Intelligent tool management method and system

Abstract

The present invention relates to the technical field of intelligent tool management, and discloses an intelligent tool management method and system. The method comprises: collecting and encoding information of multiple tools and establishing a tool database; performing spatial planning on a tool cabinet to obtain a hierarchical storage scheme and three-dimensional coordinate position information; performing multimodal recognition and authority allocation to obtain a user identity recognition result and tool use authority data; when a user requests to borrow a tool, performing verification to generate a tool borrowing instruction and borrowing record data; when a user requests to return a tool, performing status detection on the returned tool to obtain current status information of the tool, and inputting the current status information of the tool into a predictive maintenance model for life assessment to obtain predictive maintenance data; performing multi-dimensional analysis on the borrowing record data and the predictive maintenance data to generate a tool use heat map and a target tool management strategy. The present invention improves the intelligent level of tool management.

Description

Technical Field

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

[0002] Tool management has always been an important and complex issue. With the expansion of enterprise scale and the complexity of processes, the types and quantities of tools continue to increase, and traditional manual management methods can no longer meet the needs of efficient and accurate management. The borrowing, returning, maintenance and updating of tools often have problems such as inefficiency, unclear records, and waste of resources, which not only affects production efficiency, but also may bring safety hazards and economic losses.

[0003] At the same time, it is often difficult to accurately grasp the service life and maintenance timing of tools, resulting in either premature scrapping of tools, or excessive use, which can cause safety issues. In addition, the management of tool usage rights for employees of different types of work and skill levels is also a complex issue. How to maximize tool utilization while ensuring safety has become a major challenge for enterprise managers. Summary of the invention

[0004] The present invention provides an intelligent tool management method and system, and the present invention improves the intelligence level of tool management.

[0005] In a first aspect, the present invention provides an intelligent tool management method, the intelligent tool management method comprising:

[0006] Collect and encode information on multiple tools and build a tool database containing tool type, specification size, usage status, priority index, standard service life and wear index;

[0007] Performing spatial planning on the tool cabinet according to the tool information in the tool database to obtain a hierarchical storage plan for the tool cabinet, and assigning three-dimensional coordinate position information to each tool;

[0008] Perform multi-modal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data;

[0009] When a user requests to borrow a tool, the user identification result and the tool use authority data are verified, and a tool borrowing instruction and borrowing record data are generated;

[0010] When a user requests to return a tool, the returned tool is checked to obtain current status information of the tool, and the current status information of the tool is input into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data;

[0011] A multi-dimensional analysis is performed on the borrowing record data and the predictive maintenance data to generate a tool usage heat map and a target tool management strategy.

[0012] In a second aspect, the present invention provides an intelligent tool management system, the intelligent tool management system comprising:

[0013] A collection module is used to collect and encode information of multiple tools and build a tool database containing tool type, specification size, usage status, priority index, standard service life and wear index;

[0014] A planning module, used for performing space planning on the tool cabinet according to the tool information in the tool database, obtaining a hierarchical storage plan for the tool cabinet, and assigning three-dimensional coordinate position information to each tool;

[0015] The allocation module is used to perform multimodal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data;

[0016] A verification module, used for verifying the user identification result and the tool use authority data when the user requests to borrow a tool, and generating a tool borrowing instruction and borrowing record data;

[0017] A detection module, for, when a user requests to return a tool, performing a status detection on the returned tool to obtain current status information of the tool, and inputting the current status information of the tool into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data;

[0018] A generation module is used to perform multi-dimensional analysis on the borrowing record data and the predictive maintenance data, and to generate a tool usage heat map and a target tool management strategy.

[0019] In the technical solution provided by the present invention, a comprehensive tool database including tool type, specification size, usage status, priority index, standard service life and wear index is constructed through multiple methods such as optical scanning, usage history analysis, and life data analysis. The hierarchical storage scheme of the tool cabinet and the precise positioning of the tools are realized by using a three-dimensional space filling algorithm, thereby improving space utilization and access convenience. By combining biometric features such as face, fingerprint and palm vein, accurate identification of users is realized, and reasonable tool usage rights are allocated to different users through multi-factor weight calculation, thereby improving the security and flexibility of management. Through comprehensive analysis of various factors, intelligent recommendation and visual guidance of the optimal borrowing tool are realized, thereby improving borrowing efficiency and user experience. By using weight sensing technology, comprehensive status detection of returned tools is realized, and high-precision prediction of the remaining service life of tools is realized by using a fully connected neural network, thereby effectively reducing tool maintenance costs and safety risks. Through multi-dimensional feature extraction, a tool usage heat map and optimized management strategy are generated, and the management strategy can be dynamically adjusted according to tool usage and maintenance requirements, thereby realizing continuous optimization and efficiency improvement of tool management, and improving the intelligent level of tool management. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0021] Figure 1 A schematic diagram of an embodiment of an intelligent tool management method in an embodiment of the present invention;

[0022] Figure 2 Schematic diagram of an embodiment of an intelligent tool management system in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] Embodiments of the present invention provide an intelligent tool management method and system. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the intelligent tool management method in the embodiment of the present invention includes:

[0025] Step S1, collecting and encoding information of multiple tools, and constructing a tool database including tool type, specification size, usage status, priority index, standard service life and wear index;

[0026] It is understandable that the execution subject of the present invention may be an intelligent tool management system, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0027] Specifically, the physical features of multiple tools are optically scanned, and the shape, size and surface texture of the tools are captured by high-precision scanning equipment to generate a three-dimensional digital model of each tool. According to the characteristics of the three-dimensional digital model, the tools are classified to determine their tool type and specification size data. The use history data of multiple tools are analyzed to extract the use status indicators. By analyzing the performance of the tools under different conditions and the historical use frequency, information about the wear and tear of the tools, the use frequency, etc. is obtained. Based on this information, a multi-factor weighted calculation is performed on the tools to determine the priority index of each tool. The priority index can help the management system identify which tools need to be maintained or replaced first. When performing life assessment, the reliability analysis of the tool life data is performed to obtain the standard service life estimate of each tool. According to the material properties and expected use conditions of the tools, the finite element analysis method is used to establish a theoretical wear model. By performing computational analysis on the theoretical wear model, the probability distribution of the tool wear index is obtained. To ensure the uniqueness and traceability of each tool in the management system, a unique identification code is assigned to each tool, and these identification codes are encoded into electronic tags using RFID technology. The electronic tags can not only be recognized by the tool cabinet, but also facilitate the borrowing and returning of tools. When constructing the tool database, the tool type information, specification size data, usage status index, priority index, standard service life estimate and wear index probability distribution data are structured to form specific data entries. The tool database is constructed based on the data entries and electronic tag information to support the operation of the intelligent tool management system.

[0028] Step S2, performing space planning on the tool cabinet according to the tool information in the tool database, obtaining a hierarchical storage plan for the tool cabinet, and assigning three-dimensional coordinate position information to each tool;

[0029] Specifically, a cluster analysis is performed on the priority index in the tool database, and the tools are divided into different groups according to their importance. The importance grouping results can effectively guide the hierarchical division of the tool cabinet and form a preliminary hierarchical scheme. In this scheme, more important and commonly used tools may be assigned to a location that is easier to access to improve the efficiency of use. A time series analysis is performed on the usage status indicators in the tool database, and a usage frequency prediction model is established to predict the usage frequency of each tool. According to the prediction results, multiple tools are sorted and the storage priority of tools in each layer is determined to ensure that commonly used tools can be quickly accessed and reduce the time for searching and accessing. When determining the storage location of each tool, the specification size data is filled in three-dimensional space. The three-dimensional space filling calculation is performed using these data to develop the optimal tool arrangement plan in each layer. This plan aims to maximize space utilization while ensuring that the spacing between tools is reasonable to avoid inconvenience caused by too close storage. After formulating the optimal tool arrangement plan, specific three-dimensional coordinate position information is assigned to each tool, and the coordinate information is used to guide the actual storage and access of the tool. Detect conflicts in the three-dimensional coordinate location information, evaluate the space utilization and accessibility of the storage solution, and find any potential space waste or access difficulties. According to the evaluation results, iteratively optimize the optimal tool arrangement solution in each layer. Obtain a hierarchical storage solution for the tool cabinet. The optimization process may include adjusting the storage level or location of the tools and replanning the layout of some tools to ensure that the final hierarchical storage solution for the tool cabinet is both efficient in space utilization and convenient for users to quickly access tools.

[0030] Step S3: Perform multimodal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data;

[0031] Specifically, the user's face image is processed by deep convolutional neural network to extract the face feature vector, and the user's fingerprint image is subjected to feature extraction and pattern matching to obtain fingerprint feature data. The user's palm vein image is subjected to vascular texture analysis, and a vein feature template is formed by feature point extraction. The face feature vector, fingerprint feature data and vein feature template are subjected to multimodal feature fusion to generate a comprehensive biometric feature identifier. The comprehensive biometric feature identifier is used to match the user database to obtain the user identity recognition result. After confirming the user's identity, the user's job position and skill level information are extracted from the user information database, and the basic authority coefficient is obtained by weighted calculation. The authority coefficient represents the basic authority level of the user in the use of tools, which is determined according to his job responsibilities and skill level. The user's historical tool use records are data mined, analyzed and the tool use proficiency score is obtained. The score reflects the proficiency and reliability shown by the user in the past tool use. By combining the basic authority coefficient with the tool use proficiency score, logical reasoning is performed to generate the user's use authority level for different types of tools, and determine the types and scope of tools that the user can operate in the system. Each tool in the tool database is classified by security level to assess the potential risks of using these tools. The assessment results include information such as the tool's operational complexity, usage risks, and maintenance requirements. According to the user's usage permission level and the tool's usage risk assessment results, permission matching is performed to generate tool usage permission data.

[0032] Step S4: When a user requests to borrow a tool, the user identification result and the tool use authority data are verified, and a tool borrowing instruction and borrowing record data are generated;

[0033] Specifically, when a user requests to borrow a tool, structured borrowing intention data including the target tool type and the expected usage time is obtained. According to the structured borrowing intention data, a query is performed in the tool database to generate a list of available tools that meet the conditions. The user identification result and the tool usage permission data are cross-validated to confirm whether the user has the permission to operate the target tool. According to the operation permission confirmation result, the list of available tools is filtered to form a subset of tools that the user can borrow. The current status of each tool in the tool subset is evaluated to obtain the real-time availability information of these tools. Based on the real-time availability information and the priority index of the tool, multi-objective optimization is performed to determine the tool that best suits the user's current needs and generate the optimal borrowing tool recommendation. The recommendation takes into account the tool usage frequency, maintenance status and user's permission level to ensure that the recommended tool meets the user's needs and is safe and reliable. The optimal borrowing tool recommendation is presented to the user, and after the user confirms, the final tool borrowing selection is determined. According to the user's tool borrowing selection, a tool positioning instruction is generated to control the LED indicator system of the tool cabinet to the corresponding tool position, provide visual guidance signals, and help users quickly find the required tools. The use of visual guidance signals greatly improves the efficiency of tool borrowing and reduces the time users spend looking for tools. The tool positioning instructions and visual guidance signals are combined to generate a complete tool borrowing instruction sequence. The tool borrowing instruction sequence, the association information between the user and the tool, the borrowing time and the expected return time are integrated to generate comprehensive borrowing record data. The borrowing record data is not only used to record the detailed information of the current borrowing, but also for subsequent tool management, maintenance plan formulation, and analysis of user permissions and usage.

[0034] Step S5: When the user requests to return the tool, the returned tool is checked to obtain the current status information of the tool, and the current status information of the tool is input into a preset predictive maintenance model to perform life assessment to obtain predictive maintenance data;

[0035] Specifically, when a user requests to return a tool, the RFID tag of the returned tool is scanned to obtain the unique identification code of the tool. Through the unique identification code, the historical status information and standard parameters of the tool are queried in the tool database, including the standard weight, geometric model, material characteristics, etc. of the tool. The returned tool is weight-checked, and the current weight data is obtained through the weight sensor, and compared with the standard weight of the tool to calculate the weight deviation value, which helps to identify whether the weight of the tool has changed due to use or other reasons, such as material loss or increase. The returned tool is geometrically measured using three-dimensional scanning technology to generate the current three-dimensional geometric model. By comparing this model with the standard geometric model, the deformation degree data of the tool is calculated to reflect the physical stress and deformation of the tool during use. At the same time, the tool surface is scanned using multi-spectral imaging technology to obtain material feature maps. After image analysis and processing, these images reveal the wear and corrosion degree data of the tool surface. Based on the above detection data, a multi-dimensional status assessment is performed to generate the current status information of the tool. This information includes the weight deviation value, deformation degree data, and wear and corrosion degree data of the tool. The tool usage time, usage frequency and working environment data are analyzed to calculate the usage intensity coefficient. The usage intensity coefficient is a comprehensive indicator that reflects the total workload of the tool during its life cycle and can be used to predict the remaining service life of the tool. The current status information of the tool and the usage intensity coefficient are input into the preset predictive maintenance model for processing. The model uses historical data and machine learning algorithms to predict the remaining service life of the tool. The predicted value is an important reference for formulating maintenance strategies and can help decide whether the tool needs maintenance, servicing or even replacement. Based on the predicted remaining service life, the corresponding tool maintenance recommendations are extracted from the maintenance strategy database. These recommendations may include routine inspections, specific parts replacement or other maintenance operations to ensure that the tool is in optimal condition before the next use. The current status information of the tool, the remaining service life prediction value and maintenance recommendations are integrated to generate predictive maintenance data.

[0036] The current state information of the tool is processed by feature extraction to obtain the state feature vector including weight deviation value, deformation degree and wear and corrosion degree. The use intensity coefficient is normalized to obtain the standardized use intensity index. The state feature vector is concatenated with the standardized use intensity index to form a complete input feature vector. The input feature vector is input into the preset predictive maintenance model for processing. The model consists of multiple layers of neural network structure, including the first layer of fully connected neural network, the second layer of fully connected neural network, attention mechanism, long short-term memory network (LSTM) and fully connected layer. The input feature vector is processed by the first layer of fully connected neural network to generate the output of the first hidden layer, and the output of the first hidden layer is processed by the second layer of fully connected neural network to generate the output of the second hidden layer. The output of the second hidden layer is processed by the attention mechanism to obtain the weighted feature representation. The attention mechanism increases the focus on key features by calculating the weight of each feature, so that the model can more accurately predict the remaining service life. The weighted feature representation is input into the long short-term memory network (LSTM) to capture the trend of tool status over time, use the historical status data and usage of the tool to improve the accuracy of prediction and generate time series features. The time series features are processed through the fully connected layer to generate the probability distribution of the remaining service life of the tool, reflecting the possible service life range of the tool in the future. The expectation calculation is performed on the probability distribution to obtain the point estimate of the remaining service life. The confidence interval calculation is performed on the point estimate and probability distribution to obtain the predicted value of the remaining service life.

[0037] The input feature vector is dimensionally analyzed to determine its feature dimension. According to the feature dimension and the number of neurons in the first hidden layer preset in the first fully connected neural network, the first layer weight matrix is ​​initialized to generate the initial weight matrix. In order to optimize the training effect of the neural network, the initial weight matrix is ​​processed by the Xavier initialization method. The Xavier initialization method can effectively control the signal transmission between different layers, make the distribution of gradients more uniform, avoid gradient explosion or disappearance, and obtain the first layer optimized weight matrix. The first layer optimized weight matrix is ​​used to perform matrix multiplication with the input feature vector to calculate the first layer linear transformation result. In order to introduce nonlinear features, the ReLU activation function is applied to the linear transformation result to filter out all negative values ​​and set them to zero, while the positive part remains unchanged to generate the first hidden layer output. The application of the ReLU activation function helps to improve the learning ability and training speed of the network, and is particularly suitable for processing large-scale data. According to the dimension of the first hidden layer output and the number of neurons in the second hidden layer preset in the second fully connected neural network, the second layer weight matrix is ​​initialized again and optimized using Xavier, so that the second layer neural network can effectively capture the characteristics and patterns of the input data. The second layer optimized weight matrix is ​​matrix multiplied with the output of the first hidden layer to obtain the second layer linear transformation result. When processing the second layer linear transformation result, the LeakyReLU activation function is used. This activation function allows a very small slope when the input is less than zero, avoiding the "dead neuron" problem of the ReLU function, thereby generating more information flow. In order to improve the generalization ability of the model, dropout regularization is applied after the LeakyReLU activation function is processed to randomly discard a part of the neurons to prevent overfitting. After LeakyReLU activation and dropout processing, the second hidden layer output is obtained.

[0038] Step S6: Perform multi-dimensional analysis on the borrowing record data and predictive maintenance data to generate a tool usage heat map and a target tool management strategy.

[0039] Specifically, the borrowing record data is analyzed in time series to determine the frequency curve and periodic pattern of each tool. By analyzing these curves and patterns, the peak and trough periods of tool use are identified, and the utilization efficiency of the tools is evaluated. According to the obtained frequency curve and periodic pattern, the tools are clustered and grouped into tools with similar usage patterns to form tool groups with different usage patterns. Grouping helps to understand the actual demand for tools and user preferences. At the same time, the remaining service life prediction values ​​in the predictive maintenance data are analyzed in probability distribution to obtain the life risk assessment index of each tool. The life risk assessment index reflects the risk level of possible failure of the tool. The comprehensive maintenance priority ranking of the tool is obtained by weighted calculation based on the life risk assessment index and the importance index of the tool. Social network analysis is performed on the user-tool association information in the borrowing record data to construct a correlation network diagram of tool use. The network diagram shows the correlation relationship between tools and reveals the matching of tools in actual operation. Using this information, combined with the three-dimensional coordinate position information of the tools, spatial correlation analysis is performed to optimize the physical placement strategy of the tools, and tools that are frequently used together are placed adjacent to each other, thereby reducing the user's access time and space waste. Multi-dimensional feature extraction is performed on the borrowing record data and predictive maintenance data to generate high-dimensional feature vectors that reflect the tool usage and status. The high-dimensional feature vectors are reduced in dimension and visualized to obtain a tool usage heat map. The tool usage heat map, comprehensive maintenance priority sorting, and optimized tool proximity placement suggestions are used to perform multi-objective optimization and obtain a target tool management strategy. This strategy takes into account both the tool usage efficiency and the reasonable allocation of maintenance resources, ensuring that the tool life is extended while meeting user needs and improving overall management efficiency.

[0040] In the embodiment of the present invention, a comprehensive tool database including tool type, specification size, usage status, priority index, standard service life and wear index is constructed through optical scanning, usage history analysis, life data analysis and other methods. The hierarchical storage scheme of the tool cabinet and the precise positioning of the tools are realized by using the three-dimensional space filling algorithm, which improves the space utilization and access convenience. By combining biometric features such as face, fingerprint and palm vein, accurate identification of users is realized, and reasonable tool usage rights are allocated to different users through multi-factor weight calculation, which improves the security and flexibility of management. Through comprehensive analysis of various factors, intelligent recommendation and visual guidance of the optimal borrowing tool are realized, which improves the borrowing efficiency and user experience. By using weight sensing technology, comprehensive status detection of returned tools is realized, and high-precision prediction of the remaining service life of tools is realized by using a fully connected neural network, which effectively reduces the maintenance cost and safety risk of tools. Through multi-dimensional feature extraction, a tool usage heat map and optimized management strategy are generated, which can dynamically adjust the management strategy according to the tool usage and maintenance needs, realize the continuous optimization and efficiency improvement of tool management, and improve the intelligent level of tool management.

[0041] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0042] (1) Optically scanning the physical features of multiple tools to obtain a three-dimensional digital model including tool contours, dimensions, and surface textures, and classifying the multiple tools based on the three-dimensional digital model to obtain tool type information and specification size data;

[0043] (2) Analyze the usage history data of multiple tools to obtain usage status indicators, and perform multi-factor weighted calculations on multiple tools based on the usage status indicators to obtain a priority index;

[0044] (3) Perform reliability analysis on the life data of multiple tools to obtain standard service life estimates. Perform finite element analysis on multiple tools based on tool material properties and expected usage conditions to obtain theoretical wear models. At the same time, perform computational analysis on the theoretical wear model to obtain the probability distribution of wear indicators.

[0045] (4) Assign a unique identification code to each tool based on the tool type information, and perform RFID encoding on the identification code to obtain electronic tag information that can be recognized by the tool cabinet;

[0046] (5) Structural processing is performed on tool type information, specification size data, usage status indicators, priority indexes, standard service life estimates, and wear index probability distributions to obtain data entries, and a tool database is constructed based on the data entries and electronic tag information.

[0047] Specifically, optical scanning technology is used to obtain detailed physical characteristics of each tool. Optical scanning equipment can accurately capture the contour, size and surface texture of the tool to generate a three-dimensional digital model. These models not only contain the shape characteristics of the tool, but also reflect its surface details, such as roughness and tiny wear marks. Through the three-dimensional model, the tools are classified to obtain tool type information and specification size data. For example, for common tools such as screwdrivers and wrenches, optical scanning can accurately measure their length, width, thickness and surface texture characteristics of the handle to distinguish different tool types and models. The use history data of multiple tools is analyzed. By recording the frequency, time and environmental conditions of the tool, the use status index of each tool is generated. These indicators include information such as the cumulative use time of the tool, the most recent use date, and the intensity of use. According to the use status index, a multi-factor weighted calculation is performed on multiple tools to obtain a priority index. The reliability analysis of the tool life data is performed to determine the standard service life estimate of each tool. The service life data not only includes the actual use time, but also combines the material properties and expected use conditions of the tool to establish a theoretical wear model through finite element analysis. Finite element analysis can simulate the stress and strain of tools under different loads and environments, thereby predicting their wear. Using the results of finite element analysis, the probability distribution of tool wear indicators is generated to reflect the life expectancy of the tool under different usage conditions. Specifically, the results of finite element analysis can be expressed as:

[0048] ;

[0049] in, Indicates the degree of wear, It's time The stress distribution function on is the total time the tool is used. Analyze and obtain the probability distribution of wear indicators, so as to predict the remaining service life of the tool. A unique identification code is assigned to each tool according to the type information of the tool. The identification code is RFID-encoded to generate electronic tag information that can be recognized by the tool cabinet. RFID technology makes the tracking and management of tools more convenient. The tool cabinet can quickly identify the tool and record its use by scanning the electronic tag. All tool information is structured, including tool type information, specification size data, usage status indicators, priority index, standard service life estimation value and wear indicator probability distribution. These data are organized into specific data entries and combined with electronic tag information to build a comprehensive tool database. The database not only supports daily tool management, but also provides data support for maintenance planning and inventory management. For example, by analyzing the priority index and usage status indicators, it is determined which tools need priority maintenance and which tools can extend their service life.

[0050] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0051] (1) Cluster analysis is performed on the priority indexes in the tool database to obtain the importance grouping results, and the tool cabinets are divided into levels according to the importance grouping results to obtain a preliminary stratification scheme for the tool cabinets;

[0052] (2) Perform time series analysis on the usage status indicators in the tool database to obtain a usage frequency prediction model, and sort multiple tools according to the usage frequency prediction model to obtain the storage priority of tools in each layer;

[0053] (3) Filling the three-dimensional space with the specification size data to obtain the optimal tool arrangement plan in each layer, and assigning three-dimensional coordinates to each tool according to the optimal tool arrangement plan to obtain three-dimensional coordinate position information;

[0054] (4) Conflict detection is performed on the three-dimensional coordinate position information to obtain the evaluation results of space utilization and access convenience. Based on the evaluation results, the optimal tool arrangement plan in each layer is iteratively optimized to obtain a hierarchical storage plan for the tool cabinet.

[0055] Specifically, a cluster analysis is performed on the priority index in the tool database. Tools are grouped according to their importance for more efficient management and storage. The priority index can be calculated based on multiple factors, including the frequency of use, importance, and maintenance requirements of the tool. The K-means clustering algorithm is used to divide the tools into different importance groups. The formula is as follows:

[0056] ;

[0057] in, Indicates clusters, It is The priority index of the tools, It is The center of each cluster. In this way, the tools are divided into several groups, such as high priority, medium priority, and low priority. According to the importance grouping results, the tool cabinet is divided into levels to form a preliminary hierarchical scheme. Usually, high-priority tools are placed in a more accessible level, while low-priority tools can be placed in a relatively inaccessible area. Time series analysis is performed on the usage status indicators in the tool database to obtain a usage frequency prediction model. Time series analysis helps identify tool usage trends and periodic characteristics, which helps to predict future tool needs. The usage frequency prediction model can be constructed based on the ARIMA (autoregressive integrated moving average) model, and the formula is as follows:

[0058] ;

[0059] in, It's time The amount of tool usage and are the coefficients of the autoregressive and moving average parts, respectively. is the error term. The model is used to predict the frequency of use of each tool in a certain period of time in the future. Multiple tools are sorted according to the frequency prediction model to determine the storage priority of tools in each layer. For example, tools that are predicted to be used more frequently should be placed in a more accessible position for frequent use. The specification size data of the tools are filled in three-dimensional space. According to the space size of each layer and the physical size of the tools, the optimal tool arrangement scheme is designed. The goal of three-dimensional space filling is to maximize space utilization while ensuring that there is enough clearance between tools for easy access. For each tool, its three-dimensional coordinates are calculated and arranged in a specific position. For example, a greedy algorithm or a linear programming method is used to optimize the position of each tool to maximize space utilization. Conflict detection is performed on the three-dimensional coordinate position information of all tools to ensure that there is no overlap or excessive closeness between tools. Conflict detection can be achieved by calculating the minimum distance between tools. If the distance between some tools is less than the preset safety threshold, the position of the tools needs to be readjusted. Through conflict detection, the space utilization and access convenience of the tool cabinet are evaluated, and the optimal tool arrangement scheme in each layer is iteratively optimized based on these evaluation results. During the optimization process, the hierarchical position of the tools is adjusted or the storage order of some tools is rearranged to ensure that all tools are easily accessible and the space utilization is maximized. Through multiple iterations of optimization, an optimized hierarchical storage solution for the tool cabinet is obtained. In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0060] (1) Perform deep convolutional neural network processing on the user's face image to obtain a face feature vector, and perform feature extraction and pattern matching on the user's fingerprint image to obtain fingerprint feature data;

[0061] (2) Performing vascular texture analysis and feature point extraction on the user's palm vein image to obtain a vein feature template, and then performing multimodal feature fusion based on the facial feature vector, fingerprint feature data and vein feature template to obtain a comprehensive biometric identification;

[0062] (3) Perform database matching on the comprehensive biometric identification to obtain the user identification result, and extract the user's job position and skill level information from the user information database based on the user identification result to perform weight calculation to obtain the basic authority coefficient;

[0063] (4) Perform data mining on the user's historical tool usage records to obtain the tool usage proficiency score, and perform logical reasoning based on the basic authority coefficient and the tool usage proficiency score to obtain the user's permission level for different types of tools;

[0064] (5) Classify the security level of each tool in the tool database to obtain the use risk assessment results, and match the permissions based on the user's permission level and the tool use risk assessment results to obtain the tool use permission data.

[0065] Specifically, the user's face image is processed by a deep convolutional neural network to extract a unique face feature vector. The deep convolutional neural network model can capture key features in the face image, such as the texture and shape features of the eyes, nose, mouth and other areas through multi-layer convolution and pooling operations. In the fingerprint recognition part, the user's fingerprint image is subjected to feature extraction and pattern matching. By using a feature extraction algorithm, such as the Minutiae feature extraction method, the detailed feature points of the fingerprint are obtained, which include bifurcation points and endpoints. By matching these feature points, a fingerprint feature data vector is generated. The palm vein image of the user is subjected to vascular texture analysis and feature point extraction. Palm vein recognition is based on the light absorption characteristics of veins under near-infrared light irradiation. By analyzing the texture pattern of veins, a vein feature template is generated. Multimodal feature fusion is performed based on the face feature vector, fingerprint feature data and vein feature template, and different feature vectors are merged into a comprehensive biometric identification. The feature fusion is performed by weighted average or deep learning method, so that the fused features are the most representative in distinguishing different users. Through the comprehensive biometric identification, database matching is performed to verify the identity of the user and obtain the user identity recognition result. Based on the identity recognition results, the user's job position and skill level information is extracted from the user information database. Based on this information, the weight calculation is performed to obtain the basic authority coefficient The basic authority coefficient reflects the user's basic authority scope and responsibility level. For example, managers may have broader tool usage authority, while ordinary technicians have relatively limited authority. Data mining is performed on the user's historical tool usage records to analyze their usage frequency, operation time, and tool types to generate tool usage proficiency scores. This score indicates the user's familiarity with and skill in using a particular tool. Assume that some characteristics of the tool usage record (such as the number of times used, success rate, etc.) are And its corresponding weight is , then the tool use proficiency score can be expressed as:

[0066] ;

[0067] According to the basic authority coefficient and tool proficiency rating , through logical reasoning, obtain the user's permission level for different types of tools For example, if a user performs well in using complex tools and has high job requirements, his or her permission level may be increased, allowing him or her to operate more types of tools or tools with higher risks. The security level of each tool in the tool database is classified to obtain the risk assessment results of each tool. Risk assessment results of tool use It can be calculated based on factors such as the tool's operational complexity, maintenance requirements, and potential danger. Assume that the risk factors of the tool are , then the risk assessment result can be expressed as:

[0068] ;

[0069] in, It is the weight of each risk factor. According to the user's permission level and the tool use risk assessment results, permission matching is performed to generate tool use permission data, determine the specific tool types and quantities that users can borrow and operate, and ensure that the work needs are met and the operation is safe during the tool use process.

[0070] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0071] (1) When a user requests to borrow a tool, the system obtains structured borrowing intention data including the target tool type and expected usage time, and performs a query and match on the tool database based on the structured borrowing intention data to obtain a list of available tools;

[0072] (2) Cross-validate the user identification results and tool usage permission data to obtain the user's permission to operate the target tool, and filter the list of available tools based on the permission confirmation results to obtain a subset of tools that the user can borrow;

[0073] (3) Evaluate the current status of each tool in the tool subset to obtain the real-time availability information of the tool, and perform multi-objective optimization based on the real-time availability information and priority index to obtain the optimal tool recommendation;

[0074] (4) The user confirms the optimal tool borrowing suggestion, obtains the tool borrowing selection, generates a tool positioning instruction based on the tool borrowing selection, controls the LED indication system of the tool cabinet, and obtains a visual guidance signal;

[0075] (5) The tool positioning instructions and visual guidance signals are combined to obtain a tool borrowing instruction sequence, and the tool borrowing instruction sequence, user-tool association information, borrowing time, and expected return time are integrated to obtain borrowing record data.

[0076] Specifically, when a user requests to borrow a tool, the user's structured borrowing intention data is obtained. This data includes the type of target tool that the user wants to borrow and the expected usage time. The information is obtained through user input or selection and stored in the system as part of the borrowing request. According to the structured borrowing intention data, the tool database is queried and matched to generate a preliminary list of available tools, and all tools that meet the user's needs are screened out to ensure that these tools meet the borrowing request in terms of technical specifications and functions. The user's identity identification results and tool use permission data are cross-verified to ensure that the user is eligible to borrow the requested tool. The identity identification results usually include the user's position, skill level and other relevant information, while the tool use permission data defines the type and level of tools that the user can operate. Through the verification process, the user's operation permission confirmation results for the target tool are generated to help screen the preliminary list of available tools, filter out those tools that the user does not have the right to use, and obtain the subset of tools that the user can borrow. The current status of the tool is evaluated. The current status evaluation includes checking the availability, maintenance status and any factors that may affect its use. Based on the real-time availability information of the tools, combined with the priority index of each tool, multi-objective optimization is performed to generate the optimal borrowing tool recommendation. It can be calculated by comprehensively considering the frequency of use, importance and maintenance requirements of the tool. The formula is as follows:

[0077] ;

[0078] in, Indicates the frequency of use of the tool, Indicates the importance of tools, Indicates maintenance needs, , and is the corresponding weight coefficient. In this way, tools that meet user needs and are in good condition are recommended. After receiving the optimal borrowing tool suggestions, the user chooses to confirm or adjust these suggestions. The confirmed tool borrowing selection will be used to generate tool positioning instructions. These instructions control the LED indication system of the tool cabinet to the corresponding tool position to provide visual guidance signals to help users quickly find the selected tools. The LED indication system makes the tool location intuitive by lighting or flashing specific lights, reducing the time and effort of users to find tools. The tool positioning instructions and visual guidance signals are combined to form a complete tool borrowing instruction sequence. At the same time, the tool borrowing instruction sequence, user-tool association information, borrowing time, and expected return time are integrated to generate complete borrowing record data.

[0079] In a specific embodiment, the process of executing step S5 may specifically include the following steps:

[0080] (1) When a user requests to return a tool, the RFID tag of the returned tool is scanned to obtain a unique identification code;

[0081] (2) Query the tool database based on the unique identification code to obtain the historical status information and standard parameters of the returned tool;

[0082] (3) Performing weight sensing on the returned tool to obtain current weight data, comparing the current weight data with the standard weight to obtain a weight deviation value, and performing a three-dimensional scan on the returned tool to obtain a current geometric model, comparing the current geometric model with the standard geometric model to obtain deformation degree data;

[0083] (4) Perform multispectral imaging on the surface of the returned tool to obtain a material feature map of the tool surface, and perform image analysis on the material feature map of the tool surface to obtain wear and corrosion degree data;

[0084] (5) Perform multi-dimensional status assessment based on weight deviation value, deformation degree data, and wear and corrosion degree data to obtain the current status information of the tool, and calculate the usage intensity of the returned tool based on the usage time, usage frequency, and working environment data to obtain the usage intensity coefficient;

[0085] (6) Inputting the current status information and usage intensity coefficient of the tool into the preset predictive maintenance model to obtain the remaining service life prediction value of the returned tool;

[0086] (7) According to the remaining service life prediction value, the corresponding tool maintenance recommendation is queried from the maintenance strategy database, and the tool current status information, remaining service life prediction value and maintenance recommendation are integrated to obtain predictive maintenance data.

[0087] Specifically, when a user requests to return a tool, the RFID tag of the tool is scanned to obtain the unique identification code of the tool. The tool database is queried based on the identification code to extract the historical status information and standard parameters of the tool. This information includes the initial status of the tool, the usage history, and the standard weight and geometric dimensions set during manufacturing. The returned tool is weight-sensed to obtain the current weight data. By comparing the current weight data with the standard weight, the weight deviation value is calculated. The deviation value helps to determine whether the tool has material loss or additional substances, such as dirt or attached foreign matter. After completing the weight detection, the tool is 3D scanned to generate the current geometric model of the tool. By comparing the current geometric model with the standard geometric model, the deformation degree of the tool is detected. The deformation degree data reflects the physical damage or deformation that the tool may suffer during use, such as bending or twisting due to excessive use or incorrect operation. This information helps to evaluate the use status of the tool and determine whether it needs maintenance or replacement. The surface of the tool is inspected using multispectral imaging technology to obtain the material feature map of the tool surface, revealing microscopic changes in the surface material, such as wear, corrosion or other surface damage. Through image analysis, these wear and corrosion data are quantified. Wear and corrosion analysis helps to identify the actual use of the tool, especially those minor damages that may affect the function of the tool. Combined with the above test results, the current status of the tool is evaluated in multiple dimensions. Including integrating weight deviation values, deformation degree data and wear and corrosion degree data to generate complete information on the current status of the tool. Combined with the tool's usage time, usage frequency and working environment data, the usage intensity is calculated. Usage intensity coefficient It can be calculated by the following formula:

[0088] ;

[0089] in, It is Duration of use, is the frequency of use, is the total number of times the tool has been used. The usage intensity factor reflects the total load that the tool has been subjected to during its life cycle and helps predict the remaining service life of the tool. The current status information of the tool and the usage intensity factor are input into the preset predictive maintenance model to calculate the remaining service life prediction value of the tool. The predictive maintenance model estimates the limit state that the tool may reach during future use by analyzing historical data and current status. According to the remaining service life prediction value, the corresponding maintenance recommendations are queried from the maintenance strategy database. The current status information of the tool, the remaining service life prediction value and the maintenance recommendations are integrated to generate predictive maintenance data.

[0090] In a specific embodiment, the execution step inputs the current status information of the tool and the usage intensity coefficient into a preset predictive maintenance model to obtain the remaining useful life prediction value of the returned tool, which may specifically include the following steps:

[0091] (1) Perform feature extraction on the current state information of the tool to obtain a state feature vector including weight deviation value, deformation degree, and wear and corrosion degree;

[0092] (2) Normalizing the usage intensity coefficient to obtain a standardized usage intensity index, and concatenating the state feature vector and the standardized usage intensity index to obtain an input feature vector;

[0093] (3) Inputting the input feature vector into a preset predictive maintenance model, the predictive maintenance model includes: a first layer fully connected neural network, a second layer fully connected neural network, an attention mechanism, a long short-term memory network, and a fully connected layer;

[0094] (4) Processing the input feature vector with the first layer of fully connected neural network to obtain the first hidden layer output, and processing the hidden layer output with the second layer of fully connected neural network to obtain the second hidden layer output;

[0095] (5) The hidden layer output is processed by the attention mechanism to obtain a weighted feature representation, and the weighted feature representation is input into the long short-term memory network to capture the trend of tool status changes over time and obtain time series features;

[0096] (6) Perform full connection layer processing on the time series features to obtain the probability distribution of the remaining useful life;

[0097] (7) Perform expectation calculation on the probability distribution to obtain a point estimate of the remaining useful life, and perform confidence interval calculation on the point estimate and probability distribution to obtain a predicted value of the remaining useful life.

[0098] Specifically, feature extraction is performed on the current status information of the tool. Key features that can reflect the health status of the tool are extracted from various detection data of the tool. These include weight deviation value, deformation degree, and wear and corrosion degree. A state feature vector S is constructed based on the above features. Reflects the quality change and deformation degree of the tool Indicates the physical deformation, wear and corrosion degree of the tool The tool usage intensity coefficient is normalized to obtain the standardized usage intensity index Normalization helps adjust the usage intensity coefficient to a uniform range, making it comparable with other features. The normalization formula is as follows:

[0099] ;

[0100] in, is the original use intensity coefficient, and are the minimum and maximum values ​​of the usage intensity coefficient, respectively. In this way, the standardized usage intensity index is adjusted to the range between 0 and 1. and standardized usage intensity indicators Spliced ​​together to form the input feature vector , which is expressed as:

[0101] ;

[0102] The input feature vector contains various key status information of the tool, which is input into the preset predictive maintenance model for processing. The structure of the predictive maintenance model includes a multi-layer neural network and a time series analysis module. Perform the first layer of fully connected neural network processing to obtain the first hidden layer output The first layer of the fully connected neural network is connected through the weight matrix and the bias term Perform linear transformation and nonlinear activation, the formula is as follows:

[0103] ;

[0104] Among them, ReLU is the rectified linear unit activation function, which is used to introduce nonlinear features. Perform the second layer of fully connected neural network processing to obtain the second hidden layer output , the calculation formula is similar:

[0105] ;

[0106] In order to better capture important features, the model introduces an attention mechanism. Perform attention mechanism processing to obtain weighted feature representation The attention mechanism emphasizes important features by calculating the weights of the features. The formula is as follows:

[0107] ;

[0108] in, is the weight coefficient of the feature, which is learned through training data. The weighted feature is represented by Input long short-term memory network (LSTM) to capture the trend of tool status changes over time and obtain time series features LSTM can effectively process time series data through the memory and forgetting mechanism. Its formula is as follows:

[0109] ;

[0110] Timing characteristics It contains the tool status information at multiple time points, which can more accurately predict the remaining service life of the tool. Perform full connection layer processing to obtain the probability distribution of remaining service life Through the linear transformation and activation function processing of the fully connected layer, the probability distribution describes the possibility of tools under different service lives. Perform expectation calculations to obtain a point estimate of the remaining useful life The expected calculation formula is as follows:

[0111] ;

[0112] in, Indicates the possible service life, is the corresponding probability. In order to provide confidence, the confidence interval of the point estimate is calculated to obtain the remaining useful life forecast. The confidence interval can be calculated by the standard deviation and expected value of the probability distribution, providing the reliability and accuracy of the forecast.

[0113] In a specific embodiment, the execution step performs a first layer of fully connected neural network processing on the input feature vector to obtain a first hidden layer output, and performs a second layer of fully connected neural network processing on the hidden layer output to obtain a second hidden layer output. The process may specifically include the following steps:

[0114] (1) Performing dimensional analysis on the input feature vector to obtain the feature dimension, and initializing the first-layer weight matrix according to the feature dimension and the number of first hidden layer neurons preset in the first-layer fully connected neural network to obtain the initial weight matrix;

[0115] (2) Perform Xavier initialization on the initial weight matrix to obtain the first-layer optimized weight matrix, and perform matrix multiplication on the first-layer optimized weight matrix and the input feature vector to obtain the first-layer linear transformation result;

[0116] (3) Perform ReLU activation function processing on the linear transformation result of the first layer to obtain the output of the first hidden layer;

[0117] (4) Initializing and performing Xavier optimization on the second-layer weight matrix according to the dimension of the first hidden layer output and the number of second hidden layer neurons preset in the second-layer fully connected neural network to obtain the second-layer optimized weight matrix;

[0118] (5) Perform matrix multiplication on the second-layer optimized weight matrix and the output of the first hidden layer to obtain the second-layer linear transformation result, and perform LeakyReLU activation function processing and dropout regularization processing on the second-layer linear transformation result to obtain the second hidden layer output.

[0119] Specifically, the input feature vector is subjected to dimensional analysis. The input feature vector contains multiple features, such as weight deviation, deformation degree, wear and corrosion degree, and standardized use intensity index. Through dimensional analysis, the dimension of the feature vector is determined, which represents the number of features of the input data. Assuming that the dimension of the input feature vector is 4, it means that the state of each tool is described by four different features. According to the dimension of the input feature and the number of neurons in the first hidden layer preset in the first layer of the fully connected neural network , initialize the first layer weight matrix. Weight matrix The dimension will be , where each element represents the connection weight between the input feature and the hidden neuron. The initial weight matrix is ​​usually randomly initialized to break the symmetry so that each neuron can learn different features. In order to optimize the initialization effect of the weight matrix, the Xavier initialization method is used to select the initialization distribution range based on the number of input and output layers, which can help the neural network train more effectively. The formula for Xavier initialization is as follows:

[0120] ;

[0121] in, represents an element in the first layer weight matrix, Indicates that in the interval After Xavier initialization, the optimized first-layer weight matrix is ​​obtained. . Using this optimized weight matrix and the input feature vector Perform matrix multiplication to calculate the first layer linear transformation result :

[0122] ;

[0123] in, is the bias term of the first layer. Linear transformation result After being processed by the ReLU activation function, the output of the first hidden layer is obtained The ReLU function is defined as follows:

[0124] ;

[0125] The ReLU activation function sets negative values ​​to zero and retains only positive values, which helps the network introduce nonlinear features and thus better fit complex data. and the preset number of neurons in the second hidden layer of the second fully connected neural network , for the second layer weight matrix Initialize. The initialization and optimization process is similar to the first layer. The Xavier initialization method is used to ensure that the model has a good starting point at the beginning of training. The weight matrix of the second layer after initialization The dimension is ,in is the dimension of the first hidden layer output. Use the optimized second layer weight matrix and the first hidden layer output Perform matrix multiplication to calculate the second-layer linear transformation result :

[0126] ;

[0127] in, is the bias term of the second layer. In order to introduce more nonlinear features, the second layer linear transformation result After LeakyRelU activation function processing, we get The LeakyReLU function is defined as follows:

[0128] ;

[0129] in, is a small constant (such as 0.01) used to keep the negative value from being completely zero, which helps avoid the "dead neuron" problem that ReLU may cause. To prevent the model from overfitting, the output of the second layer Dropout regularization is applied to the model. The dropout method randomly discards some neurons and their connections to prevent the model from being too dependent on the training data, thereby improving its generalization ability. The final output of the second hidden layer is It is the output result after dropout processing.

[0130] In a specific embodiment, the process of executing step S6 may specifically include the following steps:

[0131] (1) Conduct time series analysis on the borrowing record data to obtain the usage frequency curve and periodic pattern of each tool. Then, perform cluster analysis on the tools based on the usage frequency curve and periodic pattern to obtain tool groups with different usage patterns.

[0132] (2) Perform probability distribution analysis on the remaining service life prediction values ​​in the predictive maintenance data to obtain the life risk assessment index of each tool, and perform weighted calculation based on the life risk assessment index and the tool importance index to obtain the comprehensive maintenance priority ranking of the tools;

[0133] (3) Perform social network analysis on the user-tool association information in the borrowing record data to obtain a correlation network diagram of tool use. Perform spatial correlation analysis based on the correlation network diagram and three-dimensional coordinate location information to obtain optimized tool proximity placement suggestions.

[0134] (4) Perform multi-dimensional feature extraction on the borrowing record data and predictive maintenance data to obtain a high-dimensional feature vector that reflects the tool usage and status. Perform dimensionality reduction and visualization on the high-dimensional feature vector to obtain a tool usage heat map.

[0135] (5) Multi-objective optimization is performed based on the tool usage heat map, comprehensive maintenance priority ranking, and optimized tool proximity placement recommendations to obtain the target tool management strategy.

[0136] Specifically, the borrowing record data is analyzed in time series. The borrowing record data includes information such as the borrowing time, return time, and usage frequency of each tool. Through time series analysis, the usage frequency curve of each tool is drawn and its periodic pattern is revealed. For example, if the usage frequency of certain tools increases significantly on Mondays, it may be due to the arrangement of specific tasks. Usage frequency curve Describes the tool at time The intensity of use is analyzed by The peaks and cycles of the tools can be identified to understand the usage patterns of the tools. Cluster analysis of the tools based on the usage frequency curve and periodic pattern, and grouping of tools with similar usage patterns can help to formulate maintenance plans and optimize tool placement strategies more accurately. Common clustering algorithms such as the K-means algorithm calculate the similarity metrics of the tool usage frequency curves, such as Euclidean distance for grouping. The clustering results divide the tools into different usage pattern groups, such as high-frequency usage groups, periodic usage groups, and low-frequency usage groups. After completing the clustering of the tool usage patterns, a probability distribution analysis is performed on the remaining service life prediction values ​​in the predictive maintenance data. The remaining service life prediction value provides the service life limit that the tool may reach in a certain period of time in the future. By analyzing these data, the life risk assessment index of each tool is calculated. Life risk assessment index It reflects the potential failure risk of the tool and is calculated as follows:

[0137] ;

[0138] in, It’s a tool For tools with higher importance, consider the tool importance index. , and get the comprehensive maintenance priority ranking. The comprehensive maintenance priority ranking is obtained by weighted calculation of the life risk assessment index and the tool importance index:

[0139] ;

[0140] in, and It is the coefficient for adjusting the weights of the two indicators. Used to determine which tools need to be inspected and maintained first. Social network analysis is performed on the user-tool association information in the borrowing record data to reveal the relationship network formed by the use of tools between users, helping to understand which tools are often used together or borrowed by the same user group. By analyzing the user-tool association, a tool use association network diagram is generated. The association network diagram shows the use association between tools. For example, some tools may be frequently borrowed by the same group of users, indicating that they are used collaboratively in specific tasks. Spatial correlation analysis is performed based on the association network diagram and the three-dimensional coordinate position information of the tools. Spatial correlation analysis aims to optimize the physical placement of tools so that tools that are often used together are placed adjacent to each other, reducing the time of retrieval and improving work efficiency. By optimizing the adjacent placement suggestions of tools, the convenience of tool storage and the efficiency of retrieval can be improved. Multi-dimensional feature extraction is performed on the borrowing record data and predictive maintenance data, including the frequency of use, intensity of use, life risk, association, etc. of the tools, to form a high-dimensional feature vector. The high-dimensional feature vector is subjected to dimensionality reduction processing, usually using principal component analysis or t-SNE and other technologies, and the data after dimensionality reduction is visualized as a tool use heat map. The tool usage heat map visually displays the usage and health status of the tools, helping managers quickly identify problem tools or high-usage tools. Combine the tool usage heat map, comprehensive maintenance priority sorting, and optimized tool proximity placement suggestions to perform multi-objective optimization. Multi-objective optimization considers multiple goals, such as reducing maintenance costs, improving tool availability, and optimizing tool layout. By weighing these goals, a target tool management strategy is generated to guide the daily management and maintenance plan of the tools.

[0141] The above describes the intelligent tool management method in the embodiment of the present invention. The following describes the intelligent tool management system in the embodiment of the present invention. Figure 2 , an embodiment of the intelligent tool management system in the embodiment of the present invention includes:

[0142] A collection module is used to collect and encode information of multiple tools and build a tool database containing tool type, specification size, usage status, priority index, standard service life and wear index;

[0143] A planning module is used to plan the space of the tool cabinet according to the tool information in the tool database, obtain a hierarchical storage plan for the tool cabinet, and assign three-dimensional coordinate position information to each tool;

[0144] The allocation module is used to perform multimodal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data;

[0145] A verification module is used to verify the user's identity identification result and tool use permission data when the user requests to borrow a tool, and generate a tool borrowing instruction and borrowing record data;

[0146] A detection module is used to detect the status of the returned tool when the user requests to return the tool, obtain the current status information of the tool, and input the current status information of the tool into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data;

[0147] A generation module is used to perform multi-dimensional analysis on borrowing record data and predictive maintenance data, generate tool usage heat maps and target tool management strategies.

[0148] Through the collaboration of the above components, a comprehensive tool database including tool type, specification size, usage status, priority index, standard service life and wear index was constructed through optical scanning, usage history analysis, life data analysis and other methods. The hierarchical storage scheme of the tool cabinet and the precise positioning of tools were realized by using the three-dimensional space filling algorithm, which improved the space utilization and access convenience. Combined with biometric features such as face, fingerprint and palm vein, accurate identification of users was achieved, and reasonable tool usage rights were assigned to different users through multi-factor weight calculation, which improved the security and flexibility of management. Through comprehensive analysis of various factors, intelligent recommendation and visual guidance of the optimal borrowing tools were achieved, which improved the borrowing efficiency and user experience. The weight sensing technology was used to realize the comprehensive status detection of the returned tools, and the fully connected neural network was used to realize the high-precision prediction of the remaining service life of the tools, which effectively reduced the tool maintenance cost and safety risk. Through multi-dimensional feature extraction, the tool usage heat map and optimized management strategy were generated, which could dynamically adjust the management strategy according to the tool usage and maintenance needs, realizing the continuous optimization and efficiency improvement of tool management, and improving the intelligent level of tool management.

[0149] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0150] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0151] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent tool management method, characterized in that: include: Collect and encode information on multiple tools, and build a tool database containing tool type, specification size, usage status, priority index, standard service life and wear index; specifically including: optically scanning the physical characteristics of multiple tools to obtain a three-dimensional digital model containing tool contour, size and surface texture, and classifying the multiple tools according to the three-dimensional digital model to obtain tool type information and specification size data; analyzing the usage history data of the multiple tools to obtain usage status indicators, and performing multi-factor weighted calculation on the multiple tools according to the usage status indicators to obtain a priority index; performing life data reliability analysis on the multiple tools to obtain a standard usage life estimation value, and according to the tool material characteristics and expected use conditions, finite element analysis is performed on the plurality of tools to obtain a theoretical wear model, and at the same time, the theoretical wear model is operated and analyzed to obtain a wear index probability distribution; a unique identification code is assigned to each tool according to the tool type information, and the identification code is RFID-encoded to obtain electronic tag information that can be recognized by the tool cabinet; the tool type information, the specification size data, the use status index, the priority index, the standard service life estimation value and the wear index probability distribution are structured to obtain data entries, and a tool database is constructed according to the data entries and the electronic tag information; Performing spatial planning on the tool cabinet according to the tool information in the tool database to obtain a hierarchical storage plan for the tool cabinet, and assigning three-dimensional coordinate position information to each tool; Perform multi-modal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data; When a user requests to borrow a tool, the user identification result and the tool use authority data are verified, and a tool borrowing instruction and borrowing record data are generated; When a user requests to return a tool, the returned tool is checked to obtain current status information of the tool, and the current status information of the tool is input into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data; A multi-dimensional analysis is performed on the borrowing record data and the predictive maintenance data to generate a tool usage heat map and a target tool management strategy.

2. The intelligent tool management method according to claim 1, characterized in that: The space planning of the tool cabinet is performed according to the tool information in the tool database to obtain a hierarchical storage scheme of the tool cabinet, and three-dimensional coordinate position information is assigned to each tool, including: Performing cluster analysis on the priority indexes in the tool database to obtain importance grouping results, and hierarchically dividing the tool cabinets according to the importance grouping results to obtain a preliminary hierarchical scheme for the tool cabinets; Performing time series analysis on the usage status indicators in the tool database to obtain a usage frequency prediction model, and sorting the multiple tools according to the usage frequency prediction model to obtain a storage priority order of the tools in each layer; Performing three-dimensional space filling on the specification size data to obtain an optimal tool arrangement scheme in each layer, and assigning three-dimensional coordinates to each tool according to the optimal tool arrangement scheme to obtain three-dimensional coordinate position information; A conflict detection is performed on the three-dimensional coordinate position information to obtain evaluation results of space utilization and access convenience, and an optimal tool arrangement plan in each layer is iteratively optimized based on the evaluation results to obtain a hierarchical storage plan for the tool cabinet.

3. The intelligent tool management method according to claim 2, characterized in that: The multi-modal identification and permission allocation of user information to obtain user identity identification results and tool usage permission data includes: Perform deep convolutional neural network processing on the user's face image to obtain a face feature vector, and perform feature extraction and pattern matching on the user's fingerprint image to obtain fingerprint feature data; Performing vascular texture analysis and feature point extraction on the palm vein image of the user to obtain a vein feature template, and performing multimodal feature fusion based on the face feature vector, the fingerprint feature data and the vein feature template to obtain a comprehensive biometric feature identification; Performing database matching on the comprehensive biometric identification to obtain a user identification result, and extracting the user's job position and skill level information from the user information database according to the user identification result to perform weight calculation to obtain a basic authority coefficient; Performing data mining on the user's historical tool usage records to obtain a tool usage proficiency score, and performing logical reasoning based on the basic authority coefficient and the tool usage proficiency score to obtain the user's level of authority to use different types of tools; Each tool in the tool database is classified by security level to obtain a use risk assessment result, and authority matching is performed according to the user's use authority level and the tool use risk assessment result to obtain tool use authority data.

4. The intelligent tool management method according to claim 3, characterized in that: When a user requests to borrow a tool, the user identification result and the tool use authority data are verified, and a tool borrowing instruction and borrowing record data are generated, including: When a user requests to borrow a tool, structured borrowing intention data including the target tool type and the estimated usage time is obtained, and the tool database is queried and matched according to the structured borrowing intention data to obtain a list of available tools; Cross-verify the user identification result and the tool use permission data to obtain a user's operation permission confirmation result for the target tool, and filter the available tool list according to the operation permission confirmation result to obtain a subset of tools that the user can borrow; Performing a current status evaluation on each tool in the tool subset to obtain real-time availability information of the tool, and performing multi-objective optimization based on the real-time availability information and the priority index to obtain an optimal tool borrowing recommendation; The user confirms the optimal tool borrowing suggestion to obtain a tool borrowing selection, generates a tool positioning instruction according to the tool borrowing selection, controls an LED indication system of the tool cabinet, and obtains a visual guidance signal; The tool positioning instruction and the visual guidance signal are combined to obtain a tool borrowing instruction sequence, and the tool borrowing instruction sequence, user-tool association information, borrowing time and expected return time are integrated to obtain borrowing record data.

5. The intelligent tool management method according to claim 1, characterized in that: When the user requests to return the tool, the returned tool is checked to obtain the current status information of the tool, and the current status information of the tool is input into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data, including: When a user requests to return a tool, the RFID tag of the returned tool is scanned to obtain a unique identification code; querying the tool database according to the unique identification code to obtain historical status information and standard parameters of the returned tool; Perform weight sensing on the returned tool to obtain current weight data, compare the current weight data with the standard weight to obtain weight deviation value, perform three-dimensional scanning on the returned tool to obtain current geometric model, compare the current geometric model with the standard geometric model to obtain deformation degree data; Perform multispectral imaging on the surface of the returned tool to obtain a material feature map of the tool surface, and perform image analysis on the material feature map of the tool surface to obtain wear and corrosion degree data; Perform a multi-dimensional status assessment based on the weight deviation value, the deformation degree data, and the wear and corrosion degree data to obtain the current status information of the tool, and perform a usage intensity calculation on the usage time, usage frequency, and working environment data of the returned tool to obtain a usage intensity coefficient; Inputting the current status information of the tool and the usage intensity coefficient into a preset predictive maintenance model to obtain a predicted value of the remaining service life of the returned tool; According to the remaining service life prediction value, corresponding tool maintenance suggestions are queried from a maintenance strategy database, and data integration is performed on the tool current status information, the remaining service life prediction value and the maintenance suggestions to obtain predictive maintenance data.

6. The intelligent tool management method according to claim 5, characterized in that: The inputting of the current status information of the tool and the usage intensity coefficient into a preset predictive maintenance model to obtain a predicted value of the remaining service life of the returned tool includes: Performing feature extraction processing on the current state information of the tool to obtain a state feature vector including a weight deviation value, a deformation degree, and a wear and corrosion degree; Normalizing the usage intensity coefficient to obtain a standardized usage intensity index, and concatenating the state feature vector and the standardized usage intensity index to obtain an input feature vector; Inputting the input feature vector into a preset predictive maintenance model, the predictive maintenance model comprising: a first layer of fully connected neural network, a second layer of fully connected neural network, an attention mechanism, a long short-term memory network and a fully connected layer; Performing a first-layer fully connected neural network processing on the input feature vector to obtain a first hidden layer output, and performing a second-layer fully connected neural network processing on the hidden layer output to obtain a second hidden layer output; Performing attention mechanism processing on the hidden layer output to obtain weighted feature representation, and inputting the weighted feature representation into a long short-term memory network to capture the trend of tool status changing over time and obtain time series features; Performing full connection layer processing on the time series features to obtain a probability distribution of the remaining useful life; An expectation calculation is performed on the probability distribution to obtain a point estimate of the remaining useful life, and a confidence interval calculation is performed on the point estimate and the probability distribution to obtain a predicted value of the remaining useful life.

7. The intelligent tool management method according to claim 6, characterized in that: The step of performing a first layer of fully connected neural network processing on the input feature vector to obtain a first hidden layer output, and performing a second layer of fully connected neural network processing on the hidden layer output to obtain a second hidden layer output includes: Performing dimensional analysis on the input feature vector to obtain a feature dimension, and initializing a first-layer weight matrix according to the feature dimension and a preset number of first hidden layer neurons in the first-layer fully connected neural network to obtain an initial weight matrix; Performing Xavier initialization processing on the initial weight matrix to obtain a first-layer optimized weight matrix, and performing a matrix multiplication operation on the first-layer optimized weight matrix and the input feature vector to obtain a first-layer linear transformation result; Performing ReLU activation function processing on the first layer linear transformation result to obtain the first hidden layer output; Initialize and perform Xavier optimization processing on the second layer weight matrix according to the dimension of the first hidden layer output and the number of second hidden layer neurons preset in the second layer fully connected neural network to obtain a second layer optimized weight matrix; A matrix multiplication operation is performed on the second-layer optimized weight matrix and the first hidden layer output to obtain a second-layer linear transformation result, and a LeakyReLU activation function and a dropout regularization process are performed on the second-layer linear transformation result to obtain a second hidden layer output.

8. The intelligent tool management method according to claim 5, characterized in that: The multi-dimensional analysis of the borrowing record data and the predictive maintenance data to generate a tool usage heat map and a target tool management strategy includes: Performing a time series analysis on the borrowing record data to obtain a usage frequency curve and a periodic pattern of each tool, and performing a cluster analysis on the tools according to the usage frequency curve and the periodic pattern to obtain tool groups with different usage patterns; Performing probability distribution analysis on the remaining service life prediction values ​​in the predictive maintenance data to obtain a life risk assessment index for each tool, and performing weighted calculation based on the life risk assessment index and the tool importance index to obtain a comprehensive maintenance priority ranking for the tools; Performing social network analysis on the user-tool association information in the borrowing record data to obtain a tool usage association network diagram, and performing spatial correlation analysis based on the association network diagram and the three-dimensional coordinate position information to obtain an optimized tool proximity placement suggestion; Performing multi-dimensional feature extraction on the borrowing record data and the predictive maintenance data to obtain a high-dimensional feature vector reflecting tool usage and status, and performing dimensionality reduction and visualization on the high-dimensional feature vector to obtain a tool usage heat map; A multi-objective optimization is performed based on the tool usage heat map, the comprehensive maintenance priority ranking and the optimized tool proximity placement suggestion to obtain a target tool management strategy.

9. An intelligent tool management system, characterized in that: The method for executing the intelligent tool management method according to any one of claims 1 to 8 comprises: A collection module is used to collect and encode information of multiple tools and build a tool database containing tool type, specification size, usage status, priority index, standard service life and wear index; A planning module, used for performing space planning on the tool cabinet according to the tool information in the tool database, obtaining a hierarchical storage plan for the tool cabinet, and assigning three-dimensional coordinate position information to each tool; The allocation module is used to perform multimodal recognition and authority allocation on user information to obtain user identity recognition results and tool usage authority data; A verification module, used for verifying the user identification result and the tool use authority data when the user requests to borrow a tool, and generating a tool borrowing instruction and borrowing record data; A detection module, for, when a user requests to return a tool, performing a status detection on the returned tool to obtain current status information of the tool, and inputting the current status information of the tool into a preset predictive maintenance model for life evaluation to obtain predictive maintenance data; A generation module is used to perform multi-dimensional analysis on the borrowing record data and the predictive maintenance data, and to generate a tool usage heat map and a target tool management strategy.

Images