Intelligent warehousing system and method for full-life-cycle management of numerical control tools

Abstract

The invention discloses an intelligent warehousing system for full-life-cycle management of numerical control tools, and relates to the technical field of tool management. The intelligent storage module adopts RFID and binocular vision fusion positioning, and efficient storage is achieved in cooperation with a multi-layer shuttle vehicle and an intelligent goods shelf; the tool management module constructs a digital twinborn model and predicts the service life of a tool by using an LSTM-GRU network; the inventory optimization module applies an algorithm to predict the demand and dynamically adjust the inventory; the quality traceability module realizes data storage and second-level traceability based on a block chain; the decision support module realizes tool type selection and process optimization through multi-attribute decision; the system further comprises an energy consumption management extension module, a man-machine interaction extension module and the like, and the whole-process intelligent management of tools from warehousing, storage, use to scrapping is achieved. The full-life-cycle intelligent management of the numerical control cutter is realized, the storage space utilization rate, the cutter service life prediction precision and the inventory turnover rate are improved through multi-technology fusion and algorithm optimization, and the competitiveness of an enterprise in the numerical control machining field is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of tool management, and in particular to an intelligent storage system and method for full life cycle management of CNC tools. Background Art

[0002] In the field of modern CNC machining, CNC tools are core consumables, and their full lifecycle management directly affects machining accuracy, production costs, and production efficiency. Traditional tool management models rely on manual registration, empirical judgment, and extensive inventory management, which has many drawbacks. For example, tool in and out information is manually entered into the ERP system, which is prone to model confusion and quantity errors. The wear status of tools requires operators to stop the machine regularly for inspection, which makes it difficult to detect early abnormal wear, resulting in fluctuations in machining quality or sudden chipping accidents. Inventory management often uses fixed-cycle replenishment, which cannot cope with order fluctuations and process changes, often resulting in stagnant inventory backlogs or shortages of key tools.

[0003] With the development of intelligent manufacturing, some companies have introduced RFID technology for tool positioning or sensor-based cutting data collection. However, these systems are mostly independent modules, lacking deep data integration and collaboration. For example, single RFID positioning cannot accurately identify tool posture, affecting storage space utilization. Sensor data is only used for real-time monitoring and is not integrated with tool life prediction and process optimization, making proactive management difficult. Furthermore, existing tool quality traceability systems are mostly based on centralized databases, making data susceptible to tampering or loss, and unable to meet the stringent quality traceability requirements of industries such as aerospace and automotive manufacturing.

[0004] When it comes to inventory optimization, the traditional economic order quantity (EOQ) model assumes stable demand and is unable to adapt to the dynamic changes in actual production. Companies often fail to account for factors such as seasonal order fluctuations and supplier delivery delays, resulting in low inventory turnover and high warehousing costs. Furthermore, tool scrapping decisions rely on manual experience and lack a quantitative assessment of the tool's residual value and overall cost, resulting in wasted resources. Therefore, an integrated, intelligent tool lifecycle management system is urgently needed to address issues such as data silos, management lags, and resource waste. Summary of the Invention

[0005] The present invention proposes an intelligent storage system and method for the full life cycle management of CNC tools to solve the problems mentioned in the above-mentioned prior art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent storage system for the full life cycle management of CNC tools, comprising:

[0007] Warehousing module: Utilizing UHF RFID and binocular vision fusion positioning technology, an automatic storage system for shuttle vehicles is deployed on each floor. Shelves are equipped with integrated weight sensors and vibration monitoring modules to achieve dynamic inventory update delays. A tool posture adaptive adjustment mechanism is introduced to automatically identify tool posture and adjust it to the optimal storage posture based on the PointNet++ point cloud processing algorithm.

[0008] Tool Management Module: Constructs a digital twin model of the tool, encompassing three dimensions: geometry, physics, and behavior. It predicts the remaining tool life based on an LSTM-GRU hybrid neural network with 128 nodes in the input layer, 64 nodes in the hidden layer, and 3 nodes in the output layer. It also develops a tool sharpening management subsystem. It also innovatively introduces a tool material microstructure detection unit to monitor the coating peeling status of coated tools in real time.

[0009] Inventory optimization module: Use the XGBoost algorithm to predict tool demand, with a tree depth of 6, a learning rate of 0.1, and a number of 100 trees, combined with a dynamic safety stock formula Constructing a target optimization model to solve the economic order quantity Based on the dynamic calculation of remaining life; where β is the seasonal fluctuation coefficient, γ is the process change impact coefficient, and φ is the tool value coefficient;

[0010] Quality traceability module: A tool data notarization system is built based on the Hyperledger Fabric consortium chain, with 5 consensus nodes and 3 sorting nodes. The PBFT consensus algorithm is used to establish a causal relationship model for machining quality, including 12 input variables, 5 intermediate variables, and 3 output variables. The SHAP value is used to analyze the influence of tool parameters on workpiece quality. A digital twin subsystem of the machining process based on the Unity engine is developed to simulate the tool cutting state in real time and predict machining errors.

[0011] Decision support module: Develop tool cost analysis engine, including sensitivity analysis of various factors and break-even analysis; calculate weights using entropy weight method based on multi-attribute decision theory e i The entropy value is combined with the AHP judgment matrix consistency ratio CR < 0.1, and the closeness C is calculated by TOPSIS i Decision-making tool selection and output of process parameter optimization suggestions; introduction of process parameter-tool life Pareto optimization, based on the NSGA-II algorithm, population size N = 100, number of iterations G = 200, optimization objective function: Where x is the process parameter vector, and the Pareto solution set is updated every minute to balance efficiency and life.

[0012] Furthermore, the storage module includes an adaptive environmental control subsystem, an anti-collision scheduling subsystem, an heuristic function h(n)=d(n)+w×Δv(n), where d(n) is the Euclidean distance, w is the weight coefficient 0.3, Δv(n) is the speed difference, and a tool state detection subsystem, and the machine vision is based on YOLOv8; the tool management module includes a submodule that integrates three-axis vibration, current, and acoustic emission sensor data based on DS evidence theory, and a tool life prediction submodule to establish a multi-parameter coupling prediction model L=f(Fc,v,f,t,T), where Fc is the main cutting force, v is the cutting speed, f is the feed rate, t is the cutting time, and T is the cutting temperature; the tool scheduling submodule optimizes the tool distribution path based on an improved genetic algorithm.

[0013] Furthermore, the inventory optimization module includes an inventory early warning subsystem, which sets dynamic early warning thresholds, with the green threshold ≥ safety stock × 120%, the yellow threshold being safety stock × 80% - 120%, and the red threshold being < safety stock × 80%, a supplier collaborative replenishment subsystem, and a tool allocation optimization subsystem; the quality traceability module includes a processing process data acquisition subsystem, a quality feature extraction subsystem, and a quality anomaly traceability subsystem.

[0014] Furthermore, the decision support module includes a tool cost prediction submodule, a process parameter optimization submodule, and a supplier performance evaluation submodule, and establishes a dynamic evaluation index system. The evaluation cycle can be configured to be 1-30 days, and the evaluation algorithm adopts the fuzzy hierarchical analysis method.

[0015] Furthermore, it also includes:

[0016] Energy consumption management module: Real-time monitoring of storage equipment energy consumption, dynamic adjustment of equipment power based on reinforcement learning, development of eyewear equipment auxiliary system through human-computer interaction module, support for tool assembly guidance, abnormal alarm visualization, system integration module provides standardized RESTful API interface, and supports docking with MES, ERP, and PLM systems.

[0017] Furthermore, the adaptive environmental control subsystem in the storage module adopts a PID controller to adjust temperature and humidity, with a proportional coefficient of 0.5, an integral time of 10s, and a differential time of 5s; the anti-collision scheduling subsystem adopts a time window algorithm TWAA to plan the AGV path.

[0018] Furthermore, the tool life prediction submodule in the tool management module adopts a parameter coupling prediction model and calculates the importance weight of each parameter through the random forest algorithm (Fc is 0.35, v is 0.25, f is 0.2, T is 0.15, and t is 0.05). The tool scheduling submodule optimizes the tool distribution path based on the improved genetic algorithm, with a population size of 100, a crossover probability of 0.8, a mutation probability of 0.1, and a number of elites retained of 5.

[0019] A method for applying the intelligent storage system for full life cycle management of CNC tools, comprising:

[0020] Tool warehousing management steps: RFID and QR code dual identification are used to verify tool information, which is automatically compared with purchase orders. Machine vision is used to detect tool appearance defects, and a laser measuring instrument is used to detect geometric parameters. The optimal storage location is assigned based on the K-means clustering algorithm, with 5 clusters, 50 iterations, and Euclidean distance as the distance metric. Spectral analysis of tool materials is introduced to automatically identify the composition and uniformity of tool coating materials.

[0021] Tool storage management steps: Real-time monitoring of storage temperature, humidity, air pressure, and lighting environment parameters automatically adjusts to optimal storage conditions. RFID mobile terminals combined with UWB positioning technology are used for dynamic inventory. A tool status monitoring mechanism is established to automatically trigger maintenance procedures for expired tools. Innovative micro-vibration excitation technology is used to periodically detect the fixed status of tools.

[0022] Tool outbound management steps: Receive processing task orders, match the optimal tool combination based on the ant colony algorithm, trigger AGV scheduling instructions, and execute the path planning algorithm. The number of ants is 50, the pheromone volatility coefficient is 0.5, the heuristic factor is 2, and the pheromone factor is 1.5. Tool status is confirmed during outbound verification, abnormal tools are automatically intercepted and a processing work order is generated; tool pre-balancing technology is introduced to automatically adjust tool dynamic balance;

[0023] Tool usage monitoring steps: Real-time collection of current, vibration, temperature, and acoustic emission data during the machining process, and judgment of tool wear status based on the Transformer network. The encoder has 4 layers, the decoder has 4 layers, the number of heads is 8, and the hidden dimension is 256. When the wear exceeds the dynamic threshold, a three-level warning is triggered; an online cutting force monitoring unit is developed to calculate the cutting force coefficient Kc = Fc / (ap×f) in real time, where Fc is the main cutting force, ap is the back cutting amount, and f is the feed rate. When Kc fluctuates by more than ±15%, the cutting parameters are automatically adjusted;

[0024] Tool grinding management steps: Automatically generate a grinding work order when the grinding threshold is met, use a 3D scanner to record changes in tool geometric parameters before and after grinding, and update the tool life model; introduce laser cladding repair technology to repair the coating of carbide tools;

[0025] Tool scrap management steps: Determine scrapping based on the attribute decision model, execute the scrapping process including physical destruction and system marking to generate improvement suggestion reports, optimize tool selection and usage strategies; develop a tool material recovery subsystem, and realize tool material recovery through plasma gasification technology.

[0026] Furthermore, the tool usage monitoring step includes establishing a variable nonlinear regression model of tool wear and cutting parameters. Where ΔVb is the flank wear, Fc is the main cutting force, v is the cutting speed, f is the feed rate, ae is the cutting width, and the model parameters k = 0.001, m = 0.3, n = 0.5, p = 0.2, q = 0.1. A federated learning algorithm is used to integrate workshop tool wear data. When the predicted remaining life is lower than the safety threshold, a tool replacement warning is automatically triggered and an alternative tool is recommended. Cutting vibration modal analysis is introduced to identify cutting chatter signs in real time.

[0027] Furthermore, it also includes:

[0028] Decision support steps: Based on historical processing data, a GNN graph neural network model is trained to predict tool failure modes, with 100 nodes, 300 edges, 64 hidden layer dimensions, and 2 aggregation layers. The economic order quantity is dynamically calculated using the formula EOQ*=EOQ×(1+α), where α is the supply risk coefficient. Dynamic adjustments are made based on the supplier's historical performance, and ABC-VED joint classification management is implemented to establish a tool life cycle cost model that includes four dimensions: procurement cost, usage cost, maintenance cost, and scrap cost to evaluate the economic feasibility of different tool solutions; a process parameter recommendation system is developed to automatically generate the optimal cutting parameter combination by interacting with the digital twin model.

[0029] Compared with the existing technology, the beneficial effects of the present invention are:

[0030] At the intelligent warehousing level, RFID and binocular vision fusion positioning technology achieve millimeter-level precise identification of tools, and cooperate with the tool posture adaptive adjustment mechanism to increase the storage space utilization rate from the traditional 65% to 85%; the AGV cluster adopts a dynamic path planning algorithm, reducing the empty driving rate to below 10%, greatly shortening the time for tools to enter and exit the warehouse.

[0031] In the tool management module, a life prediction model based on an LSTM-GRU hybrid neural network keeps the error rate in tool remaining life predictions below 5%, an improvement of over 60% compared to traditional methods. Combining real-time sensor data with digital twin technology, the system provides early warning of abnormalities such as tool wear and chipping, avoiding unexpected downtime. This reduces the response time for machining anomalies from 30 minutes to 3 minutes, significantly improving production continuity.

[0032] In terms of inventory optimization, the combination of the XGBoost algorithm and a dynamic safety stock model has achieved a 92% accuracy rate in tool demand forecasting, increasing inventory turnover from 6 to 10 times per year and reducing inventory costs by over 25%. Blockchain technology enables the storage of complete tool data throughout the entire process, ensuring that quality traceability information is tamper-proof and can be retrieved within seconds, meeting the stringent quality control requirements of the high-end manufacturing industry. Furthermore, a multi-objective decision-making algorithm assists in tool selection and process optimization, extending tool life by 30% and reducing unit processing costs by 22%-28%, effectively enhancing the company's core competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a schematic block diagram of an intelligent storage system for full life cycle management of CNC tools proposed by the present invention;

[0034] Figure 2 This is a schematic block diagram of an intelligent storage method for the full life cycle management of CNC tools proposed by the present invention;

[0035] Figure 3 A schematic diagram comparing inventory turnover rates of an intelligent warehousing method for full life cycle management of CNC tools proposed in the present invention;

[0036] Figure 4 This is a schematic diagram comparing the storage space utilization of an intelligent storage method for the full life cycle management of CNC tools proposed in the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0038] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0039] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the accompanying drawings.

[0040] Reference Figures 1 to 4 :Specific implementation of an intelligent storage system and method for full life cycle management of CNC tools

[0041] The intelligent storage system and method for the full life cycle management of CNC tools described in this application achieves refined management of the entire tool process through deep software and hardware integration and multi-algorithm collaboration. Specific implementation details are as follows:

[0042] 1. Implementation method of intelligent warehousing module

[0043] The system utilizes a fusion of UHF RFID (902-928 MHz frequency band) and binocular vision (Baslerace acA2500-14gm camera, 2592×1944 resolution, ±0.3mm positioning accuracy) for positioning. High-temperature resistant RFID tags (Alien Higgs-3 chip, read range 0-3 meters) are affixed to the tool surfaces. Fixed industrial cameras are installed on the shelf levels, using the OpenCV library and the YOLOv8 deep learning algorithm to identify tool model and location. The multi-layer shuttle system is equipped with 12 aisle stacker cranes (controlled by Siemens S7-1500 PLCs, horizontal operating speed 240 m / min, vertical lifting speed 60 m / min, positioning accuracy ±5 mm). Laser ranging sensors (SICK LMS511) provide real-time position calibration. The smart shelf integrates an HBMPW20AC3M1 / 500KG weight sensor (0-500kg range, ±0.1% FS accuracy) and a PCB352C04 vibration monitoring module (20kHz sampling rate, 1-8kHz frequency response), collecting data every two minutes and uploading it to an edge computing gateway (Advantech ARK-3500). The tool posture adaptive adjustment mechanism uses a six-degree-of-freedom collaborative robot arm (UR10e, with ±0.05mm repeatability) and uses the PointNet++ point cloud processing algorithm to identify tool posture. When a tilt angle greater than 5° is detected, the robot arm's end effector (SCHUNK PGN-plus) automatically grasps and adjusts to the optimal storage posture, increasing warehouse space utilization from the traditional 65% to 85%.

[0044] 2. Tool Management Module Implementation Method

[0045] The digital twin model of the tool is built based on the Unity3D engine, integrating the geometric parameters collected by the laser measuring instrument (Renishaw XL-80, accuracy ±0.005mm), the physical properties (density, hardness, etc.) provided by the material supplier, and the real-time processing data (cutting force, temperature). The LSTM-GRU hybrid neural network is deployed on the NVIDIA Jetson AGX Xavier edge server. The input layer receives 128-dimensional sensor data (including spindle current, three-axis vibration acceleration, cutting temperature, etc.), and outputs the predicted value of the tool's remaining life after processing by the 64-node hidden layer. The model training uses 80% of the historical processing data as the training set, 10% as the validation set, and 10% as the test set. It is iterated 500 times through the Adam optimizer (learning rate 0.001), and the measured prediction error rate is less than 5%. The tool sharpening management subsystem is equipped with a KEYENCEL K-G80 laser measuring instrument with a wavelength of 635nm (scanning speed 5000 points / second), combined with a Horiba Lab RAM H REvolution Raman spectrometer (resolution 2cm -1) Detect coating thickness and composition. When the coating peeling area is greater than 10% or the thickness is less than 2μm, a grinding work order is automatically generated. The accuracy of the tool life prediction model is compared in the following table:

[0046]

[0047]

[0048] 3. Inventory Optimization Module Implementation Method

[0049] The XGBoost algorithm (tree depth 6, learning rate 0.1, number of trees 100) was used to predict tool demand. The training data included 20 feature variables such as historical orders, equipment load, and process changes. Feature engineering and model training were performed using Python's Scikit-learn library, achieving a prediction accuracy of 92%. Dynamic safety stock formula In the equation, Z is the service level coefficient (1.645 at 95% confidence level), is the average daily demand, σ d is the standard deviation of demand, is the average lead time, σ L = is the standard deviation of the lead time; β is the seasonal fluctuation coefficient extracted through Fourier transform; γ is the process change impact coefficient) manually input by the process department based on the new product introduction plan, which increases the inventory turnover rate from 6 times / year to 10 times / year. The tool value coefficient φ is introduced into the economic order quantity (EOQ) calculation, and the formula is: Where D is the annual demand, S is the cost of each order, H is the unit storage cost, d is the demand rate, and p is the supply rate. When the remaining life of the tool is less than 30%, φ takes the value of 0.8 to reduce the single purchase quantity and reduce obsolete inventory.

[0050] 4. Implementation Methods of Quality Traceability Module

[0051] A consortium chain was built based on Hyperledger Fabric, with 5 consensus nodes (Dell PowerEdge R740 server, Intel Xeon Gold 6240 CPU, 32GB memory) and 3 sorting nodes deployed, and the PBFT consensus algorithm was used to achieve data synchronization in seconds. Each tool transaction record contains the tool ID, operation time, device number and SHA-256 hash value to ensure that the data cannot be tampered with. The machining quality causal relationship model uses Python's SHAP library to analyze the influence of 12 input variables (tool model, cutting speed, feed rate, etc.) on the surface roughness of the workpiece. For example, the weight of the cutting speed on the roughness is 0.35. The digital twin subsystem is based on the Unity3D physics engine, synchronizes machining parameters at a frequency of 100Hz, simulates the cutting process through finite element analysis (FEA), predicts machining errors (accuracy of ±0.01mm), and warns of dimensional deviation risks in advance.

[0052] 5. Decision support module implementation method

[0053] The tool cost analysis engine is developed based on Python's pandas and numpy libraries, and supports single-factor sensitivity analysis (such as the impact of raw material price fluctuations on costs) and multi-factor analysis (combination parameter changes). The process parameter optimization module uses the NSGA-II algorithm (population size 100, number of iterations 200, crossover probability 0.8, mutation probability 0.1) to simultaneously optimize processing efficiency and tool life. The generated Pareto solution set covers 20 groups of parameter combinations for engineers to choose from. Supplier performance evaluation is based on the fuzzy analytic hierarchy process (FAHP) and is evaluated from three dimensions: quality (weight 0.4), delivery time (0.3), and service (0.3). The quality dimension includes five sub-indicators such as tool qualification rate and batch stability. An evaluation report is generated every quarter, and suppliers with a score of less than 70 points are eliminated.

[0054] 6. Method implementation steps

[0055] Tool inventory management: Workers use a Zebra DS3608 handheld RFID reader to scan tool tags, which the system automatically compares to purchase orders in the SAP ERP system. A Keyence VR-3200 3D laser measuring instrument is used to measure tool dimensions (accuracy ±0.003mm). Tool locations are assigned using a K-means algorithm (number of clusters: 5, distance metric: Euclidean distance). A Raman spectrometer simultaneously analyzes the tool coating. If the TiAlN coating thickness is detected to be less than 2μm, the tool is deemed defective and an audible and visual alarm is triggered, preventing the tool from entering the warehouse.

[0056] Tool storage management: A temperature and humidity sensor (SHT35) collects environmental data every 10 minutes. When the temperature exceeds 28°C or the humidity exceeds 60% RH, the PLC controller automatically activates air conditioning and dehumidification. A UWB positioning module (Decawave DW1000 chip) tracks tool positions in real time, performs a dynamic inventory every morning, and generates a discrepancy report. A micro-vibration excitation device (vibration frequency 50Hz, amplitude 0.5μm) monitors tool fixation weekly. If the vibration spectrum shows abnormal peaks, an audible and visual alarm is triggered and an abnormal work order is issued.

[0057] Tool delivery management: Based on work order requirements, the system uses an ant colony algorithm (ant population 50, pheromone volatility coefficient 0.5, heuristic factor 2, pheromone factor 1.5) to match the optimal tool combination. An AGV (laser-guided, maximum load 100 kg) executes the D*Lite path planning algorithm and verifies tool information via RFID access control before delivery. A pre-balancing device (Schenck dynamic balancing machine from Germany) dynamically balances tools with speeds greater than 10,000 rpm to ensure residual imbalance is less than 5 g·mm / kg. Otherwise, the tool returns to the resharpening process.

[0058] Tool usage monitoring: The spindle motor current sensor (ACS712) and the triaxial vibration sensor (PCB352C65) collect data at a sampling rate of 10kHz, and the wear status is analyzed in real time through the Transformer network. If the fluctuation exceeds ±15% (where Fc is the main cutting force, ap is the depth of cut, and f is the feed rate), the system automatically adjusts the feed rate to prevent chipping. If the predicted remaining life is less than the safety threshold (20% of the initial life), a three-level warning is triggered (yellow pop-up window, orange text message, and red audio and visual alarm).

[0059] Tool regrinding management: When the regrinding threshold is reached (e.g., flank wear > 0.3mm), the system generates a regrinding work order, recommending a grinding wheel grit size (#80), feed rate (0.05mm / r), and regrinding depth (0.01mm). Laser cladding equipment (IPG Photonics laser, maximum power 2000W) repairs the tool. After repair, dimensional accuracy is verified using a 3D scanner. Defective products undergo a secondary regrinding process (up to three times). If they still fail, they are marked as scrapped.

[0060] Tool scrap management: A multi-attribute decision-making model considers wear (weight 0.4), number of regrindings (0.3), and machining quality (0.3) to determine scrapping. A combined score of less than 60 triggers the scrapping process. A plasma gasification system (100kW) processes scrapped tools, recovering carbide powder at a recovery rate of 96%. The tool status is marked as "scrapped" in the blockchain system.

[0061] 7. Data effect verification

[0062]

[0063] This application increases inventory turnover by 66.7% and significantly reduces capital occupation through dynamic demand forecasting and intelligent inventory algorithms; tool loss rate is reduced by 62.5%, attributed to real-time wear monitoring and preventive maintenance strategies. The response time for machining anomalies is shortened from 30 minutes to 3 minutes, relying on the rapid root cause location capability of blockchain traceability and digital twin technology; tool life is extended by 30%, thanks to the adaptive adjustment of process parameters and optimization of grinding strategies. Storage space utilization is increased by 30.8%, mainly achieved through automatic adjustment of tool posture and intelligent storage location allocation. These improvements verify the system's significant advantages in cost reduction, efficiency improvement and quality control, and provide a highly reliable management solution for the CNC machining industry.

[0064] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. An intelligent storage system for the full life cycle management of CNC tools, characterized by: include: Warehousing module: Utilizing UHF RFID and binocular vision fusion positioning technology, an automatic storage system for shuttle vehicles is deployed on each floor. Shelves are equipped with integrated weight sensors and vibration monitoring modules to achieve dynamic inventory update delays. A tool posture adaptive adjustment mechanism is introduced to automatically identify tool posture and adjust it to the optimal storage posture based on the PointNet++ point cloud processing algorithm. Tool Management Module: Builds a digital twin model of the tool, including the three dimensions of geometry, physics, and behavior. It predicts the remaining life of the tool based on an LSTM-GRU hybrid neural network with 128 nodes in the input layer, 64 nodes in the hidden layer, and 3 nodes in the output layer. It also develops a tool sharpening management subsystem. Innovatively introduce a tool material microstructure detection unit to monitor the coating peeling status of coated tools in real time; Inventory optimization module: Use the XGBoost algorithm to predict tool demand, with a tree depth of 6, a learning rate of 0.1, and a number of 100 trees, combined with a dynamic safety stock formula Constructing a target optimization model to solve the economic order quantity Based on the dynamic calculation of remaining life; where β is the seasonal fluctuation coefficient, γ is the process change impact coefficient, and φ is the tool value coefficient; Quality traceability module: A tool data notarization system is built based on the Hyperledger Fabric consortium chain, with 5 consensus nodes and 3 sorting nodes. The PBFT consensus algorithm is used to establish a causal relationship model for machining quality, including 12 input variables, 5 intermediate variables, and 3 output variables. The SHAP value is used to analyze the influence of tool parameters on workpiece quality. A digital twin subsystem of the machining process based on the Unity engine is developed to simulate the tool cutting state in real time and predict machining errors. Decision support module: Develop tool cost analysis engine, including sensitivity analysis of various factors and break-even analysis; calculate weights using entropy weight method based on multi-attribute decision theory e i The entropy value is combined with the AHP judgment matrix consistency ratio CR < 0.1, and the closeness C is calculated by TOPSIS i Decision-making on tool selection and output of process parameter optimization suggestions; The process parameter-tool life Pareto optimization is introduced based on the NSGA-II algorithm, with a population size of N = 100 and the number of iterations G = 200. The optimization objective function is: Where x is the process parameter vector, and the Pareto solution set is updated every minute to balance efficiency and life.

2. The intelligent storage system and method for full life cycle management of CNC tools according to claim 1 is characterized in that: The storage module includes an adaptive environmental control subsystem, an anti-collision scheduling subsystem, an heuristic function h(n)=d(n)+w×Δv(n), where d(n) is the Euclidean distance, w is the weight coefficient 0.3, Δv(n) is the speed difference, and a tool state detection subsystem, and the machine vision is based on YOLOv8; the tool management module includes a submodule that integrates three-axis vibration, current, and acoustic emission sensor data based on DS evidence theory, and a tool life prediction submodule to establish a multi-parameter coupling prediction model L=f(Fc,v,f,t,T), where Fc is the main cutting force, v is the cutting speed, f is the feed rate, t is the cutting time, and T is the cutting temperature; the tool scheduling submodule optimizes the tool distribution path based on an improved genetic algorithm.

3. The intelligent storage system and method for full life cycle management of CNC tools according to claim 1 is characterized in that: The inventory optimization module includes an inventory early warning subsystem, which sets dynamic early warning thresholds, with the green threshold ≥ safety stock × 120%, the yellow threshold being safety stock × 80% - 120%, and the red threshold being < safety stock × 80%, a supplier collaborative replenishment subsystem, and a tool allocation optimization subsystem; the quality traceability module includes a processing process data acquisition subsystem, a quality feature extraction subsystem, and a quality anomaly traceability subsystem.

4. The intelligent storage system for the full life cycle management of CNC tools according to claim 1 is characterized in that: The decision support module includes a tool cost prediction submodule, a process parameter optimization submodule, and a supplier performance evaluation submodule. A dynamic evaluation index system is established, and the evaluation cycle can be configured to be 1-30 days. The evaluation algorithm adopts the fuzzy hierarchical analysis method.

5. The intelligent storage system for the full life cycle management of CNC tools according to claim 1 is characterized in that: Also includes: Energy consumption management module: Real-time monitoring of storage equipment energy consumption, dynamic adjustment of equipment power based on reinforcement learning, development of eyewear equipment auxiliary system through human-computer interaction module, support for tool assembly guidance, abnormal alarm visualization, system integration module provides standardized RESTful API interface, and supports docking with MES, ERP, and PLM systems.

6. The intelligent storage system for the full life cycle management of CNC tools according to claim 1 is characterized in that: The adaptive environmental control subsystem in the storage module adopts a PID controller to adjust temperature and humidity, with a proportional coefficient of 0.5, an integral time of 10s, and a differential time of 5s; the anti-collision scheduling subsystem adopts a time window algorithm TWAA to plan the AGV path.

7. The intelligent storage system for the full life cycle management of CNC tools according to claim 1 is characterized in that: The tool life prediction submodule in the tool management module adopts a parameter coupling prediction model and calculates the importance weight of each parameter through the random forest algorithm (Fc is 0.35, v is 0.25, f is 0.2, T is 0.15, and t is 0.05). The tool scheduling submodule optimizes the tool distribution path based on an improved genetic algorithm with a population size of 100, a crossover probability of 0.8, a mutation probability of 0.1, and a number of elites retained of 5.

8. A method for applying the intelligent storage system for full life cycle management of CNC tools according to any one of claims 1 to 7, characterized in that: include: Tool warehousing management steps: RFID and QR code dual identification are used to verify tool information, which is automatically compared with purchase orders. Machine vision is used to detect tool appearance defects, and a laser measuring instrument is used to detect geometric parameters. The optimal storage location is assigned based on the K-means clustering algorithm, with 5 clusters, 50 iterations, and Euclidean distance as the distance metric. Spectral analysis of tool materials is introduced to automatically identify the composition and uniformity of tool coating materials. Tool storage management steps: Real-time monitoring of storage temperature, humidity, air pressure, and lighting environment parameters automatically adjusts to optimal storage conditions. RFID mobile terminals combined with UWB positioning technology are used for dynamic inventory. A tool status monitoring mechanism is established to automatically trigger maintenance procedures for expired tools. Innovative micro-vibration excitation technology is used to periodically detect the fixed status of tools. Tool outbound management steps: Receive processing task orders, match the optimal tool combination based on the ant colony algorithm, trigger AGV scheduling instructions, and execute the path planning algorithm. The number of ants is 50, the pheromone volatility coefficient is 0.5, the heuristic factor is 2, and the pheromone factor is 1.

5. Tool status is confirmed during outbound verification, abnormal tools are automatically intercepted and a processing work order is generated; tool pre-balancing technology is introduced to automatically adjust tool dynamic balance; Tool usage monitoring steps: Real-time collection of current, vibration, temperature, and acoustic emission data during machining. Tool wear status is determined based on a Transformer network. The encoder has 4 layers, the decoder has 4 layers, the number of heads is 8, and the hidden dimension is 256. When the wear exceeds the dynamic threshold, a three-level warning is triggered. Developed an online cutting force monitoring unit to calculate the cutting force coefficient Kc = Fc / (ap×f) in real time, where Fc is the main cutting force, ap is the depth of cut, and f is the feed rate. Automatically adjust cutting parameters when Kc fluctuates by more than ±15%. Tool grinding management steps: Automatically generate a grinding work order when the grinding threshold is met, use a 3D scanner to record changes in tool geometric parameters before and after grinding, and update the tool life model; introduce laser cladding repair technology to repair the coating of carbide tools; Tool scrap management steps: Determine scrapping based on the attribute decision model, execute the scrapping process including physical destruction and system marking to generate improvement suggestion reports, optimize tool selection and usage strategies; develop a tool material recovery subsystem, and realize tool material recovery through plasma gasification technology.

9. The method for managing the entire life cycle of a CNC tool according to claim 8, wherein: The tool usage monitoring step includes establishing a variable nonlinear regression model of tool wear and cutting parameters Where ΔVb is the flank wear, Fc is the main cutting force, v is the cutting speed, f is the feed rate, ae is the cutting width, and the model parameters k = 0.001, m = 0.3, n = 0.5, p = 0.2, q = 0.

1. A federated learning algorithm is used to integrate workshop tool wear data. When the predicted remaining life is lower than the safety threshold, a tool replacement warning is automatically triggered and an alternative tool is recommended. Cutting vibration modal analysis is introduced to identify cutting chatter signs in real time.

10. The method for managing the entire life cycle of a CNC tool according to claim 8, wherein: Also includes: Decision support steps: Based on historical processing data, a GNN graph neural network model is trained to predict tool failure modes. The number of nodes is 100, the number of edges is 300, the hidden layer dimension is 64, the number of aggregation layers is 2, and the economic order quantity is dynamically calculated using the formula EOQ. * =EOQ×(1+α), where α is the supply risk coefficient. It is dynamically adjusted based on the supplier's historical performance, and ABC-VED joint classification management is implemented to establish a tool life cycle cost model including procurement cost, usage cost, maintenance cost, and scrap cost to evaluate the economic feasibility of different tool solutions; a process parameter recommendation system is developed to automatically generate the optimal cutting parameter combination through interaction with the digital twin model.

Images