Intelligent binding system and method for goods allocation of stereoscopic warehouse

Abstract

The invention discloses a stereoscopic warehouse goods allocation intelligent binding system and method, and relates to the technical field of stereoscopic warehouse storage management. Comprising the following steps: S1, creating a two-dimensional electronic map of a warehouse, establishing a two-dimensional coordinate system of the warehouse, and based on the two-dimensional coordinate system and shelf positions, establishing coordinates of each shelf by taking a warehouse-out station position as a coordinate origin; s2, the RFID tag is installed at the preset position of the first layer of storage location of each goods shelf, and the coordinates of the goods shelf where the RFID tag is located are pre-stored in each RFID tag; by integrating the advanced navigation technology, the AI intelligent decision-making system and the multi-mode interaction unit, accurate and efficient binding of the goods locations of the stereoscopic warehouse is achieved. The system not only fully considers the three-dimensional space characteristics of the warehouse, but also combines the multi-dimensional factors such as dynamic heat and seasonal demand change of the goods, so that the goods allocation is more scientific and reasonable, and compared with the prior art, the system can dynamically adjust the goods allocation strategy according to the real-time demand.

Description

Technical Field

[0001] This invention relates to the field of automated warehouse storage management technology, specifically to an intelligent binding system and method for automated warehouse locations. Background Technology

[0002] With the rapid development of modern logistics, automated warehouses, as the core hub of the logistics system, directly determine the response speed and operating costs of the entire supply chain through their management efficiency. In scenarios such as material management in water pollution testing laboratories, automotive parts distribution, and e-commerce retail warehousing, how to quickly and accurately bind incoming goods to their three-dimensional storage locations within an automated warehouse and achieve dynamic optimization management has always been a key challenge in the field's technological development.

[0003] Chinese invention patent application publication number CN120146768A discloses a method and system for optimizing warehouse storage layout based on big data visualization analysis. The method includes: acquiring a heatmap of recent warehouse access probabilities and data on outbound product categories from the same batch; adjusting the product layout in multiple high-frequency return routes determined by the outbound product category data based on high-correlation product combinations; after adjustment, selecting a target order for product placement in each high-frequency return route based on the smoothness of each placement order; and calculating the optimization degree of each high-frequency return route based on the smoothness of the target order and the length of the high-frequency return route. Therefore, the warehouse storage layout is adjusted based on the target order of the high-frequency return routes with higher optimization degrees. This invention can improve the retrieval efficiency of picking personnel in the warehouse.

[0004] Existing warehouse location layout optimization methods and systems involve placing highly related goods together on the same route to improve retrieval efficiency. However, existing technologies do not fully consider the three-dimensional spatial characteristics of automated warehouses and the impact of multi-dimensional factors such as dynamic heat of goods and seasonal demand changes on location allocation. Therefore, existing technologies are unable to dynamically adjust location allocation strategies according to real-time demand. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent binding system and method for storage locations in an automated warehouse to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent warehouse location binding system, comprising the following steps:

[0007] S1. Create a two-dimensional electronic map of the warehouse, establish a two-dimensional coordinate system for the warehouse, and based on the two-dimensional coordinate system and the location of the shelves, establish the coordinates of each shelf with the location of the outbound counter as the origin.

[0008] S2. Install RFID tags in the preset positions on the first floor of each shelf. Each RFID tag has the coordinates of the shelf where it is located stored in the preset position.

[0009] S3. Task Triggering and Data Preparation: The operator scans the barcode of the goods to be put into the warehouse using the PDA on the vehicle terminal to obtain the goods information. The PDA then sends the goods information to the backend server via the network, triggering a storage location allocation request.

[0010] S4. Intelligent warehouse location allocation decision: The back-end server receives the warehouse location allocation request and inputs the data into the AI ​​optimization model. The AI ​​optimization model outputs the optimal warehouse location space coordinates for the goods to be put into the warehouse based on a multi-objective optimization algorithm. The input data of the AI ​​optimization model includes at least: the goods information, historical operation data, historical inbound and outbound data stored in the warehouse for at least 90 days, goods correlation data, and real-time order pool data.

[0011] S5. Navigation guidance: The backend server sends the optimal storage location coordinates to the PDA that initiated the request. The navigation system on the PDA generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage location coordinates, and displays it on the screen.

[0012] S6. Multimodal detection and guidance guides the operator to operate the forklift to place the goods to be stored on the corresponding storage location of the target shelf;

[0013] S7. The acquired cargo information and cargo location coordinates are bound in the system database. After successful binding, the PDA's graphical interface displays the binding success status, and the voice prompt system broadcasts the operation completion confirmation information.

[0014] Furthermore, the specific method for guiding multimodal detection in step S6 is as follows:

[0015] The operator drives the forklift along the navigation path to the target shelving area;

[0016] When the forklift approaches the target shelf, the RFID reader fixedly installed on the top of the forklift reads the RFID tag fixedly installed at the preset position of the first shelf location in a directional identification manner, and obtains the planar coordinates (X, Y) of the target location after decoding.

[0017] The Z-axis measurement module installed on the forklift measures and calculates the height coordinates of the forks or goods relative to the warehouse floor.

[0018] Furthermore, the Z-axis measurement module adopts a dual-redundancy fusion scheme of visual measurement and laser ranging, the process of which includes:

[0019] Images are captured by a top camera and a first height value is calculated based on a visual algorithm. A second height value is measured and converted by a laser range sensor at the front of the forks. Finally, the two height values ​​are weighted and fused to output the final height coordinates.

[0020] Furthermore, the specific method for outputting the optimal storage location coordinates for the goods to be put into storage based on the multi-objective optimization algorithm in step S4 is as follows:

[0021] P1: Extract relevant data from the historical data warehouse and calculate the correlation score between goods, the dynamic popularity score of goods, and the seasonal predictive score;

[0022] P2: Calculate the predictive picking path cost, the cost of inter-cargo correlation, the cost of dynamic cargo heat, the seasonal predictive cost, and the real-time roadway operation cost;

[0023] P3: Calculate the total cost of an idle storage space using the total cost function formula. The total cost function formula for an idle storage space is:

[0024] C(shelf)=w1×C(dis)+w2×C(ass)-w3×C(heat)-w4×C(sea)+w5×C(con);

[0025] Among them, w1, w2, w3, w4 and w5 are weighting coefficients, C(dis) is the predictive picking path cost, C(ass) is the cost of inter-cargo correlation, C(heat) is the cost of dynamic heat of cargo, C(sea) is the seasonal predictive cost, and C(con) is the real-time operation cost of the alley.

[0026] P4: The vacant storage location with the lowest total cost value is identified as the storage location to be put into storage.

[0027] Furthermore, the method for calculating the predictive picking path cost in step P2 is as follows:

[0028] The path planning engine simulates the total path length d that the forklift needs to travel when picking orders containing the goods to be put into storage.

[0029] The predictive picking path cost is obtained by calculating the formula C(dis)=(d-dmin) / (dmax-dmin), where dmin and dmax are the shortest and longest possible paths in the simulated scenario, respectively.

[0030] Furthermore, the calculation method for the cost of inter-goods correlation in step P2 is as follows:

[0031] The top K goods with the highest correlation to the goods to be put into storage, obtained through data mining, are used to form a set of related goods.

[0032] Calculate the average Euclidean distance between the goods to be received and all goods in the associated goods set when placing the goods to be received in the candidate storage location;

[0033] The cost of inter-goods association is obtained by calculating C(ass) = min(Davg / Dmax, 1), where Davg is the average Euclidean distance of all goods in the associated goods set, and Dmax is the maximum diagonal distance between warehouses.

[0034] Furthermore, the calculation method for the dynamic heat cost of goods in step P2 is as follows:

[0035] Based on historical inbound and outbound data, calculate the dynamic heat score of goods and normalize it;

[0036] The convenience index of candidate storage locations is calculated. This index is determined by the normalized distance from the storage location to the outbound station and the normalized value of the height of the available storage location. The calculation formula is Ibin=α×(1-Dbin)+β×(1-|Zbin-Zopt|), where α and β are weighting coefficients, Dbin is the normalized distance from the storage location to the outbound station, Zbin is the normalized height of the available storage location, and Zopt is the normalized value corresponding to the optimal operating height.

[0037] The dynamic heat cost of goods is obtained by calculating the formula C(heat)=Hgui×Ibin, where Hgui is the normalized score of dynamic heat.

[0038] Furthermore, the method for calculating the seasonal predictive cost in step P2 is as follows:

[0039] Obtain the seasonal forecast score;

[0040] Obtain the seasonal handling capacity score for the area where the candidate storage location is located;

[0041] The seasonal predictive cost is obtained by calculating C(sea)=(Cseb-1)×Rbin, where Cseb is the seasonal cargo prediction score and Rbin is the seasonal handling capacity score of the area where the candidate cargo location is located.

[0042] An intelligent location binding system for automated warehouses includes:

[0043] The vehicle-mounted terminal subsystem is installed in the cab of the forklift;

[0044] The positioning and sensing module includes RFID tags and RFID readers. The RFID tags are installed at preset positions on the first floor of each shelf. Each RFID tag has its own pre-stored coordinates of the shelf. The RFID reader is fixedly installed on the top of the forklift to read the RFID tags and obtain the planar coordinates of the storage location after decoding.

[0045] The Z-axis measurement module includes a camera and a laser rangefinder. The camera is installed at the top of the forklift facing the rack to acquire images and calculate a first height value based on a visual algorithm. The laser rangefinder is installed at the front of the forks with its probe pointing vertically downward to measure the height between the forks and the ground and calculate a second height value. The Z-axis measurement module is also equipped with a processing unit to perform confidence-weighted fusion of the two height values ​​and output the final height coordinates.

[0046] The navigation subsystem, integrated into the PDA of the vehicle terminal subsystem, generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage space coordinates, and displays it on the screen;

[0047] The back-end server is connected to the vehicle terminal subsystem via the network, receives cargo space allocation requests, and inputs the data into the AI ​​optimization model;

[0048] The AI ​​intelligent decision-making system includes a big data management module, a dynamic storage location allocation engine, a path planning and simulation engine, and an adaptive learning engine. It is deployed on a central server. The AI ​​intelligent decision-making system outputs the optimal storage location coordinates for goods to be put into storage based on a multi-objective optimization algorithm, and sends the optimal storage location coordinates to the PDA that initiated the request.

[0049] The database is used to store cargo information, historical operational data, cargo-related data, real-time order pool data, and the binding relationship between cargo location coordinates and cargo information;

[0050] The voice prompt system, connected to the vehicle terminal subsystem, is used to announce a confirmation message after successful binding.

[0051] Furthermore, the navigation subsystem also includes a multimodal interaction unit, which, after performing path planning, displays the planned path in a highlighted color on a two-dimensional map and displays congestion areas predicted based on real-time data in a contrasting color.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] This intelligent warehouse location binding system and method integrates advanced navigation technology, an AI intelligent decision-making system, and a multimodal interaction unit to achieve precise and efficient location binding in automated warehouses. The system not only fully considers the three-dimensional spatial characteristics of the warehouse but also incorporates multi-dimensional factors such as the dynamic heat of goods and seasonal demand changes, making location allocation more scientific and rational. Compared to existing technologies, this invention can dynamically adjust location allocation strategies based on real-time demand, significantly improving warehouse management efficiency and response speed while reducing operating costs. Attached Figure Description

[0054] Figure 1 This is a flowchart of the present invention;

[0055] Figure 2 This is a flowchart for calculating the optimal available storage space in this invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] This invention provides a technical solution: a method for intelligent binding of storage locations in an automated warehouse, such as... Figure 1 As shown, it includes the following steps:

[0058] S1. Create a two-dimensional electronic map of the warehouse and establish a two-dimensional coordinate system for the warehouse. Based on the two-dimensional coordinate system and the location of the shelves, establish the coordinates of each shelf with the location of the outbound platform as the origin. Scan the actual layout of the warehouse using high-precision surveying equipment or LiDAR to map the shelf locations, aisle widths, and obstacle areas to the two-dimensional coordinate system. The coordinate accuracy error is controlled within ±5cm. At the same time, mark the prohibited areas of the outbound platform, charging area, and safety passage to form a dynamically updated electronic map base map.

[0059] S2. Install RFID tags in the preset positions on the first floor of each shelf. In this solution, the height of the bottom shelf from the ground is 0.5m, and the height between each shelf is 0.5m. Each RFID tag has the coordinates (X, Y) of the shelf it is located in. The RFID tag has a built-in encryption chip that stores the shelf coordinates and a unique identification code. The RFID tag can be read within a range of 0-2 meters.

[0060] S3. Task Triggering and Data Preparation: The operator scans the barcode of the goods to be put into storage using a PDA on the vehicle terminal to obtain the goods information. The PDA then sends the goods information to the backend server via the network, triggering a storage location allocation request. Upon receiving the storage location allocation request, the backend server immediately initiates a data verification process, performing multiple verifications on the completeness of the goods information, the validity of the barcode, and the stability of network transmission. The system automatically records the timestamp of the task trigger and the operator ID, providing data support for subsequent process traceability and performance analysis. During the data preparation phase, the backend server also retrieves historical operational data related to the goods from the database, including past inbound and outbound frequencies, related goods combination information, and seasonal demand fluctuation records, providing comprehensive data basis for intelligent storage location allocation decisions.

[0061] S4. Intelligent warehouse location allocation decision: The back-end server receives the warehouse location allocation request and inputs the data into the AI ​​optimization model. The AI ​​optimization model outputs the optimal warehouse location space coordinates for the goods to be put into the warehouse based on a multi-objective optimization algorithm. The input data of the AI ​​optimization model includes at least: goods information, historical operation data, historical inbound and outbound data stored in the warehouse for at least 90 days, goods correlation data, and real-time order pool data.

[0062] like Figure 2 As shown, the specific method for allocating the optimal storage location coordinates to goods awaiting warehousing based on the output of a multi-objective optimization algorithm is as follows:

[0063] P1: Extract relevant data from the historical data warehouse and calculate the correlation score between goods, the dynamic heat score of goods, and the seasonal predictive score.

[0064] The calculation method for the dynamic heat index score of goods is as follows:

[0065] To determine the attenuation weight value for each item, we calculate the total number of outbound shipments (Fout) and the total number of inbound shipments (Fin) per day over the past T = 30 working days. The formula for the attenuation weight value is as follows: Where T is the half-life of 15 days, t = 30 - d, the calculation is performed from day 1 to day 30, and the weight of the outbound data for each day is calculated. Specifically, today's value is 1, yesterday's value is 0.96, the day before yesterday's value is 0.91, day 3's value is 0.87, and for each additional day thereafter, the values ​​are 0.83, 0.79, 0.76, 0.72, 0.69, 0.66, 0.62, 0.60, ..., 0.29, 0.27, 0.26.

[0066] Calculate and weight the outbound popularity score of goods A. Assume that the total number of outbound shipments of each goods in the past T = 30 working days is as follows: today, the score is 8; yesterday, the score is 7; the day before yesterday, the score is 5; the 3rd day, the score is 6; and the scores for each day prior to that are 4, 6, 3, 2, 1, 4, 2, 6, 3, 2, 1, 4, 6, 3, 2, 1, 4, 2, 6, 3, 2, 1, 4, 6, 3, 2. The weighted outbound popularity score of goods A is 58.68.

[0067] Similarly, for ease of calculation, assume that the number of outbound shipments per day is 8, 7, 5, 6, 4, 6, 3, 2, 1, 4, 2, 6, 3, 2, 1, 4, 6, 3, 2, 1, 4, 2, 6, 3, 2, 1, 4, 6, 3, 2, and the weighted inbound popularity score of goods A is 58.68.

[0068] The dynamic heat score of goods = λ × outbound heat score of goods A + (1-λ) × inbound heat score of goods A. With λ set to 0.7, the dynamic heat score of goods A is calculated to be 58.66.

[0069] The method for calculating the correlation score between goods is as follows: Using the FP-Growth algorithm, the correlation set of goods A is found, and the top 3 goods with the highest correlation and their location coordinates are determined. Assuming the coordinates of the 3 goods are (20, 30), (40, 60), and (80, 75), the average Euclidean distance between the 3 goods is... It is approximately equal to 51.26.

[0070] The seasonal predictive score is calculated by obtaining the sales forecast data for the next 30 days of product A through the ERP system and the average monthly sales volume of product A in the historical operating data. The seasonal predictive score is obtained by dividing the sales forecast data by the average monthly sales volume. Assuming the sales forecast data is 150 units and the average monthly sales volume is 100 units, then the seasonal predictive score is 1.5.

[0071] P2: Calculate the predictive picking path cost, the cost of inter-cargo correlation, the cost of dynamic cargo heat, the seasonal predictive cost, and the real-time roadway operation cost;

[0072] Assume the warehouse is rectangular with a longest diagonal distance of 100m. The outbound platform is located in one corner of the warehouse. The height of the bottom shelf is 0.5m, and the height between each shelf is 0.5m. There are three empty storage locations: location A, location B, and location C. The height of location A is 0.5m, the height of location B is 1m, and the height of location C is 1.5m.

[0073] The method for calculating the cost of predictive picking paths is as follows:

[0074] The path planning engine simulates the total path length d that the forklift needs to travel when picking orders containing goods to be put into storage.

[0075] The predictive picking path cost is obtained by calculating the formula C(dis)=(d-dmin) / (dmax-dmin), where dmin and dmax are the shortest and longest possible paths in the simulated scenario, respectively.

[0076] Specifically, in this solution, the shortest path from the forklift to one of the candidate storage locations, as simulated in the scenario, is 100 meters (without being associated with other goods). Correspondingly, the longest path for picking a complete order (associated with other goods) is 500 meters. When the predicted path d = 120 meters for the candidate storage location, C(dis) is 0.05.

[0077] The method for calculating the cost of inter-goods linkage is as follows:

[0078] The cost of the association between goods is obtained by calculating C(ass) = min(Davg / Dmax, 1), where Davg is the average Euclidean distance of all goods in the associated goods set, and Dmax is the maximum diagonal distance of the warehouse. Referring to the above, the average Euclidean distance of the top 3 goods with the highest association is 51.26. When the maximum diagonal distance of the warehouse is 100m, C(ass) is min(0.51, 1), with a value of 0.51.

[0079] The calculation method for the dynamic heat cost of goods is as follows:

[0080] Based on historical inbound and outbound data, the dynamic heat score of goods is calculated and normalized. Referring to the above, assuming the highest dynamic heat score of goods stored in the system is 100, and the dynamic heat score of goods A is 59, then the normalized dynamic heat score of goods is 0.59. The convenience index of candidate storage locations is then calculated. This index is determined by the normalized distance from the storage location to the outbound counter and the normalized height of the available storage location. The calculation formula is Ibin=α×(1-Dbin)+β×(1-|Zbin-Zopt|), where α and β are weighting coefficients, with values ​​of 0.6 and 0.4 respectively; Dbin is the normalized distance from the storage location to the outbound counter; Zbin is the normalized height of the available storage location; and Z... `opt` is the normalized value corresponding to the optimal operating height. Assuming the straight-line distance from the storage location to the outbound platform is 30m, then `Dbin` = 30 / 100 = 0.3. The height of the vacant storage location is 1m, and the total height of the storage rack is 4m, then `Zbin` = 1 / 4 = 0.25. The optimal operating height is 2m, then `Zopt` = 2 / 4 = 0.5. In summary, the convenience index of the candidate storage location for this goods is 0.6 × (1 - 0.3) + 0.4 × (1 - 0.25) = 0.72. The dynamic heat cost of the goods is obtained by calculating the formula `C(heat) = Hgui × Ibin`, where `Hgui` is the normalized score of dynamic heat, and the dynamic heat cost of the goods is 0.59 × 0.72 = 0.42.

[0081] The calculation method for seasonal predictive costs is as follows:

[0082] Obtain the seasonal forecast score for goods;

[0083] Obtain the seasonal handling capacity score for the area where the candidate storage location is located;

[0084] The seasonal predictive cost is obtained by calculating C(sea) = (Cseb-1) × Rbin, where Cseb is the seasonal forecast score of the cargo, and Rbin is the seasonal handling capacity score of the area where the candidate cargo location is located. The area types include core picking area and storage area, and the seasonal handling capacity scores of core picking area and storage area are 1 and 0.3, respectively. Assuming that the seasonal forecast score of the cargo is 1.6, and the candidate cargo location is in the core picking area, then the seasonal predictive cost of the cargo = (1.6-1) × 1 = 0.6.

[0085] The real-time operating cost of the lane is as follows: if the traffic flow of the lane where the cargo location is located is greater than or equal to 3.5 vehicles / min, it is marked as "congested"; if the traffic flow of the lane is less than 3.5 vehicles / min but greater than or equal to 2.5 vehicles / min, it is marked as "moderate"; if the traffic flow of the lane is less than 2.5 vehicles / min, it is marked as "unobstructed". If the lane where the candidate cargo location is located is "congested", the real-time operating cost of the lane is 0.8; if the lane where the candidate cargo location is located is "moderate", the real-time operating cost of the lane is 0.3; if the lane where the candidate cargo location is located is "unobstructed", the real-time operating cost of the lane is 0.1.

[0086] P3: Calculate the total cost of an idle storage space using the total cost function formula. The total cost function formula for an idle storage space is:

[0087] C(shelf)=w1×C(dis)+w2×C(ass)-w3×C(heat)-w4×C(sea)+w5×C(con);

[0088] Among them, w1, w2, w3, w4 and w5 are weighting coefficients with values ​​of 0.3, 0.25, 0.2, 0.15 and 0.1 respectively; C(dis) is the predictive picking path cost; C(ass) is the cost of inter-cargo correlation; C(heat) is the cost of dynamic heat of cargo; C(sea) is the seasonal predictive cost; and C(con) is the real-time operation cost of the alleyway.

[0089] Specifically, the location attributes of storage locations A, B, and C are shown in the table below:

[0090] Table 1:

[0091]

[0092]

[0093] Assume the warehouse is rectangular with a longest diagonal distance of 100m. The outbound platform is located in one corner of the warehouse. The height of the bottom shelf is 0.5m, and the height between each shelf is 0.5m. There are three empty storage locations: location A, location B, and location C. The height of location A is 0.5m, the height of location B is 1m, and the height of location C is 1.5m.

[0094] C(dis) = (d-100) ÷ (500-100), resulting in C(dis)A = 0.05, C(dis)B = 0.75, and C(dis)C = 0.375;

[0095] C(ass)=min(Davg÷100,1), we get C(ass)A=0.15, C(ass)B=0.45, C(ass)C=0.6;

[0096] The three normalized dynamic heat values ​​for goods were all set to 0.59. Through calculation, IbinA = 0.22, IbinB = 0.18, IbinC = 0.5, C(heat) = Hgui × Ibin, and C(heat)A = 0.13, C(heat)B = 0.11, C(heat)C = 0.3.

[0097] The seasonal forecast scores for all three types of goods are 1.6. Calculations show that C(sea)A = 0.6, C(sea)B = 0.18, and C(sea)C = 0.6.

[0098] C(con)A=0.8, C(con)B=0.1, C(con)C=0.3.

[0099] To sum up, C(shelf)A=0.30×0.05+0.25×0.15-0.20×0.13-0.15×0.6+0.10×0.8=0.0165, C(shelf)B=0.3 0×0.75+0.25×0.45-0.20×0.11-0.15×0.18+0.10×0.1=0.2985, C(shelf)C=0.30×0.375+0.25×0.6 -0.20×0.3-0.15×0.6+0.10×0.3=0.1425.

[0100] P4: The vacant storage location with the lowest total cost value is identified as the storage location to be put into storage.

[0101] As discussed above, based on the principle of "minimizing total cost," the system selects cargo location A as the optimal binding target. Although its regional congestion cost is high, its significant advantages in predicted path and correlation, as well as its excellent seasonal matching gain, completely offset the negative impact of congestion. This demonstrates the global trade-off capability of multi-objective optimization algorithms. The reason cargo location B is the worst is that its predicted path cost and correlation cost are too high; even if its congestion level is low, it cannot compensate for these disadvantages in core efficiency indicators.

[0102] S5. Navigation guidance: The backend server sends the optimal storage location coordinates to the PDA that initiated the request. The navigation system on the PDA generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage location coordinates, and displays it on the screen.

[0103] S6. Multimodal detection guidance guides the operator to place the goods to be received into the corresponding storage location on the target shelf using a forklift. The specific method of multimodal detection guidance is as follows:

[0104] The operator drives the forklift along the navigation path to the target rack area. As the forklift approaches the target rack, an RFID reader fixed to the top of the forklift reads the RFID tag fixed to a preset position on the first shelf level using directional identification. After decoding, the planar coordinates (X, Y) of the target location are obtained. A Z-axis measurement module installed on the forklift measures and calculates the height coordinates of the forks or goods relative to the warehouse floor. The Z-axis measurement module employs a dual-redundant fusion scheme of visual measurement and laser ranging. The process includes: acquiring images through a top camera and calculating a first height value based on a visual algorithm; measuring and converting a second height value using a laser ranging sensor at the front of the forks; and finally, performing a confidence-weighted fusion of the two height values ​​to output the final height coordinates. Thus, based on the obtained three-dimensional coordinates... The system compares the information (X, Y, Z) with the target storage location coordinates received by the PDA in real time. When the forklift enters the target shelf within ±0.5m, the system activates a three-level guidance mechanism: First, the target storage location is superimposed on the real-world image with a red highlighted frame using the PDA screen's AR projection function; second, the audio-visual indicator on the forklift is activated, and the driver operates the forklift. When the fork height approaches the target layer within ±0.2m, a green indicator light is triggered, and a voice prompt of "height correct" is issued through bone conduction headphones; finally, after the forks are fully in place, the pressure sensor array installed on the shelf beam verifies the placement status of the goods. When the detected pressure distribution matches the bottom area characteristics of the goods, an entry confirmation signal is automatically sent to the WMS system. The entire process improves positioning accuracy and placement accuracy through multi-sensor fusion verification.

[0105] S7. Bind the acquired cargo information and cargo location coordinates in the system database. After successful binding, the PDA's graphical interface displays the binding success status, and the system broadcasts a confirmation message indicating that the operation is complete.

[0106] In addition, this solution also discloses an intelligent warehouse location binding system, including:

[0107] The vehicle-mounted terminal subsystem is installed in the cab of the forklift;

[0108] The positioning and sensing module includes RFID tags and RFID readers. The RFID tags are installed at preset positions on the first floor of each shelf. Each RFID tag has its own pre-stored coordinates of the shelf. The RFID reader is fixedly installed on the top of the forklift to read the RFID tags and obtain the planar coordinates of the target storage location after decoding.

[0109] The Z-axis measurement module includes a camera and a laser rangefinder. The camera is installed at the top of the forklift facing the rack to acquire images and calculate a first height value based on a visual algorithm. The laser rangefinder is installed at the front of the forks with its probe pointing vertically downward to measure the height between the forks and the ground and calculate a second height value. The Z-axis measurement module is also equipped with a processing unit to perform confidence-weighted fusion of the two height values ​​and output the final height coordinates.

[0110] The navigation subsystem, integrated into the PDA of the vehicle terminal subsystem, generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage space coordinates, and displays it on the screen;

[0111] The back-end server is connected to the vehicle terminal subsystem via the network, receives cargo space allocation requests, and inputs the data into the AI ​​optimization model;

[0112] The AI-powered intelligent decision-making system, comprising a big data management module, a dynamic storage location allocation engine, a path planning and simulation engine, and an adaptive learning engine, is deployed on a central server. Based on a multi-objective optimization algorithm, the system outputs the optimal storage location coordinates for goods awaiting entry into the warehouse and sends these coordinates to the requesting PDA. The big data management module collects, organizes, and analyzes various data from warehouse operations, including goods entry and exit records, storage location usage, and forklift travel trajectories, providing data support for the dynamic storage location allocation engine. The dynamic storage location allocation engine, based on the data provided by the big data management module... The system combines data with factors such as cargo characteristics, order requirements, and warehouse layout, employing a multi-objective optimization algorithm to calculate and output the optimal storage location coordinates for goods to be received in real time. The path planning and simulation engine, based on a 2D electronic map of the warehouse and the optimal storage location coordinates, plans the optimal path from the forklift's current position to the target shelf, and verifies the feasibility and efficiency of the path through simulation technology, providing path data for the navigation subsystem. The adaptive learning engine automatically adjusts and optimizes the parameters and strategies of the multi-objective optimization algorithm based on actual conditions and feedback data during warehouse operations, improving the accuracy and efficiency of the AI ​​intelligent decision-making system.

[0113] The database is used to store cargo information, historical operational data, cargo-related data, real-time order pool data, and the binding relationship between cargo location coordinates and cargo information.

[0114] The voice prompt system, connected to the vehicle terminal subsystem, is used to announce a confirmation message after successful binding.

[0115] The navigation subsystem also includes a multimodal interaction unit, which displays the planned route in a highlighted color on a two-dimensional map after performing route planning, and displays congestion areas predicted based on real-time data in a contrasting color.

[0116] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for intelligent binding of storage locations in an automated warehouse, characterized in that, Includes the following steps: S1. Create a two-dimensional electronic map of the warehouse, establish a two-dimensional coordinate system for the warehouse, and based on the two-dimensional coordinate system and the location of the shelves, establish the coordinates of each shelf with the location of the outbound counter as the origin. S2. Install RFID tags in the preset positions on the first floor of each shelf. Each RFID tag has the coordinates of the shelf where it is located stored in the preset position. S3. Task Triggering and Data Preparation: The operator scans the barcode of the goods to be put into the warehouse using the PDA on the vehicle terminal to obtain the goods information. The PDA then sends the goods information to the backend server via the network, triggering a storage location allocation request. S4. Intelligent warehouse location allocation decision: The back-end server receives the warehouse location allocation request and inputs the data into the AI ​​optimization model. The AI ​​optimization model outputs the optimal warehouse location space coordinates for the goods to be put into the warehouse based on a multi-objective optimization algorithm. The input data of the AI ​​optimization model includes at least: the goods information, historical operation data, historical inbound and outbound data stored in the warehouse for at least 90 days, goods correlation data, and real-time order pool data. S5. Navigation guidance: The backend server sends the optimal storage location coordinates to the PDA that initiated the request. The navigation system on the PDA generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage location coordinates, and displays it on the screen. S6. Multimodal detection and guidance guides the operator to operate the forklift to place the goods to be stored on the corresponding storage location of the target shelf; S7. The acquired cargo information and cargo location coordinates are bound in the system database. After successful binding, the PDA's graphical interface displays the binding success status, and the voice prompt system broadcasts the operation completion confirmation information.

2. The intelligent binding method for storage locations in an automated warehouse according to claim 1, characterized in that, The specific method for multimodal detection guidance in step S6 is as follows: The operator drives the forklift along the navigation path to the target shelving area; When the forklift approaches the target shelf, the RFID reader fixedly installed on the top of the forklift reads the RFID tag fixedly installed at the preset position of the first shelf location in a directional identification manner, and obtains the planar coordinates (X, Y) of the target location after decoding. The Z-axis measurement module installed on the forklift measures and calculates the height coordinates of the forks or goods relative to the warehouse floor.

3. The intelligent binding method for storage locations in an automated warehouse according to claim 2, characterized in that, The Z-axis measurement module adopts a dual-redundancy fusion scheme of visual measurement and laser ranging, the process of which includes: Images are captured by a top camera and a first height value is calculated based on a visual algorithm. A second height value is measured and converted by a laser range sensor at the front of the forks. Finally, the two height values ​​are weighted and fused to output the final height coordinates.

4. The intelligent binding method for storage locations in an automated warehouse according to claim 1, characterized in that, The specific method for outputting the optimal storage location coordinates for the goods to be put into storage based on the multi-objective optimization algorithm in step S4 is as follows: P1: Extract relevant data from the historical data warehouse and calculate the correlation score between goods, the dynamic popularity score of goods, and the seasonal predictive score; P2: Calculate the predictive picking path cost, the cost of inter-cargo correlation, the cost of dynamic cargo heat, the seasonal predictive cost, and the real-time roadway operation cost; P3: Calculate the total cost of an idle storage space using the total cost function formula. The total cost function formula for an idle storage space is: C(shelf)=w1×C(dis)+w2×C(ass)-w3×C(heat)-w4×C(sea)+w5×C(con); Among them, w1, w2, w3, w4 and w5 are weighting coefficients, C(dis) is the predictive picking path cost, C(ass) is the cost of inter-cargo correlation, C(heat) is the cost of dynamic heat of cargo, C(sea) is the seasonal predictive cost, and C(con) is the real-time operation cost of the alley. P4: The vacant storage location with the lowest total cost value is identified as the storage location to be put into storage.

5. The intelligent binding method for storage locations in an automated warehouse according to claim 4, characterized in that, The method for calculating the predictive picking path cost in step P2 is as follows: The path planning engine simulates the total path length d that the forklift needs to travel when picking orders containing the goods to be put into storage. The predictive picking path cost is obtained by calculating the formula C(dis)=(d-dmin) / (dmax-dmin), where dmin and dmax are the shortest and longest possible paths in the simulated scenario, respectively.

6. The intelligent binding method for storage locations in an automated warehouse according to claim 4, characterized in that, The method for calculating the cost of inter-cargo correlation in step P2 is as follows: The top K goods with the highest correlation to the goods to be put into storage, obtained through data mining, are used to form a set of related goods. Calculate the average Euclidean distance between the goods to be received and all goods in the associated goods set when placing the goods to be received in the candidate storage location; The cost of inter-goods association is obtained by calculating C(ass) = min(Davg / Dmax, 1), where Davg is the average Euclidean distance of all goods in the associated goods set, and Dmax is the maximum diagonal distance between warehouses.

7. The intelligent binding method for storage locations in an automated warehouse according to claim 4, characterized in that, The calculation method for the dynamic heat cost of goods in step P2 is as follows: Based on historical inbound and outbound data, calculate the dynamic heat score of goods and normalize it; The convenience index of candidate storage locations is calculated. This index is determined by the normalized distance from the storage location to the outbound station and the normalized value of the height of the available storage location. The calculation formula is Ibin=α×(1-Dbin)+β×(1-|Zbin-Zopt|), where α and β are weighting coefficients, Dbin is the normalized distance from the storage location to the outbound station, Zbin is the normalized height of the available storage location, and Zopt is the normalized value corresponding to the optimal operating height. The dynamic heat cost of goods is obtained by calculating the formula C(heat)=Hgui×Ibin, where Hgui is the normalized score of dynamic heat.

8. The intelligent binding method for storage locations in an automated warehouse according to claim 4, characterized in that, The method for calculating the seasonal predictive cost in step P2 is as follows: Obtain the seasonal forecast score; Obtain the seasonal handling capacity score for the area where the candidate storage location is located; The seasonal predictive cost is obtained by calculating C(sea)=(Cseb-1)×Rbin, where Cseb is the seasonal cargo prediction score and Rbin is the seasonal handling capacity score of the area where the candidate cargo location is located.

9. A smart warehouse location binding system, applied in the smart warehouse location binding method according to any one of claims 1-8, characterized in that, include: The vehicle-mounted terminal subsystem is installed in the cab of the forklift; The positioning and sensing module includes RFID tags and RFID readers. The RFID tags are installed at preset positions on the first floor of each shelf. Each RFID tag has its own pre-stored coordinates of the shelf. The RFID reader is fixedly installed on the top of the forklift to read the RFID tags and obtain the planar coordinates of the storage location after decoding. The Z-axis measurement module includes a camera and a laser rangefinder. The camera is installed at the top of the forklift facing the rack to acquire images and calculate a first height value based on a visual algorithm. The laser rangefinder is installed at the front of the forks with its probe pointing vertically downward to measure the height between the forks and the ground and calculate a second height value. The Z-axis measurement module is also equipped with a processing unit to perform confidence-weighted fusion of the two height values ​​and output the final height coordinates. The navigation subsystem, integrated into the PDA of the vehicle terminal subsystem, generates a visual navigation path from the current position of the forklift to the target shelf based on the two-dimensional electronic map of the warehouse and the optimal storage space coordinates, and displays it on the screen; The back-end server is connected to the vehicle terminal subsystem via the network, receives cargo space allocation requests, and inputs the data into the AI ​​optimization model; The AI ​​intelligent decision-making system includes a big data management module, a dynamic storage location allocation engine, a path planning and simulation engine, and an adaptive learning engine. It is deployed on a central server. The AI ​​intelligent decision-making system outputs the optimal storage location coordinates for goods to be put into storage based on a multi-objective optimization algorithm, and sends the optimal storage location coordinates to the PDA that initiated the request. The database, deployed on a central server, is used to store cargo information, historical operational data, cargo-related data, real-time order pool data, and the binding relationship between cargo location coordinates and cargo information. The voice prompt system, connected to the vehicle terminal subsystem, is used to announce a confirmation message after successful binding.

10. The intelligent location binding system for an automated warehouse according to claim 9, characterized in that, The navigation subsystem also includes a multimodal interaction unit, which, after performing route planning, displays the planned route in a highlighted color on a two-dimensional map and displays congestion areas predicted based on real-time data in a contrasting color.

Images