as the matching degree between goods volume and warehouse space, and the matching degree between goods weight and warehouse load-bearing capacity. The order grouping unit 1202 clusters outbound order information based on the distribution of goods warehouse areas and the correlation of goods corresponding to the order, and then reasonably reorganizes and splits large orders to form wave tasks with short picking paths and balanced task volume, reducing cross-regional repeated picking. The inventory allocation unit 1203 initially matches picking plans based on order demand and real-time inventory information, and also calls the shelf heat analysis unit 1204 to output the shelf heat prediction data. The initial plan is revised based on factors such as the probability of outbound shipments from the shelves in the future and the distance between the shelves and the picking station. Priority is given to picking goods from shelves with high demand and close to the picking station to avoid inefficient handling. The shelf demand analysis unit 1204 analyzes historical and real-time outbound order information to predict the demand level of different shelves and generates shelf location adjustment strategies accordingly. High demand shelves are moved closer to the picking station to further adapt to outbound operation needs.

[0029] The instruction execution module 130 is communicatively connected to the algorithm decision module 120 and is used to control the warehouse execution equipment to execute decision instructions.

[0030] The instruction execution module 130 first receives decision instructions output by the algorithm decision module 120. These instructions include operation types such as inbound, outbound, and shelf adjustment, and contain equipment identification, target parameters, and timing requirements. The module then parses and verifies the instructions, first matching the communication protocol of the corresponding warehouse execution equipment, and then checking the current status of the equipment to avoid conflicts. After successful verification, the instruction execution module 130 converts the decision instructions into control signals recognizable by the equipment, such as the path coordinates of the AGV (Automated Guided Vehicle), the grab position of the workstation, and the movement instructions of the shelf. It then sends these control signals to the target equipment in real time and monitors the equipment's execution progress throughout the process. For example, when the algorithm decision module 120 outputs an inbound instruction based on the warehouse recommendation unit 1201 to move three boxes of medicines with a specification of 600×400mm from the inspection area to warehouse A3-05, the instruction execution module 130 will first confirm the idle AGV number, parse out the parameters of the starting point: inspection area connection position, the destination: warehouse A3-05, and the load weight ≤90kg, convert them into navigation control signals for the AGV and send them out. During the AGV's movement, the instruction execution module 130 obtains the AGV's position through real-time communication. After the AGV reports that the goods have been put on the shelf, the instruction execution module 130 will send the execution result. The synchronous feedback algorithm decision module 120 completes the closed loop of the warehousing operation. For example, when the shelf heat analysis unit 1204 outputs an instruction to adjust the high-heat shelf B2-08 to within 10 meters of the picking workstation, the instruction execution module 130 will verify that there is currently no picking operation on the shelf and that the nearby AGV is idle. Then, it will send a control instruction to the AGV to connect the shelf B2-08 and move along the path P03 to the target area. At the same time, the picking workstation will suspend the temporary picking tasks related to the shelf until the AGV reports that the shelf is in place, and then resume the workstation operation to ensure that the equipment execution and the operation rhythm are coordinated.

[0031] Through the collaborative work of the data processing module 110, algorithm decision-making module 120, and instruction execution module 130, the intelligent warehouse integrated management system achieves automation, intelligence, and efficiency in warehouse operations. The data processing module 110, as the center for information collection and processing, ensures the real-time nature, accuracy, and completeness of warehouse data. The algorithm decision-making module 120, by running various algorithm units, performs in-depth analysis and processing of warehouse business data, generating scientifically sound decision instructions to guide the optimization of warehouse operations. The instruction execution module 130 is responsible for accurately transmitting these decision instructions to the warehouse execution equipment and monitoring the equipment's execution progress to ensure the effective execution of the decision instructions.

[0032] In practical applications, intelligent warehouse management systems can significantly improve the efficiency and quality of warehousing operations. By acquiring and processing warehousing data, problems in the warehousing process can be identified and resolved promptly, such as inventory backlog, stockouts, and outbound errors. Furthermore, intelligent warehouse management systems can dynamically adjust warehousing strategies and operational processes based on actual warehousing needs, adapting to the ever-changing market environment and customer demands.

[0033] In some embodiments, the position recommendation unit 1201 is used for: Establish a position evaluation model.

[0034] The warehouse location evaluation model uses a quantitative scoring system based on handling distance costs and the matching degree between warehouse space and cargo volume. When establishing the model, real-time data for each warehouse location is first collected, including the specific location of the location, available space, and the volume of currently stored goods. Then, handling distance costs—the length of the handling path from the goods' entry point to each location, as well as the time and labor costs required during handling—are assigned corresponding weights and quantitatively calculated. Simultaneously, the matching degree between warehouse space and cargo volume—that is, the degree to which the cargo volume fits the available warehouse space—is considered; a higher matching score is given when the cargo volume is closer to the warehouse space. By comprehensively quantitatively scoring these two key factors—handling distance costs and the matching degree between warehouse space and cargo volume—a comprehensive evaluation score is obtained for each warehouse location, which is then used to recommend the most suitable warehouse location for the goods.

[0035] The system collects the inventory quantity, volume, weight, and historical outbound frequency of goods entering the warehouse and inputs them into the warehouse evaluation model to output the scoring results.

[0036] During the data collection phase, the volume and weight of incoming goods are automatically acquired via IoT devices, while the inventory level and historical outbound frequency are retrieved from the data processing module 110. The data is then standardized and input into the warehouse location evaluation model.

[0037] The warehouse evaluation model first compares the volume with the spatial capacity of candidate warehouses to calculate the spatial matching degree; it then compares the weight with the warehouse's maximum load-bearing capacity to calculate the load-bearing suitability degree; and finally converts historical outbound frequency into distance weights, combining this with inventory levels to determine the required number of warehouses. Subsequently, the warehouse evaluation model uses a preset algorithm to quantify and score the spatial matching degree, load-bearing suitability degree, distance cost coefficient, and continuity coefficient, ultimately outputting the score results for each candidate warehouse. For example, warehouse 05 in area A scores 91 points, and warehouse 12 in area B scores 78 points.

[0038] Storage spaces are dynamically allocated to incoming goods based on the scoring results; Among them, goods with a historical outbound frequency higher than the first threshold are allocated to warehouses within a first preset range from the picking station, and goods with a volume greater than the second threshold are allocated to warehouses that meet the load-bearing and space requirements.

[0039] For inbound goods with a historical outbound frequency between the first and second thresholds and a volume smaller than the second threshold, they are prioritized for allocation to storage locations with higher scores and relatively closer to the picking station, based on the scoring results. For inbound goods with a historical outbound frequency below the second threshold and a smaller volume, they are allocated to remaining available storage locations in descending order of their comprehensive evaluation scores, provided that basic load-bearing and space requirements are met. During the dynamic allocation of storage locations, if multiple goods are suitable for the same storage location, the location is prioritized for goods with higher comprehensive evaluation scores, higher historical outbound frequencies, or larger volumes.

[0040] In some embodiments, the order grouping unit 1202 is used for: The outbound order information is clustered based on the correlation between warehouse location and inventory level.

[0041] The order grouping unit 1202 first extracts the warehouse storage location and SKU information of all outbound orders. By analyzing historical order data, it calculates the correlation between the inventory levels of different goods, i.e., the probability that two goods appear in the same order, such as the correlation between medicine A and medicine B. Subsequently, based on the dual clustering criteria of proximity of warehouse locations and high correlation between goods, outbound orders are divided into multiple cluster groups: for example, all orders involving goods in area A1 and containing highly correlated SKU combinations are clustered into one group, and orders involving low-correlation goods in area B3 are clustered into another group, ensuring that the goods in the same group are spatially concentrated and the goods combinations are synergistic, reducing cross-regional picking routes.

[0042] Dynamic programming algorithm is used to decompose and reorganize outbound order information to generate wave tasks with optimized picking paths and balanced task loads in each wave.

[0043] Based on the clustering results, the order group wave unit 1202 uses a dynamic programming algorithm to decompose and reorganize the order set within each cluster. The dynamic programming algorithm uses the shortest picking path per wave and balanced workload across waves as its objective functions, breaking down large order sets into multiple sub-task sets. For example, if a cluster contains 10 orders requiring the picking of 50 items, the dynamic programming algorithm calculates the location coordinates of different items in the warehouse area and plans three wave tasks: the first wave covers 18 items in columns 1-5 of area A1, with an estimated time of 12 minutes; the second wave covers 17 items in columns 6-10 of area A1, with an estimated time of 11 minutes; and the third wave covers 15 additional items picked at the edge of area A1, with an estimated time of 10 minutes. This ensures that the picking path in each wave is free of repetition and that the workload difference between waves is controlled within a preset threshold, preventing overload of any particular wave.

[0044] Distribute wave tasks to multiple automated guided vehicles or picking groups.

[0045] After generating wave tasks, the order group wave unit 1202 allocates tasks based on the real-time status of AGVs (Automated Guided Vehicles) or picking groups. For example, for three wave tasks covering area A1, the order group wave unit 1202 prioritizes assigning the first wave to AGV-01 currently located near the entrance of area A1, the second wave to AGV-02 which has just completed work near area A1, and the third wave to the manual picking group responsible for the edge area of ​​area A1. During allocation, detailed information about the wave tasks, such as the coordinates of the goods, picking order, and priority, is simultaneously sent out. Real-time communication ensures that the AGVs or picking groups execute along optimized paths, avoiding equipment conflicts and achieving efficient collaboration of multiple resources operating in parallel.

[0046] In some embodiments, the inventory allocation unit 1203 is used for: Collect order demand data and inventory status data. Order demand data includes the quantity of goods, delivery priority, and delivery time. Inventory status data includes the inventory quantity, batch number, and expiration date of each warehouse.

[0047] The inventory allocation unit 1203 collects order demand data and inventory status data from the order system and warehouse management system. The order demand data includes the quantity of goods, delivery priority, and delivery time for each outbound order. For example, an order might require 20 boxes of medicine A and 15 boxes of medicine B. Orders marked as urgent must be completed within 2 hours, while regular orders can be completed within 4 hours, with a latest delivery time of 2:00 PM. Regarding inventory status data, real-time inventory information for each warehouse location is simultaneously acquired, including the specific warehouse's inventory quantity, batch number, and expiration date. During the data collection process, barcode scanning or IoT sensors ensure real-time data accuracy. For instance, RFID tags on shelves dynamically update inventory quantities, and delivery priority identifiers are automatically captured through the order system interface, providing a data foundation for subsequent allocation decisions.

[0048] Establish an optimization model aimed at improving order fulfillment rate, inventory turnover rate, and the first-in-first-out (FIFO) principle.

[0049] Specifically, the inventory allocation unit 1203 establishes a multi-objective optimization model based on the collected data, transforming abstract objectives into quantifiable mathematical indicators. The order fulfillment rate objective is measured by the ratio of actual allocated quantity to order demand quantity, with weights assigned to prioritize high-priority orders; the inventory turnover rate objective is quantified by the frequency of goods leaving a warehouse location / the length of inventory backlog, prioritizing the allocation of inventory with longer backlog times; the first-in, first-out (FIFO) principle is based on the production time of goods batches, forcibly prioritizing the allocation of earlier batches.

[0050] The optimization model incorporates constraints, such as the maximum outbound volume of a single warehouse not exceeding the current inventory and the total allocation not being less than 95% of the order demand. The priorities of the three objectives are balanced through linear programming or heuristic algorithms to avoid the overall efficiency decline caused by the optimization of a single objective.

[0051] Input order demand data and inventory status data into the optimization model, and output inventory allocation strategy.

[0052] Specifically, after inputting the collected order demand data and inventory status data into the optimization model, the model calculates the optimal solution under each objective weight. For example, to meet the high priority of urgent orders, 18 boxes are allocated first from warehouse A3-05, which conforms to the first-in, first-out (FIFO) principle, and since this warehouse has a long inventory backlog time, it improves turnover rate. The remaining 2 boxes are replenished from warehouse B2-08. The final output inventory allocation strategy clearly states: Medicine A: 18 boxes are picked from warehouse A3-05, and 2 boxes are picked from warehouse B2-08, ensuring that the goods are shipped out before 14:00, which ensures timely delivery of orders while taking into account inventory turnover and batch management requirements.

[0053] In some embodiments, the shelf heat analysis unit 1204 is used for: Collect historical outbound data.

[0054] Specifically, when collecting historical outbound data, the shelf heat analysis unit 1204 prioritizes collecting historical outbound data covering the complete operational cycle from the warehouse management system, the operation logs of the instruction execution module 130, and the order management system. The collected historical outbound data includes: the SKU code of each outbound order, the precise outbound time, the quantity of each outbound order, the unique number of the shelf to which the goods belong, and the baseline distance from that shelf to each picking workstation. Simultaneously, the shelf heat analysis unit 1204 preprocesses the raw data, removing duplicate order records, supplementing shelf number information missing due to equipment malfunction, and filtering invalid data with an outbound quantity of 0 or abnormal outbound times, thus obtaining the historical outbound data.

[0055] Input historical outbound data into the Long Short-Term Memory network model, and output the predicted outbound volume.

[0056] Considering the significant time-series dependence of historical outbound data, the shelf popularity analysis unit 1204 employs a Long Short-Term Memory (LSTM) network model for outbound volume prediction. The LTM model captures long-term time-series patterns through a gating mechanism. The unit first divides the preprocessed historical data into daily time steps, using shelf number, daily average outbound volume, date type, and season label as input features to the LTM model. Then, it trains the model using 80% of the historical data and validates it with 20% of the data, iteratively optimizing by adjusting the number of hidden layer nodes and the learning rate until the prediction mean squared error is below 0.05. After training, the unit inputs the relevant features for the prediction period into the model and outputs the predicted outbound volume for each shelf within that period.

[0057] Based on the predicted outbound volume, the shelves are divided into several heat levels.

[0058] Specifically, the shelf heat analysis unit 1204, combined with the actual warehousing operation scenario, uses a quantile + business fine-tuning approach to classify heat levels. Taking 62 sets of shelves in a hospital pharmacy warehouse as an example, it first calculates the quantiles of the predicted outbound volume for all shelves, takes the 80th percentile as the high heat threshold, the 30th percentile as the medium heat threshold, and the quantile below 30% as the low heat, initially classifying them into three levels.

[0059] Further adjustments are made based on business needs, such as upgrading shelves for emergency medicines and shelves frequently used for urgent orders to higher priority levels to ensure efficient response to critical goods. Finally, a mapping table between shelf numbers and priority levels is generated.

[0060] Based on the heat level, the automated guided vehicles are scheduled to dynamically adjust the shelf positions within a preset time period; The higher the heat level, the closer the shelf is to the picking workstation. The shelf heat analysis unit 1204 first determines a preset adjustment time period, prioritizing non-peak warehouse operations to avoid conflicts with daytime inbound and outbound main processes. Then, the shelf heat analysis unit 1204 synchronizes the shelf-to-heat mapping table and the coordinates of each picking workstation to the robot scheduling system, formulating location adjustment rules. The location adjustment rules are: high-heat shelves need to be scheduled within 5 meters of the nearest picking workstation, medium-heat shelves within 8-12 meters, and low-heat shelves deeper into the warehouse area at least 15 meters away. Before adjustment, the shelf heat analysis unit 1204 first verifies the target shelf status and then assigns an idle AGV to perform the handling; after the AGV completes the shelf placement, it provides real-time feedback of the adjustment completion signal, and the shelf heat analysis unit 1204 immediately updates the shelf location information in the intelligent warehouse management system to ensure that subsequent inbound and outbound operations can accurately locate the new position.

[0061] In some embodiments, the inventory allocation unit 1203 is further configured to: Receive the predicted shelf popularity output by the shelf popularity analysis unit 1204; Construct and solve a multi-objective optimization function to dynamically balance inventory turnover objectives and picking path efficiency objectives, and output the inventory allocation strategy.

[0062] Specifically, the multi-objective optimization function comprehensively considers multiple factors such as inventory backlog time, shelf popularity, and picking path length. Through weighted summation or constraint optimization, it finds the inventory allocation scheme that meets inventory turnover requirements and maximizes picking path efficiency. In practical applications, the inventory allocation unit 1203 adjusts the inventory allocation strategy in real time based on changes in predicted shelf popularity to ensure the efficient operation of the warehousing system.

[0063] When constructing a multi-objective optimization function, the first step is to determine the weight coefficients of each objective. These weight coefficients reflect the importance of inventory turnover and picking path efficiency in the overall optimization. For example, if inventory turnover is given more importance, the weight of inventory backlog time should be increased accordingly; if picking efficiency is more important, the weight of picking path length should be increased. After determining the weights, inventory backlog time, shelf popularity, and picking path length are substituted into the multi-objective optimization function and solved using linear programming. During the solution process, the variable values ​​are continuously adjusted to find an inventory allocation scheme that maximizes the multi-objective optimization function. The inventory allocation scheme must not only meet the current order's inventory requirements but also consider inventory turnover over a future period, while minimizing picking paths and improving overall operational efficiency.

[0064] In some embodiments, such as Figure 3 As shown, the algorithm decision module 120 also includes a model adaptive learning unit 1205; The model adaptive learning unit 1205 is used for: Collect historical execution feedback data from instruction execution module 130.

[0065] Historical execution feedback data includes the execution efficiency of decision-making instructions, the actual time consumed in warehousing operations, and resource consumption. The model adaptive learning unit 1205 continuously collects historical execution feedback data reflecting the effectiveness of decision-making instructions. The execution efficiency of decision-making instructions is reflected in the responsiveness and completion rate of each device to the instruction; the actual time consumed in warehousing operations includes the actual handling time for a single batch of goods entering the warehouse, the actual picking time for a certain batch of orders, and the actual execution time for shelf adjustments. Resource consumption covers the energy consumption per unit task of the AGV, the percentage of standby time for picking equipment, and the space occupation cost during shelf adjustments. During the collection process, the model adaptive learning unit 1205 timestamps the data, filters outliers, and stores it according to algorithm units, forming structured historical execution feedback data.

[0066] The operating parameters of the warehouse recommendation unit 1201, order grouping unit 1202, inventory allocation unit 1203, and shelf heat analysis unit 1204 are corrected based on historical execution feedback data.

[0067] Specifically, the model adaptive learning unit 1205 performs targeted parameter correction based on the decision logic of different algorithm units and historical execution feedback data.

[0068] For the warehouse recommendation unit 1201, if the feedback data shows that the actual handling time for goods entering the warehouse at the recommended warehouse location is consistently higher than the predicted value, it indicates that the distance cost coefficient in the warehouse evaluation model is not set reasonably. The model adaptive learning unit 1205 will increase the distance weight of high-frequency outbound goods based on the actual time deviation rate and prioritize the allocation of goods to closer warehouse locations.

[0069] For order group wave unit 1202, if the feedback shows that the actual time taken for a certain wave of tasks is much longer than that of other waves, the task load balancing coefficient in the dynamic programming algorithm is corrected by analyzing the cargo distribution characteristics of the wave, so as to avoid task overload in subsequent wave groups.

[0070] For inventory allocation unit 1203, if the feedback high-priority order fulfillment rate is less than 90%, the target weight of delivery priority in the optimization model is increased, such as from 0.4 to 0.5, to ensure that urgent orders are fulfilled first.

[0071] For the shelf heat analysis unit 1204, if the deviation rate between the predicted outbound quantity and the actual outbound quantity is greater than 15%, the time window parameters or learning rate of the Long Short-Term Memory network model will be adjusted to improve the prediction accuracy. After correction, the model adaptive learning unit 1205 will iteratively optimize through a process of small-batch testing and effect verification to ensure that the parameter adjustment can actually improve the decision adaptability of each algorithm unit.

[0072] In some embodiments, the position recommendation unit 1201 is used for: The system receives the historical order clustering features output by the order group wave unit 1202 and uses these features as weighting parameters for the warehouse evaluation model, so that frequently co-occurring goods are allocated to adjacent warehouses.

[0073] Specifically, the warehouse recommendation unit 1201 receives historical order clustering features output by the order grouping unit 1202. These features include high-frequency co-occurring product combinations, product association strength coefficients, and the average number of co-occurring orders. Specifically, co-occurring product combinations include the percentage of times drug A and drug B appear together in the same order within the past 30 days. The product association strength coefficient quantifies the probability of different products being shipped out simultaneously, ranging from 0 to 1; a higher coefficient indicates a stronger association. The average number of co-occurring orders includes the average number of orders where A and B are shipped out together.

[0074] Subsequently, the warehouse location recommendation unit 1201 transforms the features into new weighted parameters for the warehouse location evaluation model, which participate in the scoring in conjunction with the original parameters. For example, for a combination of goods with a correlation strength coefficient > 0.8, when calculating the candidate warehouse location score, if the target warehouse locations of the two goods are adjacent, the score of the combination is increased by an additional 20% weight; if the warehouse locations are not adjacent, the weight is reduced by 10%. Through this mechanism, when allocating warehouse locations for goods, the warehouse location recommendation unit 1201 will prioritize arranging frequently co-occurring goods in adjacent or the same area, avoiding cross-regional picking caused by the dispersed storage of related goods during subsequent outbound operations, thus improving wave grouping efficiency from the storage source.

[0075] The order grouping unit 1202 is used to: call the warehouse allocation data of the warehouse recommendation unit 1201 and aggregate the goods orders of warehouses in the same area into the same wave.

[0076] Specifically, before clustering and reorganizing outbound orders, the order grouping unit 1202 retrieves the latest warehouse allocation data from the warehouse recommendation unit 1201. This data includes the specific warehouse number and regional division for each SKU, such as drug C stored in warehouses 01-05 of zone A1, drug D in warehouses 06-10 of zone A1, and drug E in warehouse 03 of zone B2. The unit extracts the warehouse region information for all goods in the order. Then, using warehouse region consistency as the clustering condition, the order grouping unit 1202 aggregates all orders involving warehouses in zone A1 into the same wave, and orders involving warehouses in zone B2 into another wave. For orders containing goods across regions, they are split into waves closer to the main goods region based on the principle of closest regional distance. For example, if goods in zone A1 account for 70% of an order, it is split into a wave in zone A1. By calling warehouse allocation data to aggregate orders in the same area, the order group wave unit 1202 can significantly reduce the cross-regional movement distance of a single wave of picking and improve the operating efficiency of AGVs or picking groups.

[0077] In some embodiments, such as Figure 3 As shown, it also includes: a digital twin optimization module 140, which includes: Digital mirroring unit 1401 is used to build a virtual warehousing system based on warehousing execution equipment and business rules.

[0078] The digital mirror unit 1401 collects real-time data from warehousing execution equipment, such as shelf status, AGV vehicle position and operating status, and conveyor belt operation, and combines this data with preset business rules to construct a virtual warehousing system that is highly consistent with the actual warehousing system. This virtual warehousing system can reflect the dynamic changes in the actual warehousing in real time.

[0079] Simulation engine 1402 is used to load historical business data in the virtual warehouse system and run virtual algorithm units that are consistent with the algorithm unit logic in algorithm decision module 120.

[0080] The simulation engine 1402 can read and analyze historical business data, including historical order information, goods storage status, and equipment operation records. By running a virtual algorithm unit with the same logic as the algorithm unit in the algorithm decision module 120, the simulation engine 1402 can simulate warehousing operations under different business scenarios in the virtual warehousing system. For example, it can simulate the operational efficiency and resource utilization of the warehousing system under specific order volumes, goods distribution, and equipment operating conditions. Using the simulation results from the simulation engine 1402, managers can predict the impact of different business strategies on the warehousing system in advance, thereby providing data support for optimizing actual warehousing operations.

[0081] The parameter optimization unit 1403 is used to adjust the running parameters of the virtual algorithm unit and to perform iterative simulation through the simulation engine 1402, using system-level key performance indicators as the optimization target to search for and correct parameter combinations.

[0082] The parameter optimization unit 1403 can modify various parameters in the virtual algorithm unit according to different business scenarios and optimization needs. After each parameter adjustment, iterative simulation operations are quickly carried out with the help of the simulation engine 1402, taking system-level key performance indicators, such as the accuracy of warehousing operations, order processing time, and equipment failure rate, as important optimization targets. By continuously trying different parameter combinations and using the simulation results fed back by the simulation engine 1402, the corrected parameter combination that can make the system achieve the best performance is found, thereby providing a basis for the optimization of the actual warehousing system algorithm.

[0083] The parameter synchronization interface 1404 is used to send the modified parameter combination to the algorithm decision module 120 to update the running parameters of the corresponding algorithm unit.

[0084] The parameter synchronization interface 1404 has data transmission capabilities, ensuring that the corrected parameter combination is accurately and quickly transmitted to the algorithm decision module 120. Upon receiving the corrected parameter combination, the algorithm decision module 120 immediately updates the operating parameters of the corresponding algorithm unit, thereby improving the operational efficiency and accuracy of the entire intelligent warehouse management system.

[0085] In some embodiments, the parameter optimization unit 1403 is further configured to: In a virtual warehouse system, historical business data from the past N days is used as simulation input.

[0086] Set multiple different combinations of candidate parameters for the virtual algorithm unit; For each set of candidate parameter combinations, run simulation engine 1402 to simulate warehouse operations over the next M days and record the simulation results; The simulation results of all candidate parameter combinations are sorted in descending order to select the correct parameter combination based on the sorting results.

[0087] Specifically, historical business data from the past N days is extracted from the historical database of the warehousing system. This historical business data includes various aspects of warehousing operations, such as inbound, outbound, and inventory counts. Next, the parameter optimization unit 1403 generates multiple sets of different candidate parameter combinations for the virtual algorithm unit based on preset rules or experience. Each parameter combination represents an operating strategy.

[0088] Then, for each set of candidate parameter combinations, the simulation engine 1402 will run a simulation in the virtual warehouse system to predict the warehouse operation situation in the next M days and record the simulation results, including key performance indicators such as the accuracy of warehouse operations, order processing time, and equipment utilization.

[0089] Finally, the parameter optimization unit 1403 sorts the simulation results of all candidate parameter combinations in descending order and prioritizes the parameter combination that can make the system performance optimal as the correction parameter combination to optimize the performance of the intelligent warehouse integrated management system.

[0090] As can be seen from the above technical solutions, the present invention provides an intelligent warehouse integrated management system. The data processing module 110 uniformly acquires inbound, outbound, and inventory data, avoiding decision-making biases caused by data fragmentation among different algorithm units. In the algorithm decision-making module 120, each unit works collaboratively; in particular, the inventory allocation unit 1203 calls on the predictive data from the shelf popularity analysis unit 1204 to correct the picking plan, no longer relying solely on orders and real-time inventory, and prioritizing picking from high-population, nearby shelves to reduce handling time. Each algorithm unit also works in conjunction with the entire process, such as providing reasonable layouts for order grouping and inventory allocation through warehouse location recommendations, avoiding losses from localized optimization. The instruction execution module 130 controls the equipment according to collaborative instructions, avoiding action conflicts, thus solving the shortcomings of existing warehouse management systems that cannot achieve intelligent collaborative operation throughout the entire process.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent warehouse integrated management system, characterized in that, include: The data processing module is used to acquire warehousing business data, which includes inbound information, outbound order information, and inventory information. The algorithm decision module is communicatively connected to the data processing module and is used to run a group of collaborative algorithm units to process the warehousing business data and generate decision instructions for optimizing warehousing operations. The algorithm unit includes at least: a storage location recommendation unit for generating storage locations based on the inbound information and the inventory information; an order grouping unit for clustering and reorganizing the outbound order information; an inventory allocation unit for matching picking schemes with order demand and the inventory information; and a shelf popularity analysis unit for analyzing and predicting shelf popularity based on the outbound order information and generating shelf location adjustment strategies; wherein, the inventory allocation unit is further used to: call the shelf popularity prediction data output by the shelf popularity analysis unit to correct the picking scheme; The instruction execution module is communicatively connected to the algorithm decision module and is used to control the warehouse execution equipment to execute the decision instructions.

2. The intelligent warehouse integrated management system according to claim 1, characterized in that, The berth recommendation unit is used for: A warehouse location evaluation model is established, which is based on quantitative scoring of handling distance cost and the matching degree between warehouse space and cargo volume; The inventory quantity, volume, weight, and historical outbound frequency of goods entering the warehouse are collected and input into the warehouse evaluation model, and the scoring results are output. Storage locations are dynamically allocated to the incoming goods based on the scoring results; wherein, the incoming goods with a historical outbound frequency higher than a first threshold are allocated to storage locations within a first preset range from the picking station, and the incoming goods with a volume greater than a second threshold are allocated to storage locations that meet the load-bearing and space requirements.

3. The intelligent warehouse integrated management system according to claim 1, characterized in that, The order group wave unit is used for: The outbound order information is clustered based on the correlation between warehouse location and inventory level; The outbound order information is decomposed and reorganized using a dynamic programming algorithm to generate wave tasks with optimized picking paths and balanced task loads in each wave. The wave task is assigned to multiple automated guided vehicles or picking groups.

4. The intelligent warehouse integrated management system according to claim 1, characterized in that, The inventory allocation unit is used for: Collect order demand data and inventory status data. The order demand data includes the quantity of goods, delivery priority, and delivery time. The inventory status data includes the inventory quantity, batch number, and expiration date of each warehouse. Establish an optimization model aimed at improving order fulfillment rate, inventory turnover rate, and the first-in-first-out principle; Input the order demand data and the inventory status data into the optimization model, and output the inventory allocation strategy.

5. The intelligent warehouse integrated management system according to claim 1, characterized in that, The shelf heat analysis unit is used for: Collect historical outbound data; Input the historical outbound data into the Long Short-Term Memory network model, and output the predicted outbound volume; Based on the predicted outbound volume, the shelves are divided into several popularity levels; Based on the heat level, the automated guided vehicle is scheduled to dynamically adjust the shelf position within a preset time period; wherein, the higher the heat level, the closer the shelf is to the picking workstation.

6. The intelligent warehouse integrated management system according to claim 4, characterized in that, The inventory allocation unit is also used for: Receive the predicted shelf popularity output by the shelf popularity analysis unit; Construct and solve a multi-objective optimization function to dynamically balance inventory turnover objectives and picking path efficiency objectives, and output the inventory allocation strategy.

7. The intelligent warehouse integrated management system according to claim 1, characterized in that, The algorithm decision module also includes a model adaptive learning unit; The model adaptive learning unit is used for: Collect historical execution feedback data from the instruction execution module; the historical execution feedback data includes the execution efficiency of decision instructions, the actual time consumed in warehousing operations, and resource consumption; The operating parameters of the warehouse recommendation unit, the order grouping unit, the inventory allocation unit, and the shelf popularity analysis unit are corrected based on the historical execution feedback data.

8. The intelligent warehouse integrated management system according to claim 2, characterized in that, The warehouse recommendation unit is used to: receive the historical order clustering features output by the order grouping unit, and use the historical order clustering features as weighting parameters of the warehouse evaluation model, so that frequently co-occurring goods are allocated to adjacent warehouses; The order grouping unit is used to: call the warehouse allocation data of the warehouse recommendation unit to aggregate goods orders in the same area into the same wave.

9. The intelligent warehouse integrated management system according to claim 7, characterized in that, Also includes: A digital twin optimization module, comprising: Digital mirroring units are used to build virtual warehousing systems based on warehousing execution equipment and business rules; A simulation engine is used in the virtual warehouse system to load historical business data and run virtual algorithm units that are consistent with the algorithm unit logic in the algorithm decision module. The parameter optimization unit is used to adjust the running parameters of the virtual algorithm unit and perform iterative simulation through the simulation engine to search for and correct parameter combinations with system-level key performance indicators as the optimization target. The parameter synchronization interface is used to send the modified parameter combination to the algorithm decision module to update the running parameters of the corresponding algorithm unit.

10. The intelligent warehouse integrated management system according to claim 9, characterized in that, The parameter optimization unit is also used for: In the virtual warehousing system, historical business data from the past N days is used as simulation input; Multiple sets of different candidate parameter combinations are set for the virtual algorithm unit; For each set of candidate parameter combinations, run the simulation engine to simulate warehouse operations over the next M days and record the simulation results; The simulation results of all candidate parameter combinations are sorted in descending order to select the correct parameter combination based on the sorting results.

Images