Supply chain warehouse management system and method based on AI digital intelligence

Abstract

The invention discloses a supply chain warehouse management system and method based on AI digital intelligence, and belongs to the technical field of warehouse management. A sensing unit comprises a 3D vision module, an environment sensing module and a cargo positioning module; the edge calculation unit comprises an Internet of Things gateway module, an edge database and an edge calculation module; the data twinning unit comprises a three-dimensional modeling engine module and a virtual deduction module; the intelligent decision-making unit comprises a path optimization module, a demand prediction module and an abnormality diagnosis module; the execution unit comprises an AGV conveying module and a stacking machine storage module. According to the method, closed-loop intelligent management of the whole warehousing process from cargo sensing, intelligent decision making to automatic execution is achieved, accurate prediction, scientific decision making and continuous optimization are driven based on mass multi-source data fusion and AI analysis, and the real-time performance of the whole warehousing process is improved through real-time sensing, edge calculation, digital twinborn deduction and intelligent decision making. The system has strong ability to deal with uncertainties such as order fluctuation, equipment fault and environment change, and improves operation toughness.

Description

Technical Field

[0001] This invention relates to an AI-based digital supply chain warehousing management system and method. Background Technology

[0002] In enterprise supply chain warehousing management, there are numerous unseen tasks such as procurement, inbound / outbound operations, warehouse utilization, and procurement timeliness. Modern automated warehouses handle a large volume of sorting, transportation, and inbound / outbound operations daily. If equipment malfunctions or fails, localized anomalies directly impact the overall system, significantly reducing operational efficiency. Currently, many automated warehouses distribute management, production, and warehousing data across independent operating systems, lacking unified oversight, thus increasing management costs. Traditional production resource allocation, verification, and adjustment are poorly automated and heavily influenced by human factors, directly affecting production line efficiency. In actual production, factors such as warehouse design, equipment malfunctions, multi-task switching, high-frequency access to various tools, and storage location rotation make it difficult for each sub-module to achieve its intended production efficiency. This poses a significant obstacle to analyzing the efficiency of each sub-module in the actual production process and accurately identifying inefficient modules. Therefore, this paper proposes an AI-based digital supply chain warehousing management system and its methodology. Summary of the Invention

[0003] The purpose of this invention is to solve the problems existing in the prior art by proposing an AI-based digital supply chain warehousing management system and method.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] The AI-based intelligent supply chain warehouse management system includes a sensing unit, an edge computing unit, a data twin unit, an intelligent decision-making unit, and an execution unit.

[0006] The sensing unit includes a 3D vision module, an environmental sensing module, and a cargo positioning module.

[0007] The edge computing unit includes an IoT gateway module, an edge database, and an edge computing module.

[0008] The data twin unit includes a 3D modeling engine module and a virtual simulation module;

[0009] The intelligent decision-making unit includes a path optimization module, a demand prediction module, and an anomaly diagnosis module.

[0010] The execution unit includes an AGV conveying module and a stacker crane storage module.

[0011] As a preferred embodiment, the 3D vision module includes a depth camera and an RFID reader installed at multiple angles within the warehousing center.

[0012] The environmental sensing module includes a temperature and humidity sensor and a vibration sensor;

[0013] The cargo positioning module includes a UWB positioning base station and an RFID tag.

[0014] As a preferred embodiment, the 3D modeling module is developed based on the UnityHDRP pipeline and uses the constructive solid CSG geometric algorithm to automatically generate shelf models;

[0015] The path optimization module uses a fusion of AI algorithm and NSGA-Ⅲ genetic algorithm to output Pareto optimal path with multiple objective functions, including minimizing transportation distance, equipment energy consumption and maximizing timeliness.

[0016] The demand forecasting module uses an LSTM plus Attention mechanism to fuse historical sales and external market characteristics to output replenishment decisions.

[0017] The anomaly diagnosis module detects equipment anomalies based on the isolated forest algorithm and outputs the root cause of the fault by associating it with a knowledge graph.

[0018] The warehousing management method proposed by the AI-based intelligent supply chain warehousing management system includes the following specific management methods:

[0019] The intelligent inbound management method includes the following steps:

[0020] a. Cargo perception and identification: Cargo point cloud data is acquired through a depth camera, cargo dimensions are calculated using the PnP algorithm, and cargo information is parsed using an RFID reader;

[0021] b. Dynamic warehouse allocation: The NSGA-III algorithm is used to solve the multi-objective optimization problem, and the optimal warehouse coordinates are output through TOPSIS decision-making.

[0022] c. Automated Goods Inbound: AGVs plan routes and transport goods based on AI algorithms, stacker cranes locate target storage locations, and the inbound status is verified by both RFID and 3D vision.

[0023] The intelligent inventory management method includes the following steps:

[0024] a. Automated data collection: AGVs carrying RFID readers patrol the warehouse, and the collection frequency is dynamically adjusted in conjunction with ABC classification strategies;

[0025] b. Discrepancy Intelligent Analysis: When the inventory deviation is >5%, trace the inbound and outbound records and call the video behavior analysis engine;

[0026] c. Automatic correction and calibration: The six-axis robotic arm adjusts the position of the cargo and uses the ICP algorithm to match the target posture, with a maximum deviation angle of ≤5°;

[0027] The virtual simulation optimization method includes the following steps:

[0028] a. Construct a digital twin scenario: Import the BIM model and bind it to real-time sensor data streams;

[0029] b. Start discrete event simulation: Monte Carlo generates order flow with a time step of 100ms;

[0030] c. Multi-strategy parallel derivation: Genetic algorithm, particle swarm optimization and simulated annealing are used to solve the Pareto solution set;

[0031] d. Real-time risk prediction: Detect equipment anomalies using the Isolation Forest algorithm and dynamically replan the path;

[0032] e. Solution verification output: Orthogonal experiment to compare KPI indicators, requiring virtual simulation error rate <5%.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] 1. The warehouse management system proposed in this invention realizes closed-loop intelligent management of the entire warehousing process, from cargo perception and intelligent decision-making to automated execution. Based on the fusion of massive multi-source data and AI analysis, it drives accurate prediction, scientific decision-making and continuous optimization. Through real-time perception, edge computing, digital twin inference and intelligent decision-making, the system has a strong ability to cope with uncertainties such as order fluctuations, equipment failures and environmental changes, and improves operational resilience.

[0035] 2. The warehouse management system proposed in this invention significantly improves warehouse space utilization, equipment utilization, and personnel efficiency, reduces energy consumption, losses, and operating costs, shortens order response time, and can identify equipment failures, inventory anomalies, process bottlenecks, and safety risks in advance, transforming passive response into proactive prevention, and realizing visualization, traceability, and auditability of the entire life cycle of goods and the entire operation process. Attached Figure Description

[0036] Figure 1 This is a framework diagram of the AI-based digital supply chain warehousing management system proposed in this invention;

[0037] Figure 2 This is a flowchart of the intelligent inbound management method in the AI-based digital supply chain warehousing management method proposed in this invention;

[0038] Figure 3 This is a flowchart of the intelligent inventory management method in the AI-based digital supply chain warehousing management method proposed in this invention;

[0039] Figure 4This is a flowchart of the virtual simulation optimization method in the AI-based intelligent supply chain warehousing management method proposed in this invention;

[0040] Figure 5 This is the core data flow diagram of the AI-based digital supply chain warehousing management method proposed in this invention. Detailed Implementation

[0041] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0042] In the description of this invention, it should be noted that the terms "upper," "lower," "inner," "outer," "top / bottom," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0043] Example, refer to Figures 1 to 5 The AI-based intelligent supply chain warehouse management system includes a sensing unit, an edge computing unit, a data twin unit, an intelligent decision-making unit, and an execution unit.

[0044] The sensing unit includes a 3D vision module, an environmental sensing module, and a cargo positioning module;

[0045] The edge computing unit includes an IoT gateway module, an edge database, and an edge computing module;

[0046] The data twin unit includes a 3D modeling engine module and a virtual simulation module;

[0047] The intelligent decision-making unit includes a path optimization module, a demand forecasting module, and an anomaly diagnosis module;

[0048] The execution unit includes an AGV conveying module and a stacker crane storage module.

[0049] The 3D vision module includes depth cameras and RFID readers set up in the warehouse center from multiple angles. It combines multi-view fusion and SLAM technology to build a real-time environmental map, improves positioning accuracy and scene understanding, and can also introduce event cameras to deal with high-speed movement or low-light scenes.

[0050] In addition to size recognition, point cloud analysis of the centroid is used for cargo stacking stability analysis, surface defect identification and damage detection, and personnel safety monitoring to identify personnel entering dangerous areas and falling. Temperature and humidity data are not only used for monitoring, but can also be linked to the HVAC system to achieve dynamic energy-saving control; vibration data combined with equipment knowledge graphs enables more accurate predictive maintenance for bearing wear, belt loosening, etc.

[0051] The environmental sensing module includes temperature and humidity sensors and vibration sensors; the cargo positioning module includes a UWB positioning base station and RFID tags. The integration of UWB with an inertial navigation unit (IMU) improves the robustness of AGV positioning in areas with signal obstruction. The RFID tags use active / semi-active tags to track high-value goods more accurately.

[0052] The 3D modeling module is developed based on the UnityHDRP pipeline and uses the CSG geometric algorithm to automatically generate shelf models, virtually reconstruct the physical warehouse, and bind real-time data to the real scene and virtual state to conduct scheme pre-simulation and risk verification.

[0053] In addition to CSG-generated shelves, high-precision modeling of irregular goods or special equipment is performed using Photogrammetry or laser scanning technology;

[0054] The path optimization module uses a fusion of AI algorithm and NSGA-Ⅲ genetic algorithm to output Pareto optimal path with multiple objective functions, including minimizing transportation distance, equipment energy consumption and maximizing timeliness.

[0055] The demand forecasting module uses LSTM with an attention mechanism to fuse historical sales data with external market characteristics and outputs replenishment decisions.

[0056] The anomaly diagnosis module detects equipment anomalies based on the isolated forest algorithm and outputs the root cause of the fault by associating it with a knowledge graph.

[0057] The warehousing management method proposed by the AI-based intelligent supply chain warehousing management system includes the following specific management methods:

[0058] The intelligent inbound management method includes the following steps:

[0059] Cargo perception and identification: Cargo point cloud data is acquired through a depth camera, cargo size is calculated by combining PnP algorithm, cargo information is parsed through RFID reader, and cargo appearance damage detection and tag integrity check are performed by using 3D vision and deep learning while identifying size and information.

[0060] Dynamic warehouse allocation: The NSGA-III algorithm is used to solve the multi-objective optimization problem. The optimal warehouse coordinates are output through TOPSIS decision-making, historical allocation effects are recorded, and reinforcement learning is used to continuously optimize the allocation strategy.

[0061] Automated Goods Inbound: AGVs plan routes and transport goods based on AI algorithms, stacker cranes locate target storage locations, and the inbound status is verified by both RFID and 3D vision. The verification results need to be fed back to the WMS system in real time to update the inventory status, and video clips and operation logs of the inbound process are recorded for auditing purposes.

[0062] The intelligent inventory management method includes the following steps:

[0063] Automated data collection: AGVs carrying RFID readers patrol the warehouse, and the collection frequency is dynamically adjusted in conjunction with ABC classification strategies;

[0064] Construct a root cause analysis tree or Bayesian network model for discrepancy analysis, and combine it with data such as inventory entry and exit records, video analysis, operation logs, and environmental data to automatically infer the most likely causes of discrepancies (such as data entry errors, theft, unreported damage, and system vulnerabilities).

[0065] Discrepancy Intelligent Analysis: When the inventory deviation is >5%, trace the inbound and outbound records and call the video behavior analysis engine;

[0066] Formula for calculating differences:

[0067]

[0068] Automatic correction and calibration: The position of the cargo is adjusted by a six-axis robotic arm, and the target posture is matched by the ICP algorithm, with a maximum deviation angle of ≤5°;

[0069] The virtual simulation optimization method includes the following steps:

[0070] Building a digital twin scenario: Importing the BIM model and binding it to real-time sensor data streams;

[0071] Start discrete event simulation: Monte Carlo generates order flow with a time step of 100ms. In addition to Monte Carlo, historical real order data or future order data generated by the demand forecasting module can be imported for more realistic simulation.

[0072] Parallel inference of multiple strategies: Genetic algorithm, particle swarm optimization and simulated annealing are used to solve the Pareto solution set. The key parameters of the optimization algorithms such as genetic algorithm and particle swarm optimization (such as population size, crossover / mutation rate, inertia weight, etc.) are automatically tuned (such as Bayesian optimization) to improve the optimization efficiency.

[0073] Real-time risk prediction: Detects device anomalies using the Isolation Forest algorithm and dynamically replans the path;

[0074] Solution verification output: Orthogonal experiment to compare KPI indicators, requiring virtual simulation error rate <5%.

[0075] The present invention has the following effects:

[0076] Intelligent management throughout the entire process: achieving closed-loop intelligent management of the entire warehousing process, from cargo perception and intelligent decision-making to automated execution.

[0077] Data-driven decision-making: Based on the fusion of massive multi-source data and AI analysis, it drives accurate prediction, scientific decision-making and continuous optimization.

[0078] Dynamic Adaptability and Resilience: Through real-time sensing, edge computing, digital twin simulation, and intelligent decision-making, the system has a strong ability to cope with uncertainties such as order fluctuations, equipment failures, and environmental changes, thereby enhancing operational resilience.

[0079] Resource optimization and cost reduction: Significantly improve warehouse space utilization, equipment utilization, and personnel efficiency, reduce energy consumption, losses and operating costs, and shorten order response time.

[0080] Risk Foresight and Control: Proactively identify equipment failures, inventory anomalies, process bottlenecks, and safety risks, transforming passive response into proactive prevention.

[0081] Transparency and Traceability: Achieving visualization, traceability, and auditability throughout the entire lifecycle of goods and the entire operational process.

[0082] Continuous evolution: Based on data feedback and virtual simulation, system strategies and models can be continuously iterated and optimized to adapt to the needs of business development.

[0083] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An AI-based digital supply chain warehouse management system, characterized in that, The system comprises a perception unit, an edge computing unit, a data twin unit, an intelligent decision-making unit, and an execution unit. The perception unit comprises a 3D vision module, an environment sensing module, and a cargo positioning module. The edge computing unit comprises an Internet of Things gateway module, an edge database, and an edge computing module. The data twin unit comprises a three-dimensional modeling engine module and a virtual reasoning module. The intelligent decision-making unit comprises a path optimization module, a demand prediction module, and an anomaly diagnosis module. The execution unit comprises an AGV conveying module and a stacker warehouse module.

2. The AI-based supply chain warehouse management system of claim 1, wherein, The 3D vision module comprises multi-angle depth cameras and RFID readers arranged in the warehouse center. The environment sensing module comprises temperature and humidity sensors and vibration sensors. The cargo positioning module comprises UWB positioning base stations and RFID tags. 3.The AI-based supply chain warehouse management system of claim 1, wherein, The three-dimensional modeling module is based on the Unity HDRP pipeline and uses a constructive solid geometry (CSG) algorithm to automatically generate a shelf model. The path optimization module uses an AI algorithm combined with the NSGA-III genetic algorithm to minimize the carrying distance, device energy consumption, and maximize the time efficiency as a multi-objective function to output the Pareto optimal path. The demand prediction module uses LSTM with an attention mechanism to fuse historical sales and external market characteristics to output restocking decisions. The anomaly diagnosis module uses an isolation forest algorithm to detect device anomalies and associates with a knowledge graph to output fault root causes.

4. The warehouse management method based on the AI digitalization of the supply chain warehouse management system according to any one of claims 1-3, characterized in that, The warehouse management method comprises the following management methods: The intelligent warehousing management method comprises the following steps: a. Cargo perception and identification: Obtain cargo point cloud data through a depth camera, calculate cargo dimensions using a PnP algorithm, and analyze cargo information through an RFID reader. b. Dynamic bin allocation: Use the NSGA-III algorithm to solve multi-objective optimization problems and output optimal bin coordinates through TOPSIS decision-making. c. Automatic warehousing of goods: AGV plans a path to transport goods based on an AI algorithm, a stacker positions the target bin, and the warehouse state is verified through RFID and 3D vision. The intelligent inventory management method comprises the following steps: a. Automated data collection: Use an AGV to carry an RFID reader to patrol the warehouse and dynamically adjust the collection frequency based on the ABC classification strategy. b. Intelligent difference analysis: When the inventory deviation is greater than 5%, trace the warehousing records and call the video behavior analysis engine. c. Automatic correction and calibration: Adjust the cargo position using a six-axis robot arm, match the target pose using the ICP algorithm, and ensure that the maximum deviation angle is less than or equal to 5 degrees. The virtual reasoning optimization method comprises the following steps: a. Build a digital twin scene: Import a BIM model and bind real-time sensor data streams. b. Start discrete event simulation: Generate order streams using Monte Carlo, with a time step of 100 ms. c. Parallel reasoning with multiple strategies: Use genetic algorithms, particle swarm optimization, and simulated annealing to solve the Pareto solution set. d. Real-time risk prediction: Detect device anomalies using an isolation forest algorithm and dynamically re-plan the path. e. Scheme verification output: Compare KPI indicators through orthogonal experiments, and require that the virtual reasoning error rate be less than 5%.

Images