Container type intelligent vertical warehouse control system and method

Abstract

The invention provides a container type intelligent three-dimensional warehouse control system and method, and the method comprises the steps: predicting the influence effects of different storage positions on the overall weight distribution of a goods shelf through a simulation stress distribution model for a goods combination set, and obtaining an optimized storage position scheme; according to the storage position scheme and the current equipment position data, generating an equipment operation path by adopting a path optimization method, and determining a path sequence for minimizing high-load area operation; updating a scheduling instruction of a control system through the path sequence, obtaining executed new stress distribution data, and judging whether a stress balance state of the goods shelf is reached or not; and issuing an instruction to automation equipment in real time according to the final adjustment scheme, obtaining execution feedback data, and determining the stability state of the shelf system.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a containerized intelligent automated warehouse control system and method. Background Technology

[0002] In the field of modern logistics and warehousing, containerized intelligent automated storage and retrieval systems (AS / RS) play an indispensable role as a crucial technological means to improve storage efficiency and space utilization. Their core lies in achieving efficient storage and retrieval of goods and space optimization through intelligent management, which is key to ensuring the smooth operation of the supply chain. However, with the diversification of goods types and weights, the operational stability and security of AS / RS systems face increasing challenges, making research into related technologies a focus of industry attention.

[0003] Currently, although many warehousing systems have introduced automated equipment and basic control mechanisms, they still have significant shortcomings when dealing with complex storage environments. Especially when faced with uneven weight distribution of goods and long-term load variations on racks, existing methods often lack the ability to perceive the structural stress state in real time and are unable to flexibly adjust storage strategies according to actual conditions. This deficiency makes the system prone to localized overload or structural fatigue during operation, thereby affecting overall safety and service life.

[0004] Focusing on the technical challenges, the core challenge of a containerized intelligent automated storage and retrieval system (AS / RS) lies in the dynamic balance of load distribution on the racks. As the core supporting structure of the storage system, the stress state of the racks directly affects the stability of the entire AS / RS. Due to the significant weight differences among various goods, the pressure distribution at different locations on the racks is often extremely uneven. This imbalance intensifies with frequent handling of goods, leading to stress concentration in localized areas and potentially causing structural deformation or damage. More complexly, this uneven stress distribution also affects the selection of operating paths for automated equipment, causing it to repeatedly operate in high-load areas, further increasing the structural burden.

[0005] Therefore, how to monitor the stress state of each location on the rack in real time during the dynamic process of goods storage and retrieval, and rationally plan the storage location of goods and the operation path of equipment according to the weight distribution, has become a key issue to ensure the long-term stable operation of the automated warehouse system. This issue is particularly prominent in actual business operations. For example, when storing heavy goods, if the heavy items are not arranged in time in the more stable lower area, but are placed in the upper storage position, it may lead to an imbalance of the overall stress on the rack, and even cause safety hazards.

[0006] These interconnected issues, ranging from the imbalance of forces on the racks to the need for optimized equipment operation paths, constitute the technical bottlenecks that the containerized intelligent automated storage and retrieval system urgently needs to overcome, and also point the way for subsequent research. Summary of the Invention

[0007] This invention provides a containerized intelligent automated storage and retrieval system control system and method, mainly comprising: By collecting pressure distribution data at each storage location through a sensor network deployed at key locations on the shelf, the real-time force distribution matrix of the shelf is obtained, and the overall weight stress state of the shelf is determined. Based on the stress distribution matrix, calculate the pressure difference and average load level between each cargo location, use an optimization calculation method to identify areas of uneven stress, and determine the location and extent of the local overload area. If the pressure difference in the local overload area exceeds a preset threshold, the weight attribute data of the goods to be stored is extracted from the cargo weight database to determine the appropriate cargo combination set for adjustment. For the aforementioned cargo combination set, the impact of different storage locations on the overall weight distribution of the shelf is predicted by simulating a force distribution model, thereby obtaining an optimized storage location scheme; Based on the storage location scheme and the current equipment location data, a path optimization method is used to generate equipment operation paths and determine the path sequence that minimizes operation in high-load areas; The scheduling instructions of the path sequence update control system are used to obtain the new force distribution data after execution, and to determine whether the force balance state of the shelf has been reached. If the force equilibrium state is not reached, the remaining unevenness index is extracted from the new force distribution data, and the storage location and path parameters are further fine-tuned using parameter adjustment methods to obtain the final adjustment scheme. Based on the final adjustment plan, instructions are sent to the automated equipment in real time to obtain execution feedback data and determine the stability status of the racking system.

[0008] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a real-time optimization and adjustment method for shelf weight distribution based on sensor networks. By deploying sensors at key locations on the shelf to collect pressure distribution data, a real-time force distribution matrix is ​​formed, accurately identifying the overall weight status and local overload areas. When uneven force distribution is detected where the pressure difference exceeds a threshold, this invention extracts attribute data from a cargo weight database and combines it with a simulated force distribution model to predict the impact of different storage locations, quickly generating an optimized cargo storage plan. Subsequently, path optimization generates equipment operation paths that minimize high-load areas, and real-time scheduling instructions are issued to automated equipment for adjustment. Based on the new force data from the execution feedback, a balance state is determined. If balance is not achieved, the position and path parameters are further fine-tuned until shelf force equilibrium is achieved. The core of this invention lies in the organic integration of closed-loop sensing, predictive optimization, and dynamic adjustment, effectively avoiding shelf deformation or safety hazards caused by local overload, and improving the stability and service life of the warehousing system. Attached Figure Description

[0009] Figure 1 This is a flowchart of a containerized intelligent automated storage and retrieval system and method according to the present invention.

[0010] Figure 2 This is a schematic diagram of a containerized intelligent automated storage and retrieval system and method according to the present invention.

[0011] Figure 3 This is another schematic diagram of a containerized intelligent automated storage and retrieval system and method according to the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0013] like Figures 1-3 This embodiment of a containerized intelligent automated storage and retrieval system control system and method may specifically include: Step S101: Collect pressure distribution data at each storage location by deploying a sensor network at key locations on the shelf, obtain the real-time force distribution matrix of the shelf, and determine the overall weight stress state of the shelf.

[0014] A sensor network continuously collects pressure distribution data for each storage location at key positions on the shelving system, resulting in a raw pressure data set. This raw pressure data set is then synchronized and aligned in time to obtain a storage location pressure sequence with a unified timestamp. A sliding window method is used to extract local pressure change values ​​from the storage location pressure sequence, yielding a pressure change matrix for each storage location. This storage location pressure change matrix is ​​mapped to a preset shelving structural position to generate a real-time shelving stress distribution matrix. The total pressure in each region is calculated based on the real-time shelving stress distribution matrix to obtain the overall shelving weight distribution value. If the overall weight distribution value exceeds a preset equilibrium range, the shelving is determined to be under unbalanced load, and the location of the unbalanced load area is identified. Based on the location of the unbalanced load area and the stress distribution matrix, a set of abnormal stress points on the shelving structure is determined.

[0015] The sensor network continuously collects pressure distribution data for each storage location at key locations on the shelf.

[0016] Understandably, this method utilizes multiple pressure sensors placed on load-bearing parts of the shelving, such as beams, uprights, and shelves, to form a dense monitoring network.

[0017] For example, on a standard four-layer shelf, 16 sensor points are set on each layer to collect pressure values ​​every second in real time, forming a raw pressure data set. Because the sensor sampling frequencies are slightly different, these raw data need to be synchronized to obtain a shelf location pressure sequence with a unified timestamp.

[0018] Specifically, nearest neighbor interpolation or linear interpolation methods are used to align all sensor data onto the same time axis, avoiding analysis errors caused by time deviations.

[0019] In one possible implementation, a sliding window method is used to extract local pressure change values ​​from the cargo location pressure sequence. For example, the window size is set to 10 seconds, with each step sliding for 2 seconds. The standard deviation or difference value of the pressure values ​​within the window is calculated, thus obtaining the pressure change matrix for each cargo location. This matrix reflects the dynamic load fluctuations in the short term, helping to capture instantaneous changes caused by cargo placement or removal.

[0020] For example, by mapping the pressure change matrix of the storage location to the preset shelf structure position, a real-time stress distribution matrix of the shelf can be generated.

[0021] Specifically, a coordinate mapping table for the shelving is pre-established, mapping each sensor to its three-dimensional position on the shelving. Pressure change values ​​are then filled into the corresponding grid, creating a visualized stress-heat map. This distribution matrix visually shows which areas of the shelving experience concentrated stress, effectively improving monitoring accuracy. The total pressure in each area is calculated based on the real-time stress distribution matrix of the shelving, yielding the overall weight distribution value of the shelving.

[0022] For example, the shelf can be divided into four quadrants: left front, left rear, right front, and right rear. The pressure values ​​are summed for each quadrant. If the total pressure on the left side is 1500 kg and on the right side is 900 kg, exceeding the preset balance range (if the difference between the left and right sides does not exceed 10%), then the shelf is considered to be under unbalanced load, and the unbalanced area is located on the left side. This judgment can promptly detect the risks caused by unbalanced placement and prevent the shelf from tilting.

[0023] In one embodiment, the set of abnormal stress points in the rack structure is determined based on the location of the off-center load area and the stress distribution matrix.

[0024] For example, when the load is unevenly distributed on the left side, further analysis of the matrix reveals points in the matrix where the pressure value exceeds a single-point threshold, such as 300 kg, and these are marked as anomaly sets. These anomalies are often located at beam connections or the bottom of columns. Early identification can prevent structural fatigue accumulation, extend the service life of the shelving, and reduce the risk of collapse.

[0025] It should be noted that the entire process achieves a closed loop from initial data collection to anomaly identification through continuous monitoring and data processing. The beneficial effects are real-time early warning of off-center loading risks, support warehouse managers in quickly adjusting the layout of goods, ensure safe storage, and optimize space utilization.

[0026] Step S102: Calculate the pressure difference and average load level between each cargo location based on the force distribution matrix, identify areas of uneven force distribution using an optimized calculation method, and determine the location and extent of the local overload area.

[0027] By retrieving the stress distribution matrix data from the storage system, each cargo location is scanned one by one to calculate the pressure difference between each location and its adjacent locations, yielding a preliminary pressure difference distribution result. Based on the pressure difference distribution result, the average load level of all cargo locations is calculated. Statistical methods are used to compare the load value of each cargo location with the average load level to identify cargo locations with significant load deviations. For cargo locations with significant load deviations, the stress data of surrounding cargo locations is acquired. Comparative analysis is used to identify uneven stress distribution and potential local overload areas. If a local overload area is identified, the specific stress values ​​of each cargo location within that area are extracted. Combined with the area's location information, a quantitative value of the overload degree is calculated to obtain the distribution of the overload degree. Based on the distribution of the overload degree, a preset threshold is used for comparison. If the overload degree of a certain area exceeds the threshold, that area is marked as a high-risk overload area, and its specific location is determined. By recording and classifying the specific location data of high-risk overload areas, corresponding area identification information is generated to obtain the final overload area location result.

[0028] For example, in the field of warehouse management, stress distribution analysis of racks is a crucial step, especially in ensuring the safety of stored goods and the stability of rack structures. The following will provide detailed examples and analyses on topics such as obtaining stress distribution matrix data, calculating pressure differences, identifying load deviations, determining local overload areas, quantifying overload levels, and marking high-risk areas, aiming to construct a logically rigorous explanation from multiple perspectives.

[0029] For example, to obtain force distribution matrix data from a storage system, imagine a large warehouse where multiple pressure sensors are installed on the shelves, and data from each location is uploaded to a central system in real time. Assuming a shelf has 5 layers with 10 locations per layer, the system would generate a matrix containing 50 data points, each representing the pressure value of a location in kilograms. This method comprehensively reflects the stress state of the shelf, providing fundamental data for subsequent analysis.

[0030] For example, when calculating the pressure difference between each storage location and its adjacent locations, the pressure differences between adjacent locations can be compared by scanning them one by one. Suppose the pressure value of a certain storage location is 500 kg, while the pressure of the storage location to its left is 300 kg and to its right is 400 kg. Then the pressure difference between the left and right sides of this storage location is 200 kg and 100 kg, respectively. This comparison can provide a preliminary assessment of whether there are any abnormal pressures on the storage location, offering clues for further analysis.

[0031] For example, to identify storage locations with significant load deviations, the average load level of all storage locations can be calculated first. Suppose the average pressure across 50 storage locations is 350 kg, and a particular location has a pressure of 600 kg, significantly higher than the average, then this location can be marked as a deviation point. This method helps to quickly locate areas where problems may exist.

[0032] For example, when identifying areas of localized overload, the stress conditions of surrounding storage locations can be analyzed. Suppose there are three storage locations surrounding the 600 kg location, with pressure values ​​of 550 kg, 500 kg, and 520 kg respectively, all higher than the average. This suggests a potential localized overload in that area. This analytical approach can effectively identify potential risks.

[0033] For example, to quantify the degree of overload, the specific stress values ​​of each cargo location within a localized overload area can be extracted and evaluated in conjunction with the area's location information. Assuming the average pressure of the four cargo locations in this area is 542.5 kg, exceeding the average by 192.5 kg, the degree of overload can be defined as relatively high. This quantification method provides an intuitive basis for subsequent processing.

[0034] For example, when marking high-risk overload areas, a threshold can be set, such as 1.5 times the average load, or 525 kg. If the average pressure in a certain area exceeds this value, it is marked as a high-risk area, and its specific location is recorded, such as the area on the left side of the third shelf. This marking method helps to quickly locate problem areas and take measures. Through the above multi-faceted analysis and examples, it can be seen that the processing of each step is closely centered on the core theme of shelf stress distribution, progressing step by step to ensure the comprehensiveness and logic of the analysis, while providing reliable data support for warehouse management, and helping to improve the safety and stability of shelf use.

[0035] Step S103: If the pressure difference in the local overload area exceeds a preset threshold, the weight attribute data of the goods to be stored is extracted from the goods weight database to determine a suitable set of goods combinations for adjustment.

[0036] Obtain the pressure difference value of the local overload area. If the pressure difference value exceeds a preset threshold, extract the weight attribute data of the goods to be stored from the cargo weight database. Based on the extracted weight attribute data, determine the weight sorting sequence of the goods to be stored. Use the weight sorting sequence to determine the matching degree between the weight attribute data and the local overload area. If the matching degree is lower than a preset level, use the K-means clustering algorithm to group the weight attribute data to obtain a cargo combination set. Based on the cargo combination set, determine the suitable cargo combination subset for adjustment. Obtain the adjusted weight distribution data using the cargo combination subset.

[0037] In the field of warehouse cargo storage optimization, after the system identifies a local overloaded area, it is necessary to further obtain the pressure difference between each storage location within that area to provide a basis for subsequent cargo adjustments. For example...

[0038] In one embodiment, the pressure values ​​of all adjacent cargo location pairs in the local overload area are first extracted from the force distribution matrix, and the difference between each pair of cargo locations is calculated to form a pressure difference dataset.

[0039] Specifically, assuming a localized overload area contains six storage locations with pressure values ​​of 450 kg, 520 kg, 480 kg, 610 kg, 470 kg, and 550 kg respectively, the pressure difference between adjacent storage locations could reach 70 kg, 130 kg, 140 kg, etc. If these differences exceed a preset threshold, such as 100 kg, a subsequent cargo adjustment process is triggered. This threshold setting effectively filters out minor fluctuations and avoids unnecessary adjustment operations.

[0040] It should be noted that once the pressure difference exceeds the threshold, the system immediately retrieves the weight attribute data of the goods to be stored from the cargo weight database. This data typically includes key information such as cargo number, weight, and dimensions.

[0041] For example, the weight of goods to be stored in the database might be 300kg, 420kg, 580kg, 350kg, 510kg, etc. By extracting this data, it can be ensured that the adjustment plan is based on actual available goods, avoiding unfounded assumptions.

[0042] In one possible implementation, based on the extracted weight attribute data, the system generates a weight sorting sequence of the goods to be stored, such as 300kg, 350kg, 420kg, 510kg, and 580kg arranged from lightest to heaviest. This sorting helps to quickly assess the suitability of the goods for overloaded areas, prioritizing a mix of light and heavy items to balance the load.

[0043] For example, when determining the matching degree between weight attribute data and local overload areas, the deviation ratio between the current average load of the overload area and the weight of the candidate goods can be calculated. If the average load of the overload area is 513 kg, while the average weight of the goods to be stored is 432 kg, the matching degree may only be 70%, which is lower than the preset level of 85%. This indicates that direct placement may exacerbate the unevenness and further optimization is needed.

[0044] Specifically, when the matching degree is lower than the preset level, the K-means clustering algorithm is used to group the weight attribute data.

[0045] For example, the weights of the goods can be clustered into two groups: a light group of 300kg, 350kg, and 420kg, and a heavy group of 510kg and 580kg. This grouping naturally creates combinations of similar weights, reducing sudden pressure changes caused by cross-group placement and effectively protecting the stability of the shelving structure.

[0046] In one embodiment, a subset of goods combinations suitable for adjustment is further selected based on the set of goods combinations obtained from clustering.

[0047] For example, a subset of light-duty cargo can be prioritized for placement around overloaded areas to reduce localized pressure peaks; or a mixed subset of light and heavy-duty cargo can be selected to fill underloaded cargo locations to achieve overall balance. This subset determination process can significantly improve adjustment efficiency.

[0048] For example, by selecting a subset of goods, adjusted weight distribution data can be obtained, such as adjusting the pressure values ​​in overloaded areas to 480kg, 490kg, 500kg, 510kg, 490kg, and 500kg, with the difference controlled within 20kg. This distribution data not only intuitively reflects the improvement effect but also provides a benchmark for subsequent monitoring, helping to prevent potential risks in a timely manner, extend the service life of shelving, and improve the level of warehouse safety.

[0049] Step S104: For the set of goods, the effect of different storage locations on the overall weight distribution of the shelf is predicted by simulating the force distribution model, and an optimized storage location scheme is obtained.

[0050] Obtain the weight and volume data of each item in the goods combination set. Calculate the local stress values ​​of the items when placed in different storage positions on the shelf using a stress distribution model. Simulate and predict the overall shelf weight distribution state corresponding to each storage position. Determine whether the weight distribution state exceeds the preset shelf structure stress threshold; if it does, mark the storage position as unusable. For unusable storage positions, exclude the corresponding position schemes from the goods combination set. Use a genetic algorithm to iteratively optimize the remaining storage position schemes to obtain the position combination with the most balanced weight distribution. Verify the uniformity of the overall shelf stress distribution under the optimized storage position scheme using a finite element analysis model, and determine the final optimized storage position scheme.

[0051] For example, in the business scenario of goods storage management, the weight and volume data of a collection of goods can be obtained directly through sensors and databases in the warehouse management system.

[0052] For example, a cargo combination set contains 10 goods, each weighing between 5 and 50 kilograms, with volume data recorded in cubic meters. This data provides the basis for subsequent analysis.

[0053] In one possible implementation, when calculating the local stress values ​​of goods at different locations on the shelf using a stress distribution model, it can be assumed that the shelf is divided into multiple zones, each with a different load-bearing capacity. For example, if the upper left corner of a shelf has a maximum load-bearing capacity of 200 kg, placing a 180 kg item there would result in a local stress value close to the maximum, requiring further analysis to determine if it affects overall stability.

[0054] For example, when simulating and predicting the overall weight distribution of a shelf, visualization tools can be used to present the force distribution under different placement schemes. Suppose that heavy goods are concentrated at the bottom of the shelf and light goods are placed on the upper shelf. The simulation results may show that the bottom area is under concentrated force, while the upper layer is under less force. This distribution needs to be further determined to see if it exceeds a preset threshold.

[0055] In one possible implementation, to determine whether the weight distribution exceeds the structural stress threshold of the shelving, the threshold can be set to 80% of the shelving's design load-bearing capacity. If the simulation results show that the stress in a certain area reaches 90% of the design load-bearing capacity, then that location is marked as unusable. This approach helps to identify potential risk areas in advance.

[0056] For example, to exclude unusable storage locations, the corresponding solutions can be removed from the cargo combination set. Suppose a location is marked as unusable due to excessive stress, the system will automatically remove 3 solutions involving that location from 10 candidate solutions, and the remaining solutions will proceed to the next round of optimization.

[0057] In one possible implementation, when using a genetic algorithm to iteratively optimize the remaining storage location schemes, each scheme can be treated as an individual, and the combination with the most balanced weight distribution can be selected through multiple iterations. Assuming there are 7 initial schemes, after multiple rounds of optimization, a scheme with heavy goods at the bottom and light goods at the top is obtained, ensuring a more reasonable overall force distribution.

[0058] For example, when verifying the uniformity of the overall stress distribution on the racking system using a finite element analysis model, the deformation of the racking system under different loads can be simulated. If the final solution shows that the stress difference between different areas is controlled within 10%, then the solution can be confirmed as the final optimized storage location solution. This verification method can effectively improve the reliability of the storage solution.

[0059] In one possible implementation, the above method can also be combined with the pressure difference analysis of local overload areas mentioned in the historical dialogue to ensure that the cargo storage plan matches the actual load-bearing capacity of the rack.

[0060] For example, when adjusting cargo combinations, priority should be given to the impact of weight distribution on local areas to avoid overall structural risks caused by overloading in a single area. This comprehensive consideration helps improve the safety and efficiency of warehouse storage.

[0061] Step S105: Based on the storage location scheme and the current equipment location data, a path optimization method is used to generate the equipment operation path and determine the path sequence that minimizes the operation in high-load areas.

[0062] Step 1: Acquire equipment location data and a preset storage scheme. The data is then structured using a data parsing module to obtain the equipment's location distribution information in the current environment. Step 2: Based on the location distribution information, a preliminary work path sequence is generated using preset path planning rules and high-load area identification data. The operational nodes involved in the path are then identified. Step 3: The preliminary work path sequence is optimized using Dijkstra's algorithm to calculate a shorter path that avoids high-load areas, resulting in an optimized work path. Step 4: The optimized work path is used to extract the operational sequence. If high-load area nodes are still present in the operational sequence, the priority of the path nodes is readjusted to determine the final operational sequence. Step 5: Based on the final operational sequence, a detailed time schedule for equipment operations is generated. The operation time for high-load areas is compressed to obtain a time-reduced work plan. Step 6: The work plan is used to generate execution instructions for the equipment's operational path. Considering the constraints of area operations, if a temporary high-load area is encountered during path execution, a backup path is dynamically switched to determine the final execution path.

[0063] For example, in warehouse automation scenarios, equipment location data includes the real-time coordinates of AGVs and the fixed location points of shelves. The data parsing module converts this raw coordinate information into a structured grid distribution map, which facilitates subsequent path calculation.

[0064] Specifically, the preset storage scheme may define multiple pick-up and drop-off points. After parsing, the location distribution information of the equipment in the current environment can be obtained, such as the closest distance between the AGV and the high-load area being 2.5 meters.

[0065] In one possible implementation, preset path planning rules are used based on location distribution information, such as prioritizing paths with shorter straight-line distances. This is combined with high-load area identification data, such as marking frequently used forklift lanes as high-load areas; this data comes from historical traffic statistics. When generating the initial operation path sequence, the operational nodes involved in the path are determined. For example, from the starting point to pickup point A and then to drop-off point B, the sequence contains 5 nodes, avoiding direct passage through high-load areas.

[0066] For example, Dijkstra's algorithm is used to optimize the initial work path sequence. This algorithm finds the path with the minimum cost by expanding point by point. Here, the passage cost of nodes in high-load areas is set to three times that of normal areas, thus calculating a shorter path that avoids high-load areas. After optimization, the path length may be reduced from the original 15 meters to 12 meters, resulting in a more efficient work path.

[0067] In one possible implementation, the operation sequence is extracted through the optimized job path, such as visiting nodes 1, 3, and 5 sequentially. If there are still high-load nodes in the sequence, the priority of the path nodes is readjusted, for example, increasing the weight of low-load nodes from 0.5 to 0.8, and finally determining the operation sequence without high-load interference. This helps reduce equipment waiting time and improves the overall smoothness of the operation.

[0068] Specifically, a detailed time schedule for equipment operations is generated based on the final operation sequence. For example, the dwell time at each node is preset to 30 seconds, and the original sequence of 6 nodes would take 3 minutes. However, the operation time in high-load areas is compressed, such as by compressing the time of relevant nodes by 20%, resulting in a reduced operation plan with a total duration of 2.4 minutes. This compression effectively reduces the risk of congestion during peak periods and improves warehouse throughput efficiency.

[0069] For example, the system generates execution instructions for equipment operation paths through work plans, such as issuing specific coordinate sequences and speed commands to AGVs, combined with area operation constraints, such as a 20% speed limit in high-load areas. If a temporary high-load area is encountered during path execution, such as a sudden forklift occupation, a backup path is dynamically switched. This backup path is pre-calculated and stored, and its length only increases by 1 meter after switching, ensuring operational continuity. This dynamic mechanism significantly improves system robustness, reduces operational interruptions caused by temporary disturbances, and thus optimizes the overall warehouse operational efficiency.

[0070] Step S106: Obtain the new force distribution data after execution by updating the scheduling instructions of the path sequence update control system, and determine whether the force balance state of the shelf has been reached.

[0071] The path sequence generation module acquires the latest path sequence data to determine the basis for updating scheduling instructions. Based on the path sequence data, the control system's built-in logic processing unit generates corresponding scheduling instructions, obtaining a preliminary execution plan. The instruction execution module sends the generated scheduling instructions to relevant equipment on the rack structure, collects post-execution force distribution data, and determines whether it meets preset balance conditions. If the force distribution data exceeds the preset balance threshold, the system feedback mechanism obtains deviation information to determine the path sequence portion requiring adjustment. Based on the deviation information, a support vector machine algorithm is used to optimize the path sequence, obtaining adjusted path sequence data. The adjusted path sequence data updates the control system's scheduling instructions, acquires new force distribution data, and determines whether the rack structure has reached a balance state. If the new force distribution data still does not reach a balance state, the distribution analysis module extracts key features of the force distribution to determine the optimization direction for subsequent path sequences.

[0072] For example, in an intelligent scheduling system for warehouse racking structures, the latest path sequence data can be obtained through the path sequence generation module, which can capture the changes in racking position after equipment movement in real time, thereby providing an accurate basis for updating scheduling instructions.

[0073] Specifically, after an automated guided vehicle completes a goods handling operation, the path sequence data records the new coordinate distribution of the shelves, which directly determines the accuracy of subsequent scheduling instructions.

[0074] In one embodiment, the corresponding scheduling instructions are generated by the logic processing unit built into the control system based on the path sequence data, which can quickly form a preliminary plan.

[0075] For example, if the current path sequence shows that multiple shelves need to be adjusted at the same time, the logic processing unit will first calculate the movement order with the least overlap, generate instructions to move shelf A to the target point in sequence, and then process shelf B, thus avoiding conflicts caused by simultaneous operation and thus initially achieving efficient scheduling.

[0076] For example, after the scheduling instructions are sent to the relevant equipment of the rack structure through the instruction execution module, the collected force distribution data after execution becomes a key basis for judgment.

[0077] Understandably, when the equipment moves the shelf, it will collect the force value of each support point in real time through pressure sensors installed on the base. If the data shows that the force on one side reaches 1500 Newtons while the force on the other side is only 800 Newtons, it is determined that it exceeds the preset balance threshold of 1200 Newtons. This helps to detect potential tilting risks in a timely manner.

[0078] In one possible implementation, if the force distribution exceeds a threshold, deviation information is obtained through a system feedback mechanism, which can accurately locate the part of the path sequence that needs to be adjusted.

[0079] For example, if the deviation information shows that the left side of the shelf is under too much force, the feedback mechanism will mark the sequence segment in the corresponding path that involves left-side movement as the adjustment target. This provides a clear direction for subsequent optimization and effectively reduces the extra time consumption caused by blind adjustments.

[0080] For example, when using the support vector machine algorithm to optimize a path sequence, the algorithm learns from historical force deviation samples and constructs a classification hyperplane to predict the effect of the adjusted path.

[0081] Specifically, the algorithm takes the previous path sequences and corresponding force results as input into the training set, outputs optimized weights, and prioritizes the movement order that can balance the left and right forces in the new path, thus obtaining the adjusted path sequence data, which significantly improves the balance convergence speed.

[0082] In one embodiment, after updating the scheduling instructions with the adjusted path sequence data, new force distribution data is obtained again. If the equilibrium state is still not reached, the distribution analysis module is entered to extract key features.

[0083] For example, if the analysis shows that the peak force is concentrated at the front of the shelf, the module will extract the peak position and amplitude as features to determine the subsequent optimization direction as prioritizing the timing of the movement of the front shelf. This forms a closed-loop iteration to ensure that the final force is evenly distributed at around 1000 Newtons.

[0084] Understandably, the entire process emphasizes beneficial effects. For example, through real-time feedback and machine learning optimization, the shelving structure can quickly reach a balanced state, avoiding structural fatigue or safety hazards caused by uneven stress. It can also reduce equipment idle waiting time and improve the overall operational efficiency and stability of the warehouse.

[0085] Step S107: If the force balance state is not achieved, the remaining unevenness index is extracted from the new force distribution data, and the storage location and path parameters are further fine-tuned using the parameter adjustment method to obtain the final adjustment scheme.

[0086] Residual non-uniformity indices are extracted from the new force distribution data to obtain the location coordinates of multiple local non-uniformity regions. Based on the location coordinates of these regions, the offset of the storage location parameter corresponding to each region is determined. A gradient descent algorithm is used to calculate the joint adjustment direction of the storage location parameter offset and the path parameter, resulting in a preliminary parameter correction vector. The storage location parameter and path parameter are updated using this preliminary parameter correction vector to obtain the updated force distribution data. Residual non-uniformity indices are extracted from the updated force distribution data, and it is determined whether all remaining non-uniformity indices are below a preset threshold. If all remaining non-uniformity indices are below the preset threshold, the current storage location parameter and path parameter are output as the final adjustment scheme. If any remaining non-uniformity indices are still above the preset threshold, the fine-tuning range of the storage location parameter and path parameter is re-determined based on the location coordinates of the remaining non-uniformity indices above the threshold, resulting in the final adjustment scheme.

[0087] In one possible implementation, residual unevenness indicators are extracted from the new force distribution data. This is mainly achieved by using sensors to collect the pressure values ​​of each support point of the shelf in real time, calculating the percentage deviation of each area from the ideal uniform force, and thus obtaining multiple local unevenness indicators.

[0088] For example, assuming the total load capacity of the shelf is 1000kg, and each support point ideally bears 250kg, if a certain area actually bears 280kg, then the remaining unevenness index is 12%.

[0089] Specifically, these indicators can accurately identify localized areas of overload or underload, avoiding unnecessary disturbances caused by overall adjustments. For example...

[0090] In one embodiment, the offset of the storage location parameter corresponding to each region is determined based on the location coordinates of multiple local uneven regions.

[0091] Understandably, the storage location parameters include the horizontal and vertical coordinate offsets of the goods on the shelf. The system will map uneven areas to the nearest storage point and calculate the offset to redistribute the weight.

[0092] For example, if the unevenness index is high in the left area, the corresponding goods are shifted 10cm to the right, thus guiding the weight to transfer to the right. This method helps to quickly locate and adjust the object, improving balancing efficiency.

[0093] Preferably, a gradient descent algorithm is used to calculate the joint adjustment direction of the storage location parameter offset and the path parameter to obtain a preliminary parameter correction vector. This algorithm iteratively evaluates the gradient of the effect of parameter changes on the overall force uniformity to find the fastest descent direction.

[0094] For example, the initial offset is set to 5cm, and the path parameters are adjusted to shorten the pickup path by 2 meters. After several iterations, the correction vector is obtained, such as shifting the storage location to the right by 8cm and reducing the path detour by 1.5 meters, thereby initially reducing the unevenness index to 60% of the original value. This joint optimization is beneficial to simultaneously take into account force balance and path efficiency, and avoids secondary imbalances caused by single adjustments.

[0095] In one possible implementation, after updating the stored position parameters and path parameters through the initial parameter correction vector, the updated force distribution data is obtained, and the remaining non-uniformity index is extracted from it.

[0096] For example, after the update, the unevenness index on the left side decreased from 12% to 4%, and on the right side from 8% to 3%, resulting in a more uniform overall performance. This iterative update gradually approaches the ideal state, significantly improving shelf stability.

[0097] For example, it determines whether all remaining unevenness indicators are below a preset threshold, such as 3%. If all are below, the current parameters are directly output as the final solution to ensure the shelving quickly reaches equilibrium and reduces the risk of structural fatigue. If there are still indicators above the threshold, such as a corner point still reaching 5%, the fine-tuning range is re-determined based on its position coordinates, such as a further offset of 3cm or a fine-tuning path of 0.5 meters, to obtain the final adjustment solution. This conditional judgment mechanism is beneficial for achieving precise convergence and avoiding increased energy consumption caused by over-adjustment.

[0098] Specifically.

[0099] In one embodiment, the above process can be repeated until a threshold is met, forming a closed-loop optimization. Through multiple rounds of fine-tuning, not only can the overall unevenness index be controlled within 2%, but the rationality of path parameters can also be maintained, ensuring the efficient operation of the warehousing system and the long-term safety of the shelves.

[0100] Step S108: Based on the final adjustment plan, send instructions to the automated equipment in real time, obtain execution feedback data, and determine the stability status of the racking system.

[0101] The final adjustment plan generates and issues instructions. These instructions are then sent to the automated equipment in real time. Execution feedback data is acquired. The execution status of the instructions is determined based on the feedback data; if the feedback data matches the issued instructions, a normal execution result is obtained; otherwise, an abnormal execution result is obtained. Deviation data is extracted from the abnormal execution results. A time series analysis algorithm is used to analyze the trend of the deviation data, obtaining the stability change trend. The stability state of the racking system is determined based on the stability change trend; if the trend exceeds a preset threshold, an unstable state is obtained; otherwise, a stable state is obtained.

[0102] For example, after the optimization and adjustment of the shelving system is completed, the process of generating and issuing instructions based on the final adjustment plan can be understood as converting parameters into control signals that the equipment can execute.

[0103] Specifically, the final adjustment scheme includes optimized storage location coordinates and path parameters, which are encapsulated into a standardized instruction format, such as a position offset of 0.05 meters in the x-direction and 0.03 meters in the y-direction, as well as path curve coefficient adjustment values.

[0104] In one embodiment, the system sends instructions to automated equipment in real time, such as via a wireless network to an AGV (Automated Guided Vehicle) robot or robotic arm controller. Upon receiving the instructions, the equipment immediately performs cargo repositioning or path-following operations, thereby making the force distribution on the shelf more uniform. This real-time nature ensures that adjustments take effect quickly, avoiding the potential risks caused by persistent unevenness.

[0105] Understandably, obtaining execution feedback data is a crucial step in verifying the effectiveness of adjustments. Automated equipment will return sensor data in real time during execution, such as actual distance traveled, force sensor readings, or position deviation values.

[0106] For example, feedback data might show an actual offset of 0.048 meters in the x-direction and 0.031 meters in the y-direction, which is basically consistent with the command. By analyzing the feedback data, the execution status of the command is determined. If the feedback data is consistent with the issued command, a normal execution result is obtained, indicating that the adjustment plan has been accurately implemented and the force balance has been effectively maintained.

[0107] For example, under normal circumstances, the overall stability index of the shelving improves by more than 15%, reducing the risk of tilting. Otherwise, if discrepancies exist, abnormal execution results are obtained. In this case, deviation data is extracted from the abnormal execution results, for example, calculating a positional deviation of 0.012 meters or a force distribution deviation of 5%. Time series analysis algorithms are used to perform trend analysis on the deviation data to obtain the stability change trend.

[0108] Specifically, the deviation data from multiple consecutive feedback sessions are sorted by time, and their fluctuation amplitude and direction are analyzed. For example...

[0109] In one possible implementation, if the deviation sequence shows a gradually increasing trend, such as rising from 0.005 meters to 0.015 meters, it indicates that external disturbances are accumulating. The stability state of the shelving system is determined based on the trend of stability changes. If the trend exceeds a preset threshold, such as a volatility exceeding 10%, an unstable state is identified, indicating the need for further intervention; otherwise, a stable state is identified, confirming that the system has returned to reliable operation. This judgment mechanism is beneficial for early detection of problems and preventing small deviations from evolving into major instability.

[0110] Preferably, under stable conditions, the system records this adjustment as a successful case for subsequent optimization and learning, improving overall automation efficiency. Under unstable conditions, an alarm is triggered and some parameters are rolled back to ensure rack safety. This feedback loop design significantly improves the system's robustness and adaptability, enabling the racks to maintain long-term balanced stress in dynamic warehousing environments, reducing maintenance costs and enhancing operational safety.

[0111] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. The present invention has been described in detail with reference to preferred embodiments. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A containerized intelligent automated storage and retrieval system control system and method, characterized in that, The method includes: By collecting pressure distribution data at each storage location through a sensor network deployed at key locations on the shelf, the real-time force distribution matrix of the shelf is obtained, and the overall weight stress state of the shelf is determined. Based on the stress distribution matrix, calculate the pressure difference and average load level between each cargo location, use an optimization calculation method to identify areas of uneven stress, and determine the location and extent of the local overload area. If the pressure difference in the local overload area exceeds a preset threshold, the weight attribute data of the goods to be stored is extracted from the cargo weight database to determine the appropriate cargo combination set for adjustment. For the aforementioned cargo combination set, the impact of different storage locations on the overall weight distribution of the shelf is predicted by simulating a force distribution model, thereby obtaining an optimized storage location scheme; Based on the storage location scheme and the current equipment location data, a path optimization method is used to generate equipment operation paths and determine the path sequence that minimizes operation in high-load areas; The scheduling instructions of the path sequence update control system are used to obtain the new force distribution data after execution, and to determine whether the force balance state of the shelf has been reached. If the force equilibrium state is not reached, the remaining unevenness index is extracted from the new force distribution data, and the storage location and path parameters are further fine-tuned using parameter adjustment methods to obtain the final adjustment scheme. Based on the final adjustment plan, instructions are sent to the automated equipment in real time to obtain execution feedback data and determine the stability status of the racking system.

2. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, The process of collecting pressure distribution data at each storage location through a sensor network deployed at key locations on the shelving, obtaining the real-time force distribution matrix of the shelving, and determining the overall weight stress state of the shelving includes: By continuously collecting pressure distribution data of each storage location at key locations on the shelf through a sensor network, a raw pressure data set is obtained. Time synchronization and alignment are performed on the original pressure data set to obtain a cargo location pressure sequence under a unified timestamp; The sliding window method is used to extract local pressure change values ​​from the pressure sequence of cargo locations, and the pressure change matrix of each cargo location is obtained. By mapping the pressure change matrix of the storage location to the preset shelf structure position, a real-time stress distribution matrix of the shelf is generated; The total pressure in each area is calculated based on the real-time force distribution matrix of the shelving to obtain the overall weight distribution value of the shelving. If the overall weight distribution value exceeds the preset balance range, it is determined that the shelf is under unbalanced load and the location of the unbalanced load area is obtained. Based on the location of the off-center load area and the force distribution matrix, the set of abnormal stress points in the rack structure is determined.

3. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, The step of calculating the pressure difference and average load level between each cargo location based on the force distribution matrix, identifying unevenly stressed areas using an optimized calculation method, and determining the location and extent of the local overload areas includes: By obtaining the force distribution matrix data from the storage system, scanning each cargo point one by one, calculating the pressure difference between each cargo point and its adjacent cargo points, and obtaining the preliminary pressure difference distribution results; Based on the pressure difference distribution results, the average load level of all cargo locations is calculated. Statistical methods are used to compare the load value of each cargo location with the average load level to identify cargo locations with larger load deviations. For cargo locations with significant load deviations, obtain the stress data of surrounding cargo locations, and identify potential areas of localized overload through comparative analysis to determine if there are any uneven stress distributions. If a local overload area is identified, the specific stress value of each cargo location in that area is extracted. Combined with the area location information, the quantitative value of the overload degree is calculated to obtain the distribution of the overload degree. Based on the distribution of overload levels, a preset threshold is used for comparison. If the overload level of a certain area exceeds the threshold, the area is marked as a high-risk overload area, and its specific location is determined. By recording and classifying the specific location data of high-risk overload areas, corresponding area identification information is generated, and the final overload area location result is obtained.

4. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, If the pressure difference in the local overload area exceeds a preset threshold, the weight attribute data of the goods to be stored is extracted from the cargo weight database to determine a suitable set of cargo combinations for adjustment, including: Obtain the pressure difference value of the local overload area; If the pressure difference exceeds the preset threshold, the weight attribute data of the goods to be stored will be extracted from the cargo weight database. Based on the extracted weight attribute data, determine the weight sorting sequence of the goods to be stored; By sorting the weight sequence, determine the degree of matching between weight attribute data and local overload areas; If the matching degree is lower than the preset level, the K-means clustering algorithm is used to group the weight attribute data to obtain a cargo combination set; Based on the set of goods combinations, determine the appropriate subset of goods combinations to adjust; Obtain adjusted weight distribution data by combining a subset of goods.

5. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, For the aforementioned goods combination set, the impact of different storage locations on the overall weight distribution of the shelving is predicted using a simulated force distribution model, resulting in an optimized storage location scheme, including: Obtain the weight and volume data of each item in the cargo combination set; The local force values ​​of goods placed in different storage positions on the shelf were calculated using a force distribution model. Simulate and predict the overall shelf weight distribution for each storage location; Determine whether the weight distribution exceeds the preset stress threshold of the shelf structure; if it does, mark the storage location as unusable. For unavailable storage locations, exclude the corresponding location from the cargo combination set; A genetic algorithm is used to iteratively optimize the remaining storage location schemes to obtain the location combination with the most balanced weight distribution; The uniformity of the overall stress distribution of the shelving under the optimized storage location scheme was verified by using a finite element analysis model, and the final optimized storage location scheme was determined.

6. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, The step of generating equipment operation paths using a path optimization method based on the storage location scheme and current equipment location data, and determining the path sequence that minimizes operation in high-load areas, includes: Step 1: Obtain device location data and a preset storage scheme. Then, use the data parsing module to perform structured processing on the data to obtain the device's location distribution information in the current environment. Step 2: Based on the location distribution information, using preset path planning rules and combined with the identification data of high-load areas, generate a preliminary work path sequence and determine the operation nodes involved in the path. Step 3: For the initial job path sequence, the Dijkstra algorithm is used to optimize the path, calculate the shorter path that avoids the high-load area, and obtain the optimized job path; Step 4: Extract the operation sequence from the optimized job path. If there are still high-load nodes in the operation sequence, readjust the priority of the path nodes to determine the final operation sequence. Step 5: Based on the final operation sequence, generate a detailed time schedule for equipment operations, compress the operation time for high-load areas, and obtain a time-reduced operation plan; Step Six: Generate execution instructions for the equipment operation path through the work plan. Combined with the constraints of the area operation, determine if a temporary high-load area is encountered during the path execution, and then dynamically switch to the backup path to determine the final execution path.

7. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, The step of updating the scheduling instructions of the control system through the path sequence, obtaining the new force distribution data after execution, and determining whether the force balance state of the shelf has been reached includes: The path sequence generation module obtains the latest path sequence data to determine the basis for updating scheduling instructions. Based on the path sequence data, the built-in logic processing unit of the control system generates corresponding scheduling instructions to obtain a preliminary plan for instruction execution. The command execution module sends the generated scheduling commands to the relevant equipment of the rack structure, collects the force distribution data after execution, and determines whether it meets the preset balance conditions. If the force distribution data exceeds the preset equilibrium state threshold, the deviation information is obtained through the system feedback mechanism to determine the path sequence that needs to be adjusted. Based on the deviation information, the support vector machine algorithm is used to optimize the path sequence to obtain the adjusted path sequence data; By updating the control system's scheduling instructions with the adjusted path sequence data, new force distribution data is obtained to determine whether the rack structure has reached a balanced state. If the new force distribution data still has not reached equilibrium, the key features of the force distribution are extracted through the distribution analysis module to determine the optimization direction of the subsequent path sequence.

8. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, If the force equilibrium state is not reached, the remaining unevenness index is extracted from the new force distribution data, and the storage location and path parameters are further fine-tuned using parameter adjustment methods to obtain the final adjustment scheme, including: The remaining unevenness index is extracted from the new force distribution data to obtain the location coordinates of multiple local unevenness regions; Based on the location coordinates of multiple local uneven regions, determine the offset of the storage location parameters corresponding to each region; The gradient descent algorithm is used to calculate the joint adjustment direction of the storage location parameter offset and the path parameter to obtain the preliminary parameter correction vector; The stored position and path parameters are updated by modifying the initial parameter correction vector to obtain the updated force distribution data; Extract the remaining unevenness index from the updated force distribution data and determine whether all the remaining unevenness indexes are below the preset threshold. If all remaining unevenness indicators are below the preset threshold, the current storage location parameters and path parameters will be output as the final adjustment scheme. If there are still remaining unevenness indicators higher than the preset threshold, the fine-tuning range of the storage location parameters and path parameters will be re-determined based on the location coordinates of the remaining unevenness indicators that are higher than the threshold, and the final adjustment scheme will be obtained.

9. The containerized intelligent automated storage and retrieval system control system and method according to claim 1, characterized in that, The step of issuing instructions to automated equipment in real time according to the final adjustment plan, obtaining execution feedback data, and determining the stability status of the racking system includes: The instructions will be generated and issued based on the final adjustment plan; Send instructions to automated equipment in real time; Obtain execution feedback data; The execution status of the instruction is determined by the execution feedback data. If the execution feedback data is consistent with the issued instruction, a normal execution result is obtained; otherwise, an abnormal execution result is obtained. Extract deviation data from abnormal execution results; Time series analysis algorithms are used to perform trend analysis on the deviation data to obtain the stability change trend; The stability state of the shelving system is determined based on the trend of stability changes. If the trend of change exceeds a preset threshold, an unstable state is obtained; otherwise, a stable state is obtained.

Images