Intelligent tool warehouse management system

Abstract

The invention provides an intelligent tool warehouse management system, which belongs to the field of tool management and comprises a multi-mode sensing module, an intelligent access control module, an edge calculation module, an anomaly analysis module and a visualization module. The multi-mode sensing module realizes identity recognition, real-time positioning and integrity verification of the tool through RFID and UWB dual-mode positioning, visual recognition and sensor fusion technologies; the intelligent access control module is based on dynamic authority control and gravity sensing goods shelves, and tool storing and taking compliance is ensured; the edge computing module adopts localized data processing and low-power-consumption communication, and supports network disconnection disaster recovery; the abnormity analysis module identifies violation operation through an LSTM model and triggers grading alarm; the visualization module displays the tool state and the warehouse thermodynamic diagram in real time based on the digital twinning technology. The problems that traditional warehouse management depends on manpower, efficiency is low and errors are prone to occurring are solved, and high-precision tracking, intelligent early warning and optimal scheduling of the tool in the whole life cycle are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of tool management, and in particular to an intelligent tool warehouse management system. Background Art

[0002] In industries like electric power, petrochemicals, and rail transit, tool management directly impacts operational safety and efficiency. Traditional warehouse management relies primarily on manual registration and barcode scanning, which presents the following issues: Insufficient positioning accuracy: Conventional RFID or QR code technology cannot track the movement of tools within the warehouse in real time, which can easily lead to tool loss or misplacement. Weak authority control: Manual verification makes it difficult to ensure that borrowed tools strictly match employee qualifications, posing the risk of illegal operations. Lack of status monitoring: Issues such as damaged tools, missing accessories, or aging batteries are difficult to detect in a timely manner, impacting operational safety. Therefore, an intelligent tool warehouse management system is proposed to address these issues. Summary of the Invention

[0003] In view of this, an object of the present invention is to provide an intelligent tool warehouse management system to at least solve the above problems.

[0004] The technical solution adopted in the present invention is as follows:

[0005] An intelligent tool warehouse management system, the system comprising:

[0006] Multimodal sensing module, intelligent access control module, edge computing module, anomaly analysis module and visualization module;

[0007] The multimodal sensing module uses RFID and UWB dual-mode positioning technology, visual recognition technology, and sensor fusion detection technology to perform tool identity recognition, real-time positioning, and integrity verification;

[0008] The intelligent access control module includes a dynamic permission control unit and a gravity-sensing shelf for tool access authorization and access behavior recording based on personnel qualifications;

[0009] The edge computing module includes a localized edge gateway and a low-power communication unit for real-time data processing and offline cache synchronization;

[0010] The anomaly analysis module uses a machine learning model to identify illegal operations and trigger an alarm;

[0011] The visualization module uses digital twin technology to display tool status and warehouse heat map.

[0012] Furthermore, the multimodal sensing module includes:

[0013] RFID tags embedded in the tool surface are used to identify the status of tools entering and leaving the warehouse in batches;

[0014] The UWB positioning base station deployed in the warehouse uses the Time Difference of Arrival (TDOA) algorithm to achieve centimeter-level positioning and trajectory tracking of the tool;

[0015] AI cameras and infrared curtain sensors are installed at the entrances and exits. When the infrared curtain sensor detects that a tool is approaching, it triggers the AI camera to capture the tool image. The convolutional neural network (CNN) is used to identify the tool's shape, color, and surface damage, and compares it with the pre-stored image database to verify the tool's integrity.

[0016] Furthermore, the intelligent access control module includes:

[0017] A dynamic permission control unit linked to the work order management system generates a list of borrowable tools based on employee qualifications;

[0018] Biometric access control terminals, used to verify facial or fingerprint information and unlock authorized shelves;

[0019] The gravity-sensing shelf integrates a pressure sensor array. When a tool is removed, the pressure sensor detects the weight change and compares it with the pre-stored tool weight threshold, automatically recording the tool name, removal time and operator information; the weight difference is checked again when the tool is returned. If the weight deviation is detected to be more than 5%, the accessory is determined to be missing and an alarm is triggered.

[0020] Furthermore, the edge computing module further includes:

[0021] A local data processing unit deployed on the NVIDIA Jetson edge device to interpret sensor and camera data in real time;

[0022] A disconnected cache unit stores operation records during network interruptions and synchronizes them to the cloud after network recovery;

[0023] Tags using LoRaWAN or NB-IoT communication have a battery life of no less than 3 years, and the warehouse gateway supports solar power supply.

[0024] Furthermore, the abnormality analysis module includes:

[0025] Establish a normal operation baseline based on the behavioral pattern learning unit of the long short-term memory network LSTM;

[0026] A real-time detection unit to flag the behavior of multiple people picking up the same tool in a short period of time or moving the tool during non-working hours;

[0027] The alarm linkage unit triggers sound and light alarms and pushes alarm information including on-site images to the administrator.

[0028] Furthermore, the visualization module includes:

[0029] A digital twin warehouse model based on the Unity engine displays tool location, status, and lifecycle progress in real time;

[0030] Heat map generation unit to analyze tool access frequency and optimize shelf layout;

[0031] Overdue marking units are not returned, and overdue tools are indicated by a flashing red light and a text message reminder is sent.

[0032] Furthermore, it also includes a predictive maintenance module:

[0033] For the helmet life prediction subunit, the Weibull distribution algorithm is used to calculate the number of times used and UV exposure data, and a replacement reminder is generated 30 days in advance;

[0034] For the battery monitoring subunit of power tools, the capacity attenuation rate is calculated based on the charge and discharge curves and temperature and humidity data, and a maintenance alarm is triggered when it is lower than 70%.

[0035] Furthermore, the collaborative positioning of RFID and UWB includes:

[0036] RFID readers scan tags in batches to generate lists when tools enter and leave the warehouse;

[0037] The UWB base station continuously tracks the location of tools in the warehouse and binds the coordinates of the shelf when the tools are stationary for more than 5 minutes;

[0038] When the UWB signal is lost, the latest RFID records and gravity sensing shelf data are called for comprehensive positioning.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. High-precision multi-modal positioning:

[0041] RFID and UWB collaborative positioning realizes batch entry and exit identification (RFID) and centimeter-level trajectory tracking within the warehouse (UWB). When the UWB signal is lost, three levels of compensation are used to ensure positioning continuity through RFID recent records, gravity sensing shelves and visual positioning.

[0042] 2. Dynamic intelligent access control:

[0043] The work order system is linked to the employee qualification database to automatically match the list of borrowable tools. Gravity-sensing shelves analyze weight differences (>5%) to prevent the wrong removal of tools or missing accessories.

[0044] 3. AI-driven abnormal warning:

[0045] The LSTM model learns normal operating patterns and identifies anomalies such as tool movement during non-working hours and multiple people picking up items within a short period of time.

[0046] Three-level sound and light alarm (strobe light + 80dB siren) is linked to access control and locking, and the alarm information includes on-site snapshots and biometric features.

[0047] 4. Visualization and predictive maintenance

[0048] The digital twin warehouse displays tool location and lifespan progress (e.g., number of days remaining on a hard hat) in real time;

[0049] The Weibull distribution algorithm predicts helmet aging, and the charge and discharge curves analyze battery health, triggering a replacement reminder 30 days in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only preferred embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0051] Figure 1 This is a schematic diagram of the overall structure of an intelligent tool warehouse management system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present invention more apparent, exemplary embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described in the present invention, all other embodiments obtained by those skilled in the art without creative work should fall within the scope of protection of the present invention.

[0053] In the following description, numerous specific details are provided to provide a more thorough understanding of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced without one or more of these details. In other instances, certain technical features well known in the art are not described to avoid confusion with the present invention.

[0054] It should be understood that the present invention can be implemented in different forms and should not be interpreted as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to make disclosure thorough and complete and to fully convey the scope of the present invention to those skilled in the art.

[0055] The purpose of the terms used herein is only to describe specific embodiments and is not intended to limit the present invention. When used herein, the singular forms "a", "an", and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "comprising" and / or "comprising", when used in this specification, determine the presence of the features, integers, steps, operations, elements and / or parts, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, parts and / or groups. When used herein, the term "and / or" includes any and all combinations of the relevant listed items.

[0056] In order to fully understand the present invention, a detailed structure will be provided in the following description to illustrate the technical solution proposed by the present invention. Optional embodiments of the present invention are described in detail below. However, in addition to these detailed descriptions, the present invention may also have other implementations.

[0057] Reference Figure 1 The present invention provides an intelligent tool warehouse management system, the system comprising:

[0058] Multimodal sensing module, intelligent access control module, edge computing module, anomaly analysis module and visualization module;

[0059] The multimodal sensing module uses RFID and UWB dual-mode positioning technology, visual recognition technology, and sensor fusion detection technology to perform tool identity recognition, real-time positioning, and integrity verification;

[0060] The intelligent access control module includes a dynamic permission control unit and a gravity-sensing shelf for tool access authorization and access behavior recording based on personnel qualifications;

[0061] The edge computing module includes a localized edge gateway and a low-power communication unit for real-time data processing and offline cache synchronization;

[0062] The anomaly analysis module uses a machine learning model to identify illegal operations and trigger an alarm;

[0063] The visualization module uses digital twin technology to display tool status and warehouse heat map.

[0064] For example, the RFID and UWB dual-mode positioning in the multimodal sensing module can realize batch tool entry and exit identification through embedded RFID tags, and use the TDOA algorithm in combination with the UWB base station to achieve centimeter-level positioning. For visual recognition, it can set up AI cameras and infrared curtain sensors at the entrance and exit to link. For sensor fusion detection: it can achieve multi-dimensional monitoring of tool status by integrating data such as gravity sensing. The Sino-Russian dynamic permission unit of the intelligent access control module can be linked with the work order management system and dynamically generate a list of operable tools based on employee qualifications. It can also unlock authorized shelves through biometric access control terminals. Gravity sensing shelves can use pressure sensor arrays to record tool placement behavior in real time, detect missing accessories through weight difference verification, and automatically generate operation logs. The localized edge gateway in the edge computing module deploys a local data processing unit based on the NVIDIA Jetson platform, enabling real-time analysis of sensor and camera data streams. The low-power communication network utilizes the LoRaWAN / NB-IoT protocol to achieve a three-year battery life for tool tags, and the warehouse gateway supports solar power. For network outage recovery, a local cache unit can be configured to store operation records during network outages and automatically synchronize them to the cloud upon recovery. The anomaly analysis module uses an LSTM long-short-term memory network to establish a baseline for normal operations, learn time series characteristics of tool usage, and flag abnormal behaviors. The visualization module enables three-dimensional visual management and control, using digital twin modeling to map tool location, status, and lifespan in real time, and display a heat map of the warehouse, providing decision support for optimizing shelf layouts.

[0065] The multimodal sensing module includes:

[0066] RFID tags embedded in the tool surface are used to identify the status of tools entering and leaving the warehouse in batches;

[0067] The UWB positioning base station deployed in the warehouse uses the Time Difference of Arrival (TDOA) algorithm to achieve centimeter-level positioning and trajectory tracking of the tool;

[0068] AI cameras and infrared curtain sensors are installed at the entrances and exits. When the infrared curtain sensor detects that a tool is approaching, it triggers the AI camera to capture the tool image. The convolutional neural network (CNN) is used to identify the tool's shape, color, and surface damage, and compares it with the pre-stored image database to verify the tool's integrity.

[0069] For example, RFID tags can be embedded on the surface of tools, and each RFID tag stores a unique EPC code and basic tool information: model, specifications, and calibration reference values. When a tool passes through the warehouse entrance or exit, the RFID reader deployed on the door frame performs a 360° coverage scan using an 8dBi circularly polarized antenna and generates a tool entry and exit event list, including a timestamp, tool list, and operator ID. It can also be compared with the cloud asset ledger through a timed polling mechanism (every 15 minutes) to generate an inventory difference report. The UWB centimeter-level positioning layer deploys 6 Decawave The UWB base station of the DW1000 chip constitutes a TDOA positioning network. The positioning algorithm can calculate the tag coordinates through the arrival time difference TDOA, and cooperate with the Kalman filter to achieve dynamic trajectory smoothing. Its accuracy is controlled within the horizontal positioning error ≤ 10cm and the vertical height error ≤ 5cm. When the tool is stationary for more than 5 minutes, the coordinates are automatically bound to the nearest gravity-sensing shelf to form a "virtual electronic fence". When the UWB signal is lost, the RFID location record of the last 30 seconds can be called as compensation data; when the tool enters / leaves, the infrared curtain sensor is blocked and triggers the AI camera to capture. Its image processing process is as follows: convolutional neural network extracts tool contour features, color space conversion detects surface color difference changes, defect detection sub-network identifies cracks, wear and other anomalies, and performs feature matching with the pre-stored 3D model library. For integrity verification, the tool weight baseline value recorded by RFID can be compared with the actual value measured by the gravity-sensing shelf at the same time. Its specific technical effects are:

[0070] When maintenance personnel pass through entrances and exits wearing insulating gloves;

[0071] The RFID reader identifies the tag and records the outbound event;

[0072] The UWB base station continuously tracks the movement trajectory of the gloves to shelf No. 3;

[0073] The infrared curtain triggers the camera to capture the image of the glove, and the CNN detects that the surface is intact and the color is normal.

[0074] The gravity-sensing shelf detects a weight reduction of 150g (consistent with the baseline value) and automatically links to the work order system;

[0075] All data is verified at the edge gateway and the release decision is made.

[0076] The intelligent access control module includes:

[0077] A dynamic permission control unit linked to the work order management system generates a list of borrowable tools based on employee qualifications;

[0078] Biometric access control terminals, used to verify facial or fingerprint information and unlock authorized shelves;

[0079] The gravity-sensing shelf integrates a pressure sensor array. When a tool is removed, the pressure sensor detects the weight change and compares it with the pre-stored tool weight threshold, automatically recording the tool name, removal time and operator information; the weight difference is checked again when the tool is returned. If the weight deviation is detected to be more than 5%, the accessory is determined to be missing and an alarm is triggered.

[0080] For example, a dynamic permission control unit can implement a work order linkage mechanism that interacts with an external work order management system in real time via an API. This mechanism dynamically generates a list of borrowable tools based on the current task type, employee qualifications (such as high-voltage electrician certificates and safety officer certificates), and historical operation records. For example, when a "10kV line maintenance" work order is detected, only employees with the corresponding qualifications are allowed to borrow tools such as insulating gloves and electroscopes. A role-based access control model can also be used to categorize tools into three levels: "normal," "high-risk," and "privileged." High-risk tools require dual verification and authorization, while privileged tools require remote approval from the department head before they can be unlocked. Biometric access control terminals can integrate a facial recognition module with a liveness detection algorithm and a semiconductor fingerprint recognition unit to achieve dual biometric verification. For example, employees first confirm their identity through facial recognition, then press their fingerprint to match it to a pre-stored template. Passing this dual verification triggers the release of the shelf's electromagnetic lock. Furthermore, a 3D structured light sensor deployed at the top of the door frame can construct a human silhouette model. If an unauthorized person is detected following, the door opening command is immediately terminated and an audible and visual alarm is triggered. A 4×4 array of thin-film pressure sensors is embedded on the surface of each pallet on the shelf, forming 16 independent detection areas. When a tool is placed, the system identifies the tool's outline through a pressure distribution cloud map, automatically matches it to a pre-stored database of tool 3D models, and uses a dynamic threshold calibration algorithm for retrieval detection. Specifically, when a tool is removed, the pressure mutation value ΔP is collected in real time and compared with the standard weight G of the tool. When |ΔP / G| exceeds 3%, it is determined to be an abnormal removal (such as mistakenly taking another manual tool). Double verification is performed when the tool is returned. Weight difference verification: If the deviation between the detected weight G and the standard value G exceeds 5%, an accessory missing alarm is triggered. Piezoelectric sensors can also be used to collect vibration signals when the tool is placed and compare them with pre-stored spectrum templates to identify whether it is a similar tool, preventing Class A tools from being returned as Class B tools. Chain-based evidence storage can be performed during return and retrieval operations. That is, each access operation generates a blockchain evidence record containing the tool's unique code, the operator's biometric hash value, and a timestamp. The data is encrypted in the TEE trusted execution environment before being uploaded to the chain to ensure tamper-proof and traceable.

[0081] The edge computing module also includes:

[0082] A local data processing unit deployed on the NVIDIA Jetson edge device to interpret sensor and camera data in real time;

[0083] A disconnected cache unit stores operation records during network interruptions and synchronizes them to the cloud after network recovery;

[0084] Tags using LoRaWAN or NB-IoT communication have a battery life of no less than 3 years, and the warehouse gateway supports solar power supply.

[0085] For example, the local data processing unit deployed on the NVIDIA Jetson edge device can pre-process the raw data transmitted by the multimodal sensing module (such as RFID tag sequence, UWB base station signal strength, camera video stream). For example, when a worker removes insulating gloves from the shelf, the edge device simultaneously parses the RFID tag information, UWB positioning coordinates and gravity sensor weight changes, and completes the "tool removal-personnel verification-status confirmation" ternary verification locally; the offline cache unit can adopt a dual-mode storage architecture, and under normal network conditions, the operation records are uploaded to the cloud database in real time; when the 4G / Wi-Fi signal is detected to be interrupted, it automatically switches to the local NAND flash array storage. By using LoRaWAN or NB-IoT communication tags in conjunction with energy harvesting circuits, the battery life can reach 42 months under the condition of uploading positioning data 3 times a day, and the warehouse gateway supports the synergistic effect of solar power supply technology.

[0086] The abnormality analysis module includes:

[0087] Establish a normal operation baseline based on the behavioral pattern learning unit of the long short-term memory network LSTM;

[0088] A real-time detection unit to flag the behavior of multiple people picking up the same tool in a short period of time or moving the tool during non-working hours;

[0089] The alarm linkage unit triggers sound and light alarms and pushes alarm information including on-site images to the administrator.

[0090] For example, the anomaly analysis module uses a multi-level behavioral analysis architecture to achieve intelligent identification and handling of warehouse operation anomalies. It can use a long short-term memory network (LSTM) architecture to build a normal behavior baseline model based on 30 days of historical operation records. The training data covers the tool collection / return time series, operator qualification codes, tool type weight values, and spatial movement trajectories. The operation frequency features are extracted through the temporal convolution layer, and the attention mechanism is used to strengthen the learning of behavioral patterns during non-working time periods. The model is automatically incrementally trained every 24 hours, and the normal operation threshold range is dynamically updated (confidence interval ≥ 95%).

[0091] The real-time detection unit can analyze the behavior of video streams in real time, detect abnormal scenarios such as multi-person gatherings, and set three-level alarm rules:

[0092] (1) Level 1 warning: The same tool is claimed by ≥3 people within 10 minutes (based on RFID scan serial number comparison)

[0093] (2) Level 2 Alarm: Tool movement is detected during unauthorized periods (22:00-06:00) (combined with UWB positioning base station activity analysis)

[0094] (3) Level 3 emergency: The gravity-sensing shelf experiences a sudden change in weight within 15 seconds and the visual recognition fails to match the corresponding tool

[0095] Introducing spatiotemporal correlation analysis to jointly determine if UWB positioning drift (≥3 base station signal jumps) occurs in the operation trajectory and the RFID is not updated;

[0096] The alarm linkage unit can trigger the sound and light alarm, which can adopt a graded response mechanism:

[0097] (1) Level 1 warning: yellow strobe light + voice prompt

[0098] (2) Second level alarm: red flashing light + 80dB siren

[0099] (3) Level 3 emergency: Linked fire alarm system + access control forced lock

[0100] The alarm information push includes: event timestamp, digital twin model highlight positioning (3D coordinates + tool number), on-site video stream key frames (automatically capture 15 seconds before and after the abnormality occurs), operator biometric snapshot (the most recent face / fingerprint record that passed the access control); after the authorized personnel confirms the alarm through the mobile APP, the system automatically records the handling process video and links it to the work order system.

[0101] The visualization module includes:

[0102] A digital twin warehouse model based on the Unity engine displays tool location, status, and lifecycle progress in real time;

[0103] Heat map generation unit to analyze tool access frequency and optimize shelf layout;

[0104] Overdue marking units are not returned, and overdue tools are indicated by a flashing red light and a text message reminder is sent.

[0105] For example, a digital twin warehouse model based on the Unity engine can adopt a double-buffered rendering architecture to achieve real-time data-driven. The main thread of the model renders the three-dimensional scene at 60 frames per second, and an independent thread receives tool status data (positioning coordinates, battery voltage, number of uses, etc.) pushed by the edge computing module through the WebSocket protocol. In the tool status visualization coding, the green transparent outline: normal usable state, the yellow pulse effect: low power warning (remaining <20%), the red flashing mark: overdue (synchronized smart access control module borrowing and returning records), the life progress bar: integrated predictive maintenance module data, showing the remaining service life of consumables such as helmets / batteries; the thermal map generation unit integrates multi-source information through the data acquisition layer: RFID scanning frequency (average daily number of times each tool enters and exits the warehouse), UWB positioning trajectory (personnel movement hotspot area), gravity sensing shelf pressure change curve (tool access time distribution), and uses kernel density estimation to generate the initial thermal distribution. , identify high-frequency access areas through the DBSCAN clustering algorithm and optimize shelf layout using genetic algorithms, for example, use the fitness function to minimize walking distance (weight 60%) + tool relevance (weight 30%) + weight balance (weight 10%) to optimize shelf layout; for overdue marked units, multi-level alarms can be issued through overdue judgment logic, and corresponding text messages can be sent to remind, where the overdue judgment logic is: (1) Basic threshold: preset standard borrowing period of tool type (such as 72 hours for insulating gloves), (2) Dynamic adjustment: automatically extend 30% according to the priority of the work order (requires electronic approval by the project manager), (3) Holiday compensation: recalculate after deducting the length of statutory holidays, where the multi-level alarm mechanism is specifically as follows: Level 1 warning (overdue for 12 hours): tool icon yellow gradient fill, Level 2 alarm (overdue for 24 hours): digital twin model flashes red, Level 3 disposal (overdue for 48 hours): automatically lock the associated work order and push text message notification.

[0106] This embodiment also includes a predictive maintenance module:

[0107] For the helmet life prediction subunit, the Weibull distribution algorithm is used to calculate the number of times used and UV exposure data, and a replacement reminder is generated 30 days in advance;

[0108] For the battery monitoring subunit of power tools, the capacity attenuation rate is calculated based on the charge and discharge curves and temperature and humidity data, and a maintenance alarm is triggered when it is lower than 70%.

[0109] For example, the helmet life prediction subunit can collect data from a piezoelectric film sensor deployed on the helmet lining, which can record the number of impacts and cumulative force values in real time, as well as data from an ultraviolet light intensity sensor integrated on the brim. The unit then calculates the number of uses and ultraviolet exposure data using a Weibull distribution algorithm to generate a multi-level warning 30 days in advance, including a system notification on the 15th day, locking high-risk operation authorization on the 7th day, and mandatory replacement on the 3rd day.

[0110] The power tool battery monitoring subunit can analyze the charging and discharging characteristics and temperature and humidity data to calculate the decay rate of the power tool battery capacity, and trigger a maintenance alarm when it is lower than 70%. The battery capacity decay rate is calculated as follows: η = (C_new-C_current) / C_new×100%, where C_new is the initial capacity of the battery and C_current is the current actual available capacity of the battery. When η≥30%, it is determined that the battery health is lower than 70%, triggering a maintenance alarm.

[0111] The RFID and UWB collaborative positioning includes:

[0112] RFID readers scan tags in batches to generate lists when tools enter and leave the warehouse;

[0113] The UWB base station continuously tracks the location of tools in the warehouse and binds the coordinates of the shelf when the tools are stationary for more than 5 minutes;

[0114] When the UWB signal is lost, the latest RFID records and gravity sensing shelf data are called for comprehensive positioning.

[0115] For example, when the RFID reader scans the tags in batches to generate a list when the tools enter and leave the warehouse, it can read the tags in batches and automatically associate the specification parameters, calibration cycle and last maintenance record in the electronic file of the tools to improve the recognition efficiency of the tags; when it is detected that the tool has been stationary for more than 5 minutes, the system will automatically bind its coordinates to the grid coordinates of the nearest gravity-sensing shelf; in the signal loss compensation positioning stage, a multi-source data fusion engine can be built to start a three-level compensation mechanism when the UWB signal is interrupted (such as metal obstruction or electromagnetic interference): (1) call the RFID's most recent valid scanning record to generate a spatial position probability cloud map; (2) query the gravity-sensing shelf pressure sensor array and locate the shelf layer according to the tool weight distribution characteristics; 3) activate the AI camera visual positioning and use the improved YOLOv7 model to perform three-dimensional reconstruction and positioning of the tool feature points, so as to achieve accurate positioning of the tool with virtual and real fusion.

[0116] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent tool warehouse management system, characterized in that: The system comprises: Multimodal sensing module, intelligent access control module, edge computing module, anomaly analysis module and visualization module; The multimodal sensing module uses RFID and UWB dual-mode positioning technology, visual recognition technology, and sensor fusion detection technology to perform tool identity recognition, real-time positioning, and integrity verification; The intelligent access control module includes a dynamic permission control unit and a gravity-sensing shelf for tool access authorization and access behavior recording based on personnel qualifications; The edge computing module includes a localized edge gateway and a low-power communication unit for real-time data processing and offline cache synchronization; The anomaly analysis module uses a machine learning model to identify illegal operations and trigger an alarm; The visualization module uses digital twin technology to display tool status and warehouse heat map.

2. The system according to claim 1, wherein: The multimodal sensing module includes: RFID tags embedded in the tool surface are used to identify the status of tools entering and leaving the warehouse in batches; The UWB positioning base station deployed in the warehouse uses the Time Difference of Arrival (TDOA) algorithm to achieve centimeter-level positioning and trajectory tracking of the tool; AI cameras and infrared curtain sensors are installed at the entrances and exits. When the infrared curtain sensor detects that a tool is approaching, it triggers the AI camera to capture the tool image. The convolutional neural network (CNN) is used to identify the tool's shape, color, and surface damage, and compares it with the pre-stored image database to verify the tool's integrity.

3. The system according to claim 1, wherein: The intelligent access control module includes: A dynamic permission control unit linked to the work order management system generates a list of borrowable tools based on employee qualifications; Biometric access control terminals, used to verify facial or fingerprint information and unlock authorized shelves; The gravity-sensing shelf integrates a pressure sensor array. When a tool is removed, the pressure sensor detects the weight change and compares it with the pre-stored tool weight threshold, automatically recording the tool name, removal time and operator information; the weight difference is checked again when the tool is returned. If the weight deviation is detected to be more than 5%, the accessory is determined to be missing and an alarm is triggered.

4. The system according to claim 1, wherein: The edge computing module also includes: A local data processing unit deployed on the NVIDIA Jetson edge device to interpret sensor and camera data in real time; A disconnected cache unit stores operation records during network interruptions and synchronizes them to the cloud after network recovery; Tags using LoRaWAN or NB-IoT communication have a battery life of no less than 3 years, and the warehouse gateway supports solar power supply.

5. The system according to claim 1, wherein: The abnormality analysis module includes: Establish a normal operation baseline based on the behavioral pattern learning unit of the long short-term memory network LSTM; A real-time detection unit to flag the behavior of multiple people picking up the same tool in a short period of time or moving the tool during non-working hours; The alarm linkage unit triggers sound and light alarms and pushes alarm information including on-site images to the administrator.

6. The system according to claim 1, wherein: The visualization module includes: A digital twin warehouse model based on the Unity engine displays tool location, status, and lifecycle progress in real time; Heat map generation unit to analyze tool access frequency and optimize shelf layout; Overdue marking units are not returned, and overdue tools are indicated by a flashing red light and a text message reminder is sent.

7. The system according to claim 1, wherein: Also includes predictive maintenance modules: For the helmet life prediction subunit, the Weibull distribution algorithm is used to calculate the number of times used and UV exposure data, and a replacement reminder is generated 30 days in advance; For the battery monitoring subunit of power tools, the capacity attenuation rate is calculated based on the charge and discharge curves and temperature and humidity data, and a maintenance alarm is triggered when it is lower than 70%.

8. The system according to claim 2, wherein: The RFID and UWB collaborative positioning includes: RFID readers scan tags in batches to generate lists when tools enter and leave the warehouse; The UWB base station continuously tracks the location of tools in the warehouse and binds the coordinates of the shelf when the tools are stationary for more than 5 minutes; When the UWB signal is lost, the latest RFID records and gravity sensing shelf data are called for comprehensive positioning.

Images