3D digital factory equipment management system and method

Abstract

The invention relates to the technical field of equipment management, in particular to a 3D digital factory equipment management system and method, and provides the following scheme: obtaining current structure information of equipment, and constructing a high-precision three-dimensional visual model through a laser scanning and image feature fusion technology; the operation data of the equipment is acquired through multiple sensors, and the three-dimensional model is linked, so that the state of the equipment is monitored in real time; a dynamic baseline concentration model is generated in combination with a time sequence processing and spatial interpolation method, and the hydrogen leakage risk is subjected to multi-dimensional evaluation; and optimizing operation parameters of the production line by utilizing multi-equipment correlation analysis. The system integrates a three-dimensional modeling module, a data monitoring module, a data analysis module, an anomaly detection module and an operation optimization module, and can comprehensively improve the equipment management efficiency, safety and process optimization capability. The method overcomes the problems of insufficient modeling precision, isolated monitoring data and weak dynamic risk assessment capability in the prior art, and is suitable for high-risk production scenes such as hydrogen storage tanks and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment management, and in particular to a 3D digital factory equipment management system and method. Background Art

[0002] The cross-integration of intelligent operation and maintenance theory and hydraulic technology, combined with the characteristics of mobile equipment systems such as a wide distribution of operating points, mobile operations, and harsh operating environments, has made it a common scientific problem in the equipment manufacturing industry to achieve system status monitoring and intelligent operation and maintenance while ensuring that the equipment meets operating requirements.

[0003] For example, a Chinese patent with the authorization announcement number CN104614247B discloses a visual triaxial test system, wherein a transparent pressure chamber (5) is supplied with oil by a manual pump (17), a hydraulic cylinder (7) is arranged directly above the transparent pressure chamber (5), the hydraulic cylinder (7) is supplied with oil by a variable plunger device (34), a loading head (8) is arranged at the lower end of the piston of the hydraulic cylinder (7), the loading head (8) is directly opposite to the pressure rod of the transparent pressure chamber (5), a linear differential transformer (9) is arranged directly above the hydraulic cylinder (7), the probe of the linear differential transformer (9) is in contact with the upper end of the piston of the hydraulic cylinder (7); four cameras (11) and three video cameras (12) are arranged on the periphery of the transparent pressure chamber (5). The invention has a simple and compact structure, is easy to assemble, has a low cost and good stability; the hardware-controlled test has high precision, good reliability and long holding time; the radial deformation of the specimen can be visually observed, and the radial deformation of the specimen can be more accurately measured by shooting equipment combined with image processing system software.

[0004] The above-mentioned prior art has the problems raised by this background technology: the working conditions of the production line equipment are complex and changeable, the operating differences between different individuals of the same model are large, it is difficult to describe the working status of the equipment with a mathematical model, and the large number of monitoring signals generated during the equipment operation stage show multi-source, multi-dimensional and multi-modal characteristics. An accurate and effective method is needed to analyze and process the signals. In order to solve the above problems, this application designs a 3D digital factory equipment management system and method. Summary of the invention

[0005] The technical problem to be solved by the present invention is to address the deficiencies of the prior art, and to provide a 3D digital factory equipment management system and method, obtain the current structural information of the equipment, and build a high-precision three-dimensional visualization model through laser scanning and image feature fusion technology; collect the operating data of the equipment through multiple sensors and link the three-dimensional model to monitor the equipment status in real time; combine time series processing and spatial interpolation methods to generate a dynamic baseline concentration model, and conduct a multi-dimensional assessment of the risk of hydrogen leakage; and optimize the production line operating parameters using multi-equipment correlation analysis. The system integrates a three-dimensional modeling module, a data monitoring module, a data analysis module, an anomaly detection module, and an operation optimization module, which can comprehensively improve the equipment management efficiency, safety, and process optimization capabilities. The present invention overcomes the problems of insufficient modeling accuracy, isolated monitoring data, and weak dynamic risk assessment capabilities in the prior art, and is suitable for high-risk production scenarios such as hydrogen storage tanks.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A 3D digital factory equipment management method, the method comprising:

[0008] Obtain the current structural information of each device in the hydrogen storage tank production line;

[0009] Processing the current structural information, digitally modeling the hydrogen storage tank production line through 3D modeling software, generating a three-dimensional model of the production line, and constructing a visual model of the production line;

[0010] Collect the operating data of each device through multiple sensors, and link the three-dimensional model of the production line to monitor the working status of the equipment;

[0011] Analyze the equipment operation status in the production process based on the visually displayed data;

[0012] Assess the risk of hydrogen leakage during operation based on the visually displayed data;

[0013] Based on the analysis and evaluation results, measures to optimize the production line are formulated through optimization strategies.

[0014] The obtaining of the current structural information of each device in the hydrogen storage tank production line includes:

[0015] The hydrogen storage tank production line is scanned in sections by a laser scanning device to obtain a laser point cloud, wherein the scanning angle of each area is adjusted according to the complexity and occlusion characteristics of the equipment;

[0016] Determine the geometric characteristics of the equipment according to the laser point cloud, plan the acquisition path of the image sensor according to the geometric characteristics, and perform image acquisition on the hydrogen storage tank production line according to the acquisition path, wherein the acquisition path includes:

[0017] For welding equipment, collect texture images of the welding head and welding area surface;

[0018] For cleaning equipment, collect gloss images of nozzles, liquid flow paths, and housing surfaces;

[0019] For heat treatment equipment, collect structural images inside the furnace;

[0020] For servo motors, capture trajectory images of bearings and couplings.

[0021] The construction of a visual model of the production line includes:

[0022] Preprocessing the laser point cloud, determining the spatial position and geometric characteristics of each device in the production line according to the preprocessed point cloud data, and generating a preliminary three-dimensional model;

[0023] Extract image features of texture images, gloss images, structure images and trajectory images;

[0024] Matching feature points of the image features with the preliminary three-dimensional model, optimizing the preliminary three-dimensional model according to the matching results, and constructing a visualization model, wherein the optimization includes:

[0025] Optimize welding detail features of welding joints and welding areas based on texture image features;

[0026] Optimizing the cleaning efficiency characteristics of the nozzle, liquid flow path and housing surface based on the gloss image characteristics;

[0027] Optimize the heat distribution characteristics inside the furnace based on structural image features;

[0028] Optimize the stability characteristics of servo motor bearings and couplings based on trajectory image features.

[0029] The step of extracting image features of a texture image, a gloss image, a structure image and a trajectory image comprises:

[0030] Preprocessing the texture image, extracting the texture features of the welding surface through the local binary pattern, extracting the edge direction features of the welding area through the gradient direction histogram, combining the frequency characteristics of the texture features and the edge direction features through Fourier transform, and calculating the texture image features;

[0031] Preprocessing the gloss image, calculating the surface reflectivity of the gloss image according to a light reflection model, extracting the reflection characteristics of the nozzle and the liquid flow path, calculating the highlight distribution characteristics by analyzing the light spot distribution of the gloss image, and calculating the cleaning efficiency characteristics according to the reflection characteristics and the highlight distribution characteristics;

[0032] Preprocessing the structural image, extracting corner features of the structural image, processing the corner features according to the SURF algorithm, and calculating feature matching points of the furnace inner wall and the heating element;

[0033] The trajectory image is preprocessed, and the trajectory path of the servo motor bearing and the coupling during the movement is extracted by an optical flow algorithm, and the radial runout of the bearing and the synchronous offset of the coupling are calculated according to the trajectory path.

[0034] The specific steps for analyzing the equipment operation status in the production process are as follows:

[0035] The operating data of each device in the hydrogen storage tank production line is collected through multiple sensors, including temperature, current, voltage and vibration data, and dynamic feature extraction is performed on the collected data;

[0036] By establishing an adaptive anomaly detection model, the extracted features are analyzed in real time to determine whether the working status of the equipment is normal;

[0037] The long short-term memory network prediction model is used to predict the trend of key parameters of the equipment and evaluate the remaining life of the equipment.

[0038] The specific steps of extracting dynamic features from the collected data are as follows:

[0039] Collect temperature, current and voltage data from welding equipment, cleaning equipment, heat treatment equipment and servo motors, extract their change curves through time series analysis methods, construct temperature gradient diagrams, current fluctuation diagrams and voltage fluctuation diagrams of the equipment, and monitor the temperature, current and voltage stability of the equipment during the production process;

[0040] The short-time Fourier transform and wavelet transform techniques are used to perform frequency domain analysis on the equipment vibration signal, extract the characteristic frequency, amplitude, energy distribution and acceleration indicators in the vibration signal, and construct the equipment vibration state spectrum.

[0041] The risk assessment of hydrogen leakage during operation includes:

[0042] Hydrogen concentration sensors are placed at the welding interfaces, pipeline connections and valve areas of hydrogen storage tank production lines to collect hydrogen concentration data in the equipment operating environment;

[0043] Performing time series smoothing and spatial interpolation processing on the hydrogen concentration data to obtain standardized concentration data;

[0044] Constructing a baseline concentration model, generating a reference concentration distribution according to the hydrogen concentration data under normal operating conditions of the hydrogen storage tank production line, wherein the reference concentration distribution is set as a concentration threshold 1 and a concentration threshold 2 according to the operating parameters of the hydrogen storage tank production line;

[0045] Comparing the hydrogen concentration data with the reference concentration distribution, and calculating a weighted value based on historical hydrogen leakage conditions;

[0046] The hydrogen leakage risk rate is calculated according to the comparison result and the weighted value, wherein the calculation formula of the hydrogen leakage risk rate is:

[0047]

[0048] Where L represents the hydrogen leakage risk rate, j represents a single historical hydrogen leakage, n represents the total number of historical hydrogen leakages, η represents the hydrogen leakage safety factor at the collection node, and L j represents the hydrogen leakage risk rate of the jth historical hydrogen leakage, G k represents the mean hydrogen concentration of the acquisition node within the acquisition time range, M1 represents the concentration threshold 1, M2 represents the concentration threshold 2, max(·) represents the maximum value function, ΔG represents the rate of change of the hydrogen concentration of the acquisition node within the acquisition time range, G t Indicates the concentration change threshold.

[0049] The specific steps of the optimization strategy are as follows:

[0050] Use the correlation matrix to analyze the relationship between devices and calculate the mutual impact of multiple devices working together;

[0051] Based on the results of multi-equipment correlation analysis, provide production process optimization suggestions, including optimizing welding equipment temperature to improve welding quality, adjusting cleaning equipment flow to improve cleaning effect, and optimizing servo motor output pressure to improve the overall working efficiency and safety of the equipment;

[0052] The diffusion process of hydrogen leakage is simulated by a fluid mechanics diffusion model, and the diffusion path of hydrogen in the production line is predicted based on the location of the leakage source, the concentration change rate and the equipment layout;

[0053] The diffusion analysis results are mapped into a 3D visualization model, using color gradients to show the leakage concentration levels in different areas.

[0054] 3D digital factory equipment management system, the system includes a three-dimensional modeling module, a data monitoring module, a data analysis module, an anomaly detection module and an operation optimization module;

[0055] The three-dimensional modeling module is configured with a production line modeling strategy, and the production line modeling strategy is used to generate a three-dimensional visualization model of the hydrogen storage tank production line;

[0056] The data monitoring module is used to collect and monitor various equipment data of the hydrogen storage tank production line in real time;

[0057] The data analysis module is configured with a data analysis strategy, which is used to perform dynamic feature extraction and association analysis on the collected data;

[0058] The abnormality detection module is used to detect abnormal conditions of the equipment and generate an early warning signal;

[0059] The operation optimization module provides optimization suggestions for operation parameters based on the equipment operation status analysis results.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1. The present invention combines laser point cloud and image feature fusion technology to achieve accurate modeling of complex equipment in the hydrogen storage tank production line, especially greatly improving the modeling accuracy in key details of welding areas, cleaning equipment and heat treatment equipment, providing a reliable foundation for the visual management of the production line.

[0062] 2. Construct a dynamic baseline concentration model, combine time series smoothing, spatial interpolation and concentration change rate analysis methods to conduct real-time assessment and dynamic adjustment of hydrogen leakage risks, accurately identify leakage areas, and avoid false alarms and missed alarms of traditional fixed threshold methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0064] Figure 1 This is a flow chart of a 3D digital factory equipment management method in Embodiment 1 of the present invention;

[0065] Figure 2 This is a schematic diagram of a corner feature determination area according to Embodiment 1 of the present invention;

[0066] Figure 3 This is a schematic diagram of the structure of an adaptive anomaly detection model in Example 1 of the present invention;

[0067] Figure 4 This is a module diagram of the 3D digital factory equipment management system in Example 2 of the present invention. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0069] Embodiment 1:

[0070] See also Figure 1, an embodiment provided by the present invention: a 3D digital factory equipment management method, comprising the following steps;

[0071] S1: Obtain the current structural information of each device in the hydrogen storage tank production line;

[0072] In this step, the hydrogen storage tank production line is scanned in sections using high-precision laser scanning equipment, and the scanning angle and parameters are dynamically adjusted based on the complexity and occlusion characteristics of the equipment to obtain laser point cloud data containing all the geometric details of the equipment. Subsequently, for different types of equipment (such as welding equipment, cleaning equipment, etc.), the sensor acquisition path is planned, and high-resolution image acquisition of texture, gloss and structural features is further utilized to achieve accurate data capture of key equipment areas.

[0073] S2: Processing the current structural information to construct a visualization model of the production line;

[0074] In this step, the laser point cloud is first preprocessed, including noise removal, point cloud thinning and registration, and key geometric feature points are extracted according to the equipment classification. Then, the image features of the equipment are fused with the point cloud model through the feature point matching algorithm by combining the features of the texture image, gloss image and structure image, and the accuracy and details of the 3D model are optimized. Through the above steps, a real-time interactive production line visualization model is constructed, which enables managers to intuitively understand the layout and equipment status of the production line, thereby improving the management efficiency of the production line.

[0075] S3: Collect the operating data of each device through multiple sensors, and link the three-dimensional model of the production line to monitor the working status of the equipment;

[0076] In this step, multiple types of sensors such as temperature, current, voltage and vibration are arranged on the key equipment of the production line (such as welding equipment, cleaning equipment, heat treatment equipment and servo motors) to collect the operating status data of the equipment in real time. At the same time, these sensor data are linked with the three-dimensional visualization model through the network, and the dynamic operating status of each device is intuitively presented in the model, so as to realize real-time monitoring of the operating status of the production line equipment and quickly locate the faulty equipment when an abnormality is found.

[0077] S4: Analyze the equipment operation status in the production process based on the visually displayed data;

[0078] In this step, by building an adaptive analysis model of the equipment's operating status, the collected equipment data is subjected to dynamic feature extraction, including temperature gradients, current fluctuations, and vibration status, and trend analysis is performed in combination with the equipment's historical data and operating conditions. The long short-term memory network (LSTM) model is used to predict key equipment parameters and assess possible failure risks and the remaining life of the equipment in advance. In this way, not only is the accurate analysis of the equipment status achieved, but the level of intelligent production line management is also significantly improved, effectively avoiding the problem of lagging equipment status assessment and relying on manual judgment in traditional methods.

[0079] S5: Assess the risk of hydrogen leakage during operation based on the visually displayed data;

[0080] In this step, hydrogen concentration data is collected in real time through hydrogen concentration sensors arranged at welding interfaces, pipeline connections and hydrogen storage tank valve areas, and stable standardized concentration data is generated through time series smoothing and spatial interpolation processing. Combined with the baseline concentration model, the concentration threshold range (threshold one and threshold two) under normal operating conditions is set, and the real-time data is compared with the baseline model, and the leakage risk rate is calculated through historical weighting and dynamic rate analysis. This can accurately identify potential risk areas for hydrogen leakage, and further combine with the three-dimensional visualization model to intuitively present the risk diffusion path, greatly improving safety management capabilities and overcoming the problems of slow response and difficult positioning of traditional leak detection.

[0081] S6: Based on the analysis and evaluation results, formulate measures to optimize the production line through optimization strategies;

[0082] In this step, the collaborative working relationship between the equipment is analyzed by constructing a correlation matrix, and the key process parameters are optimized based on the correlation of the operating status of multiple equipment. For example, the temperature of the welding equipment is dynamically adjusted to improve the welding quality, the flow of the cleaning equipment is optimized to improve the cleaning efficiency, and the output pressure of the servo motor is accurately controlled to ensure the overall performance of the equipment. At the same time, the optimized safety control strategy is generated by combining the three-dimensional visualization model with the simulation results of hydrogen leakage and diffusion to further guide the operation and scheduling of the equipment. This can not only effectively improve the overall efficiency of the production line, but also significantly improve the safety of the equipment, solving the lack of data support and real-time response capabilities in traditional optimization solutions.

[0083] Specifically, although collecting equipment data through laser scanning or image sensors is a commonly used 3D modeling method, which can realize the monitoring of industrial production lines to a certain extent, there are still many shortcomings in complex production environments. For example, the complex geometric structure and occlusion characteristics of the equipment often lead to incomplete data collection, making it difficult to capture key details of the equipment; in stacked, staggered or space-constrained scenarios, redundancy or errors may occur between data, affecting the accuracy of modeling; on the other hand, it is difficult for the data collected by multiple sensors to accurately correspond to the actual position of the equipment, especially in data analysis, it is impossible to dynamically link the equipment status and efficiently detect equipment abnormalities. In addition, in terms of leakage risk assessment, existing technologies mostly rely on static threshold judgments and cannot dynamically analyze the diffusion path and risk level of the leakage, resulting in untimely response and major safety hazards.

[0084] Furthermore, in this embodiment, laser scanning and image sensor technology are combined to design dedicated acquisition paths and data processing algorithms for different devices (such as welding heads of welding equipment, nozzles of cleaning equipment, etc.), ensuring that detailed information of complex equipment can be fully captured. For example, traditional laser scanning technology is difficult to capture fine surface features of welding areas, while this patent combines feature extraction technology of texture images and gloss images to not only achieve high-precision modeling of the geometric characteristics of welding heads, but also optimize the welding details of the weld area, significantly improving the accuracy and availability of three-dimensional models. On the other hand, a real-time updated baseline concentration model is constructed in this embodiment to effectively solve the defect of false alarms caused by overly fixed threshold judgments in the prior art. For example, in cases where the risk of leakage is low but the concentration change rate is fast, traditional systems may ignore potential risks, while this embodiment achieves early identification of potential leakage areas through sensitive capture of dynamic rates.

[0085] The specific steps of S1 are as follows:

[0086] S1.1: Scan the hydrogen storage tank production line in sections using a laser scanning device to obtain a laser point cloud, wherein the scanning angle of each area is adjusted according to the complexity and occlusion characteristics of the equipment;

[0087] Furthermore, combined with the pre-collected equipment layout data (such as the overall model of the production line or CAD drawings), angle calculation is performed based on the complexity of the equipment surface and the characteristics of the occluded area. For high-complexity equipment such as welding equipment, a multi-angle scanning strategy is pre-set, and the laser scanning equipment repeatedly scans the same area from multiple angles (such as 45°, 90°, 135°, etc.), capturing complex surface features such as the grooves of the welding head and the surface undulations of the weld by switching angles. Complexity parameters are calculated based on the geometric characteristics of the equipment. For example, by quickly analyzing the surface curvature of the point cloud data, high curvature areas are marked as priority scanning areas;

[0088] For equipment with obstructions, such as the connection of hydrogen tank pipes and the inside of servo motors, the scanning path is planned as a multi-level structure. Specifically, the angle of the scanning equipment is dynamically adjusted using the prior equipment layout data so that the laser beam bypasses the obstructed area and captures a complete point cloud. For example, when the intensity of the laser scanning return signal is detected to be weakened, the scanning direction is automatically adjusted to avoid information loss.

[0089] This dynamic adjustment mechanism can achieve complete point cloud data acquisition in complex and blocked areas of the device by adjusting the scanning angle in real time, solving the problem of information loss or insufficient accuracy caused by fixed angles in traditional scanning methods. At the same time, partition scanning avoids the tedious process of preliminary low-resolution scanning and directly generates a high-precision point cloud model, improving scanning efficiency and data quality.

[0090] S1.2: Determine the geometric characteristics of the equipment according to the laser point cloud, plan the acquisition path of the image sensor according to the geometric characteristics, and perform image acquisition on the hydrogen storage tank production line according to the acquisition path, wherein the acquisition path includes:

[0091] For the welding head and welding area of ​​the welding equipment, a high-resolution industrial camera is used to collect texture images, and the acquisition path perpendicular to the welding head surface ensures that the details of the weld are fully captured. This path combines the curvature changes of the laser point cloud with the geometric shape of the welding area.

[0092] For the nozzle, liquid flow path and shell surface of the cleaning equipment, a multi-angle acquisition path is used to capture the gloss image. The light reflection model is used to optimize the path so that the relative angle between the light source and the camera reaches the best position, thereby clearly presenting the gloss characteristics of the surface and the liquid flow trajectory, and improving the visual analysis ability of the cleaning efficiency.

[0093] For the interior of the furnace of the heat treatment equipment, an industrial camera with a high dynamic range (HDR) function is used. The point cloud data is used to guide the camera's path deep into the furnace to capture structural images and heat distribution characteristics, ensuring that details of key parts of the furnace (such as heating elements and insulation layers) are not missed.

[0094] For the bearings and couplings of the servo motor, trajectory images are collected in real time. The path planning is based on the principle of being parallel to the bearing rotation direction, capturing data on dynamic motion states to provide support for evaluating the stability and operating efficiency of the equipment.

[0095] Furthermore, by calculating the curvature of the laser point cloud data, the key geometric feature points on the equipment surface are extracted, such as the edge points of the welding head, the cylindrical boundary points of the cleaning equipment nozzle, and the key connection points inside the furnace of the heat treatment equipment. The points with prominent curvature among the key geometric feature points are clustered into feature areas and marked as target acquisition areas.

[0096] The image sensor acquisition angle and path are calculated based on the shape features of the target acquisition area. For example, for the welding head area of ​​the welding equipment, the path is planned to be perpendicular to the weld surface as the priority direction to ensure the complete details in the texture image; for the nozzle and liquid flow path of the cleaning equipment, a multi-angle surround acquisition path is planned to capture the surface gloss distribution and liquid flow characteristics.

[0097] Specifically, the image sensor can accurately collect image data of key areas based on the geometric characteristics of the device, avoiding the problems of low collection efficiency and lack of details caused by unreasonable paths in traditional methods.

[0098] The specific steps of S2 are as follows:

[0099] S2.1: preprocessing the laser point cloud, determining the spatial position and geometric characteristics of each device in the production line according to the preprocessed point cloud data, and generating a preliminary three-dimensional model;

[0100] The preprocessing of laser point cloud includes denoising, point cloud thinning and point cloud registration. Through the preprocessed point cloud data, the spatial position, outline and geometric characteristics of each equipment in the hydrogen storage tank production line are accurately determined, avoiding the redundancy and inconsistency problems in the original point cloud data.

[0101] S2.2: Extract image features of texture image, gloss image, structure image and trajectory image;

[0102] S2.3: matching feature points of the image features with the preliminary three-dimensional model, optimizing the preliminary three-dimensional model according to the matching results, and constructing a visualization model, wherein the optimization includes:

[0103] Optimize welding detail features of welding joints and welding areas based on texture image features;

[0104] Optimizing the cleaning efficiency characteristics of the nozzle, liquid flow path and housing surface based on the gloss image characteristics;

[0105] Optimize the heat distribution characteristics inside the furnace based on structural image features;

[0106] Optimize the stability characteristics of servo motor bearings and couplings based on trajectory image features.

[0107] Furthermore, the Euclidean distance between image features and point cloud features is calculated through the fast neighbor search algorithm to determine the best matching pair. At the same time, RANSAC is used to eliminate incorrect matching pairs and retain high-confidence matching point pairs. The fast neighbor search algorithm improves the efficiency of high-dimensional feature matching by constructing a neighbor search tree. RANSAC fits the best transformation model by iteratively randomly sampling point pairs, while eliminating incorrect matching points that do not conform to the model.

[0108] The surface shape and weld boundary of the welding head and welding area are corrected by matching the texture features, and the missing details in the preliminary model are supplemented. According to the distribution of reflection features of the gloss image, the geometric accuracy of the nozzle and liquid flow path is optimized, and the smoothness of the shell surface is adjusted. Combined with the inner wall corner point features, the 3D structural model inside the furnace is corrected, and the key areas of heat distribution are marked in the model. The radial runout model of the bearing is adjusted according to the dynamic trajectory data, and the synchronous offset model of the coupling is corrected to ensure the stability of the dynamic part.

[0109] Specifically, the 3D model optimization is based on the displacement vector of feature point matching, which converts the geometric position information of the image features into the adjustment parameters of the model, thereby achieving shape correction and accuracy improvement of the model. The generated visualization model not only fully presents the spatial layout and geometric characteristics of each device on the production line, but also combines the dynamic characteristics of the equipment in operation, so that the model can dynamically reflect the status and performance of the equipment.

[0110] The specific steps of S2.2 are as follows:

[0111] S2.2.1: Preprocessing the texture image, including grayscale, denoising and contrast enhancement, to eliminate illumination non-uniformity and noise interference during the shooting process, extracting texture features of the welding surface through local binary patterns for analyzing the roughness and texture direction distribution of the weld area, extracting edge direction features of the welding area through gradient direction histograms to capture the edge integrity and shape characteristics of the weld, combining the frequency characteristics of the texture features and edge direction features through Fourier transform, and calculating texture image features;

[0112] Specifically, the texture and edge features of the welding area are directly related to the welding quality. By extracting and analyzing these features, the stability and consistency of welding can be accurately evaluated, thereby avoiding equipment failure caused by weld defects. Although the laser point cloud can capture the geometric shape of the welding area, its data is essentially a set of discrete three-dimensional coordinate points, which is difficult to accurately reflect the texture details of the welding surface and the edge characteristics of the weld. For example, in practical applications, the recognition of the weld edge by the laser point cloud may be blurred or transitionally smooth, resulting in the difficulty in accurately capturing the integrity and details of the weld edge. By introducing the preprocessing and feature extraction methods of texture images, it can not only make up for the shortcomings of laser point clouds in capturing fine features in the welding area, but also establish multimodal correlation analysis between texture images and point cloud data. For example, the roughness features extracted from the texture image can help determine the welding uniformity of the weld surface, while the edge direction features can accurately characterize the integrity and continuity of the weld boundary shape. This information is crucial for evaluating the welding quality. In addition, the frequency domain fusion of texture and edge features using Fourier transform can further enhance the feature differentiation ability.

[0113] S2.2.2: Enhance the gloss image using high dynamic range processing technology to ensure that accurate reflection characteristics can be captured even when lighting conditions vary greatly. Subsequently, the surface reflectance of the gloss image is calculated based on the light reflection model to extract the surface smoothness and reflection characteristics of the nozzle and liquid flow path. Combined with the morphological analysis method, the highlight area in the gloss image is separated and the highlight distribution characteristics are extracted. By analyzing the density, brightness distribution and shape characteristics of the light spot, the coverage efficiency of the cleaning area is calculated;

[0114] Specifically, the performance evaluation accuracy of cleaning equipment can be significantly improved by extracting and analyzing gloss image features, especially the real-time monitoring capability of the smoothness and cleanliness of nozzles and liquid flow paths. Although laser point clouds can capture the geometric shapes of nozzles and liquid flow paths, they are limited in reflecting the cleanliness and smoothness of the equipment surface because they cannot record the surface reflection characteristics and gloss distribution. For example, the surface of the nozzle may have residual dirt due to insufficient cleaning, but the laser point cloud can only identify the outer contour of the nozzle and cannot effectively reflect the surface dirt distribution and its impact on the cleaning efficiency. In addition, the cleaning effect of the liquid flow path depends on the uniformity of the fluid distribution and the spray coverage. These dynamic characteristics are also difficult to accurately describe through static point cloud data. By supplementing the reflection characteristics and highlight distribution characteristics of the gloss image, the cleaning status of the nozzle and flow path surface can be accurately monitored, and the efficiency of the cleaning process can be dynamically evaluated.

[0115] S2.2.3: Perform image grayscale equalization and edge enhancement processing on the structural image to improve the detection accuracy of corner features; then use the corner detection algorithm to extract the key feature points of the inner wall of the furnace, combine the SURF (Speeded Robust Features) algorithm to match and filter the feature points, further extract the spatial distribution characteristics of the heating element, and generate a relative geometric relationship diagram between the inner wall and the heating element. These features are used to analyze the heat distribution characteristics in the furnace and provide support for the optimization of the heat treatment process;

[0116] Specifically, although the laser point cloud can record the overall geometry of the furnace, its resolution in identifying fine structures is limited, especially in confined spaces where important details are missing due to occlusion or insufficient sampling. For example, the positional offset of the heating element or local damage to the inner wall has an important impact on the uniformity of the heat distribution, but it is difficult for the laser point cloud to capture these details, which may lead to overheating or uneven heating during the heat treatment process. By supplementing the corner point features and matching point analysis of the structural image, the geometric details of the inner wall of the furnace and the relative position relationship of the heating elements can be accurately extracted. The introduction of this feature not only makes up for the lack of details in the point cloud, but also enables the heat distribution in the equipment to be fully quantified and visualized, thereby optimizing the stability of the heat treatment process and the energy efficiency utilization rate.

[0117] See also Figure 2 , a schematic diagram of the judgment area of ​​the corner point feature of the inner wall of the furnace according to an embodiment of the present invention, the steps of extracting the corner point feature are as follows:

[0118] Randomly select any pixel point in the structural image, mark it as point P, and calculate the gray value I of P P ;

[0119] Draw a circle with P as the center and 4 pixels as the radius, and select 20 pixels on the circle boundary;

[0120] Set the gray threshold t. If the gray values ​​of 16 consecutive pixels out of 20 pixels are less than I P -t or greater than I P +t, then point P is determined to be a corner feature of the image.

[0121] S2.2.4: Preprocess the trajectory image, extract the trajectory path of the servo motor bearing and coupling during the movement process through the optical flow algorithm, perform time series analysis on the extracted trajectory path, and calculate the radial runout of the bearing and the synchronous offset of the coupling. These dynamic feature data are further used to generate the dynamic operation state diagram of the servo motor, thereby calculating the trajectory image features.

[0122] Specifically, the operating status of the servo motor (such as the radial runout of the bearing or the offset of the coupling) is a dynamic process, which is difficult to accurately describe through point cloud modeling. For example, when the bearing has a slight runout or the coupling has a small synchronization deviation, the static point cloud cannot perceive these subtle dynamic changes, and these anomalies may become potential failure points during long-term operation of the equipment. By supplementing the dynamic feature analysis of the trajectory image, the dynamic behavior of the equipment can be accurately captured to help identify abnormal conditions during operation. The radial runout can quantify the vibration of the bearing, and the synchronization offset reflects the coordination of the coupling. These data provide a reliable basis for equipment fault warning and maintenance optimization.

[0123] The specific steps for analyzing the equipment operation status in the production process are as follows:

[0124] S4.1: Collect the operating data of key equipment in the hydrogen storage tank production line through multiple sensors, including temperature, pressure, flow and vibration data, and extract dynamic features from the collected data;

[0125] Specifically, after data collection, the data is preprocessed using a dynamic feature extraction algorithm, including noise filtering, time series smoothing, and feature enhancement. The extracted dynamic features include temperature gradients, pressure change rates, flow fluctuations, and characteristic frequencies, amplitudes, and accelerations of vibration signals. The technical advantage of dynamic feature extraction is that it can convert multiple types of data into time series features that are easy to analyze, while retaining key information about the equipment's operating status, providing data support for subsequent analysis.

[0126] S4.2: By establishing an adaptive anomaly detection model, the extracted features are analyzed in real time to determine whether the working status of the equipment is normal;

[0127] See also Figure 3 , structural diagram of the adaptive anomaly detection model of an embodiment of the present invention. Specifically, the adaptive anomaly detection model adopts an architecture based on multi-layer data fusion, including a feature selection layer, an anomaly determination layer and an adaptive update layer:

[0128] The feature selection layer selects the key features most relevant to the equipment operation status from the dynamic features, thereby reducing the interference of redundant data on model judgment and improving the accuracy and efficiency of anomaly detection;

[0129] Furthermore, the mutual information analysis method is used to calculate the correlation score between each dynamic feature and the equipment failure event. Mutual information quantifies the degree of information sharing between the feature and the target variable. The higher the score, the stronger the correlation between the feature and the equipment state. For example, the temperature change rate may have a strong correlation with the equipment overheating failure, while the vibration frequency may have a higher correlation with mechanical wear. At the same time, the correlation between dynamic features is calculated based on the Pearson correlation coefficient, and highly correlated redundant features are eliminated. For example, if the correlation coefficient between the vibration frequency and the vibration amplitude is close to 1, only one feature is retained for subsequent analysis, and the features are sorted according to the feature correlation score. The top N (for example, the top 10) most relevant features are selected as the input feature set. The screening criteria can be adjusted dynamically. For example, when the amount of sensor data increases or the type of equipment changes, the feature selection layer can automatically recalculate the correlation and update the key feature set.

[0130] The abnormality determination layer determines whether the current device status is abnormal and determines the abnormality level based on the support vector machine and time series classification algorithm;

[0131] Furthermore, the "normal boundary" of the normal operating state of the equipment is constructed through the support vector machine algorithm, that is, a hyperplane in a multidimensional feature space, which is used to separate the normal state from the abnormal state. The normal operating data is used to train the model to generate a set of support vectors and hyperplane parameters to define the normal range of feature distribution. For example, the temperature change rate, vibration frequency and other features of the equipment during operation are distributed within the normal boundary, and data points beyond this boundary are judged as abnormal.

[0132] On the basis of SVM, a time series classification algorithm is introduced to further judge the time series change pattern of dynamic features. For example, for abnormal detection of temperature change rate, not only the current value is considered to be beyond the boundary, but also the change trend over the past period of time is combined to determine whether there is an abnormal pattern (for example, a gradually rising temperature may be a sign of overheating failure).

[0133] The adaptive update layer uses an incremental learning algorithm to dynamically adjust the anomaly threshold during operation to ensure that the model can adapt to changes in the operating status of the equipment.

[0134] S4.3: Use the long short-term memory network prediction model to predict the trend of key parameters of the equipment and then evaluate the remaining life of the equipment.

[0135] Specifically, LSTM is a deep learning model suitable for processing time series data. It can remember historical information over a long period of time and predict future trends through a special "gating mechanism". In this embodiment, the LSTM model inputs the time series data of key dynamic features (such as temperature gradient, pressure change rate, vibration amplitude, etc.), and outputs the parameter prediction value for a period of time in the future. On the basis of predicting parameter trends, the LSTM model further evaluates the remaining life of the equipment by combining historical data and equipment characteristic curves (such as welding head aging curves of welding equipment). For example, based on the growth trend of vibration frequency and amplitude, the degree of wear of the servo motor bearing is inferred, and the time range of the estimated remaining life is given.

[0136] The specific steps of extracting dynamic features from the collected data are as follows:

[0137] S4.1.1: Collect temperature, current and voltage data from welding equipment, cleaning equipment, heat treatment equipment and servo motors, and extract their change curves through time series analysis methods, construct temperature gradient diagrams, current fluctuation diagrams and voltage fluctuation diagrams of the equipment, and monitor the temperature, current and voltage stability of the equipment during the production process;

[0138] S4.1.2: Collect flow data of cleaning equipment, decompose flow signals using spectrum analysis methods, extract high-frequency, low-frequency and transient features, and construct flow fluctuation graphs to analyze flow change trends and identify abnormal flow fluctuations during equipment operation;

[0139] S4.1.3: Use short-time Fourier transform and wavelet transform techniques to perform frequency domain analysis on equipment vibration signals, extract key indicators such as characteristic frequency, amplitude, energy distribution and acceleration in the vibration signals, and construct equipment vibration state maps to detect the mechanical operating status of the equipment during the production process and identify potential early fault signals.

[0140] The specific steps of S5 are as follows:

[0141] S5.1: Hydrogen concentration sensors are arranged at the welding interfaces, pipeline connections and valve areas of hydrogen storage tank production lines to collect hydrogen concentration data in the equipment operating environment;

[0142] Specifically, a multi-sensor arrangement is used to form a concentration monitoring network covering the production line, which can accurately capture hydrogen leaks in any area of ​​the production line. In addition, these sensors upload data in real time through high-speed communication modules to ensure that the monitoring system has the ability to respond immediately to concentration changes in the production line operating environment. This design solves the problem of monitoring blind spots caused by insufficient sensor coverage in traditional technologies, and improves the quality and comprehensiveness of hydrogen leakage data collection.

[0143] S5.2: Performing time series smoothing and spatial interpolation processing on the hydrogen concentration data to obtain standardized concentration data;

[0144] Specifically, the collected hydrogen concentration data is smoothed in time series to eliminate abnormal data points caused by sensor noise or instantaneous environmental fluctuations, ensuring the continuity and reliability of the data. Time series smoothing uses a dynamic window algorithm, which can automatically adjust the window size according to the different stages of the production line operation, thereby maintaining data stability without losing the dynamic response ability of the data. Spatial interpolation processing uses an algorithm based on Kriging interpolation to convert the data of discrete sensor points into a continuous concentration distribution map. Generates concentration spatial distribution information for the entire production line, accurately reflecting the hydrogen concentration in each area.

[0145] S5.3: construct a baseline concentration model, and generate a reference concentration distribution according to the hydrogen concentration data under the normal operating conditions of the hydrogen storage tank production line, wherein the reference concentration distribution is set as a concentration threshold 1 and a concentration threshold 2 according to the operating parameters of the hydrogen storage tank production line;

[0146] Specifically, a dynamic baseline concentration model was constructed based on the historical concentration data of the hydrogen storage tank production line under normal operating conditions. Unlike the traditional fixed threshold method, this step uses a multidimensional parameter analysis method to dynamically generate a baseline concentration distribution based on the operating parameters of the hydrogen storage tank (such as temperature, pressure, and flow). Concentration threshold one is used to define the concentration fluctuation within the normal range, and concentration threshold two is used as the starting point for the dangerous concentration of leakage risk. Such a dynamic baseline model can be updated in real time as the operating conditions of the production line change, adapt to the needs of different working conditions, and solve the defects of false alarms or missed alarms caused by fixed thresholds in traditional methods.

[0147] S5.4: Compare the hydrogen concentration data with the baseline concentration distribution, and calculate a weighted value based on historical hydrogen leakage conditions;

[0148] Specifically, the calculation of the weighting factor takes into account the following factors:

[0149] Spatial location: The weight of the sensor location. The weight value is adjusted according to the importance of the area (such as key interfaces and pipeline connections).

[0150] Historical leakage frequency: By analyzing historical data, the frequency information of leakage incidents in each region is extracted, and historical data in high-risk areas are given higher weight.

[0151] S5.5: Calculate the hydrogen leakage risk rate according to the comparison result and the weighted value, wherein the calculation formula of the hydrogen leakage risk rate is:

[0152]

[0153] Where L represents the hydrogen leakage risk rate, j represents a single historical hydrogen leakage, n represents the total number of historical hydrogen leakages, η represents the hydrogen leakage safety factor at the collection node, and L j represents the hydrogen leakage risk rate of the jth historical hydrogen leakage, G k represents the mean hydrogen concentration of the acquisition node within the acquisition time range, M1 represents the concentration threshold 1, M2 represents the concentration threshold 2, max(·) represents the maximum value function, ΔG represents the rate of change of the hydrogen concentration of the acquisition node within the acquisition time range, G t Indicates the concentration change threshold.

[0154] Specifically, the calculation of the risk rate not only considers a single concentration anomaly, but also conducts a comprehensive assessment based on the distribution range and cumulative effect of the abnormal concentration, monitors the concentration change rate in real time, and combines the nonlinear growth model of the rate change to accurately capture possible high-risk leakage trends. For example, when an abnormal concentration point appears at a welding interface, not only will the degree of concentration exceeding the standard at that point be analyzed, but the comprehensive risk rate will also be calculated based on factors such as the historical leakage frequency and dynamic rate of the area. Assuming that the concentration rises rapidly in a short period of time, the risk weight of the point will be dynamically increased and it will be marked as a high-risk area. In addition, by generating step-by-step optimization suggestions, such as closing relevant valves or starting the emergency exhaust system, leakage can be controlled in a timely manner.

[0155] The specific steps of the optimization strategy are as follows:

[0156] S6.1: Use the correlation matrix to analyze the relationship between devices and calculate the mutual impact of multiple devices working together;

[0157] S6.2: Provide production process optimization suggestions based on the results of multi-equipment correlation analysis. Specifically, it includes optimizing the temperature of welding equipment to improve welding quality, adjusting the flow of cleaning equipment to improve cleaning effect, and optimizing the output pressure of equipment to improve the overall working efficiency and safety of the equipment;

[0158] S6.3: Use a fluid mechanics diffusion model to simulate the diffusion process of hydrogen leakage and predict the diffusion path of hydrogen in the production line based on the location of the leakage source, the concentration change rate and the equipment layout;

[0159] S6.4: Map the diffusion analysis results into a three-dimensional visualization model and use color gradients to display the leakage concentration levels in different areas.

[0160] Embodiment 2:

[0161] See also Figure 4 , the present invention provides an embodiment: a 3D digital factory equipment management system, the system comprising a three-dimensional modeling module, a data monitoring module, a data analysis module, an anomaly detection module and an operation optimization module;

[0162] The three-dimensional modeling module is configured with a production line modeling strategy, and the production line modeling strategy is used to generate a three-dimensional visualization model of the hydrogen storage tank production line;

[0163] The data monitoring module is used to collect and monitor various equipment data of the hydrogen storage tank production line in real time;

[0164] The data analysis module is configured with a data analysis strategy, which is used to perform dynamic feature extraction and association analysis on the collected data;

[0165] The abnormality detection module is used to detect abnormal conditions of the equipment and generate an early warning signal;

[0166] The operation optimization module provides optimization suggestions for operation parameters based on the equipment operation status analysis results.

[0167] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. 3D digital factory equipment management method, characterized in that: The method comprises: Obtain the current structural information of each device in the hydrogen storage tank production line; Processing the current structural information, digitally modeling the hydrogen storage tank production line through 3D modeling software, and constructing a visual model of the production line; Collect the operating data of each device through multiple sensors, and link the three-dimensional model of the production line to monitor the working status of the equipment; Analyze the equipment operation status in the production process based on the visually displayed data; Assess the risk of hydrogen leakage during operation based on the visually displayed data; Based on the analysis and evaluation results, measures to optimize the production line are formulated through optimization strategies.

2. The 3D digital factory equipment management method according to claim 1 is characterized in that: The obtaining of the current structural information of each device in the hydrogen storage tank production line includes: The hydrogen storage tank production line is scanned in sections by a laser scanning device to obtain a laser point cloud, wherein the scanning angle of each area is adjusted according to the complexity and occlusion characteristics of the equipment; Determine the geometric characteristics of the equipment according to the laser point cloud, plan the acquisition path of the image sensor according to the geometric characteristics, and perform image acquisition on the hydrogen storage tank production line according to the acquisition path, wherein the acquisition path includes: For welding equipment, collect texture images of the welding head and welding area surface; For cleaning equipment, collect gloss images of nozzles, liquid flow paths, and housing surfaces; For heat treatment equipment, collect structural images inside the furnace; For servo motors, capture trajectory images of bearings and couplings.

3. The 3D digital factory equipment management method according to claim 2 is characterized in that: The construction of a visual model of the production line includes: Preprocessing the laser point cloud, determining the spatial position and geometric characteristics of each device in the production line according to the preprocessed point cloud data, and generating a preliminary three-dimensional model; Extract image features of texture images, gloss images, structure images and trajectory images; Matching feature points of the image features with the preliminary three-dimensional model, optimizing the preliminary three-dimensional model according to the matching results, and constructing a visualization model, wherein the optimization includes: Optimize welding detail features of welding joints and welding areas based on texture image features; Optimizing the cleaning efficiency characteristics of the nozzle, liquid flow path and housing surface based on the gloss image characteristics; Optimize the heat distribution characteristics inside the furnace based on structural image features; Optimize the stability characteristics of servo motor bearings and couplings based on trajectory image features.

4. The 3D digital factory equipment management method according to claim 3 is characterized in that: The step of extracting image features of a texture image, a gloss image, a structure image and a trajectory image comprises: Preprocessing the texture image, extracting the texture features of the welding surface through the local binary pattern, extracting the edge direction features of the welding area through the gradient direction histogram, combining the frequency characteristics of the texture features and the edge direction features through Fourier transform, and calculating the texture image features; Preprocessing the gloss image, calculating the surface reflectivity of the gloss image according to a light reflection model, extracting the reflection characteristics of the nozzle and the liquid flow path, calculating the highlight distribution characteristics by analyzing the light spot distribution of the gloss image, and calculating the cleaning efficiency characteristics according to the reflection characteristics and the highlight distribution characteristics; Preprocessing the structural image, extracting corner features of the structural image, processing the corner features according to the SURF algorithm, and calculating feature matching points of the furnace inner wall and the heating element; The trajectory image is preprocessed, and the trajectory path of the servo motor bearing and the coupling during the movement is extracted by an optical flow algorithm, and the radial runout of the bearing and the synchronous offset of the coupling are calculated according to the trajectory path.

5. The 3D digital factory equipment management method according to claim 1 is characterized in that: The specific steps for analyzing the equipment operation status in the production process are as follows: The operating data of each device in the hydrogen storage tank production line is collected through multiple sensors, including temperature, current, voltage and vibration data, and dynamic feature extraction is performed on the collected data; By establishing an adaptive anomaly detection model, the extracted features are analyzed in real time to determine whether the working status of the equipment is normal; The long short-term memory network prediction model is used to predict the trend of key parameters of the equipment and evaluate the remaining life of the equipment.

6. The 3D digital factory equipment management method according to claim 5 is characterized in that: The specific steps of extracting dynamic features from the collected data are as follows: Collect temperature, current and voltage data from welding equipment, cleaning equipment, heat treatment equipment and servo motors, extract their change curves through time series analysis methods, construct temperature gradient diagrams, current fluctuation diagrams and voltage fluctuation diagrams of the equipment, and monitor the temperature, current and voltage stability of the equipment during the production process; The short-time Fourier transform and wavelet transform techniques are used to perform frequency domain analysis on the equipment vibration signal, extract the characteristic frequency, amplitude, energy distribution and acceleration indicators in the vibration signal, and construct the equipment vibration state spectrum.

7. The 3D digital factory equipment management method according to claim 1 is characterized in that: The risk assessment of hydrogen leakage during operation includes: Hydrogen concentration sensors are placed at the welding interfaces, pipeline connections and valve areas of hydrogen storage tank production lines to collect hydrogen concentration data in the equipment operating environment; Performing time series smoothing and spatial interpolation processing on the hydrogen concentration data to obtain standardized concentration data; Constructing a baseline concentration model, generating a reference concentration distribution according to the hydrogen concentration data under normal operating conditions of the hydrogen storage tank production line, wherein the reference concentration distribution is set as a concentration threshold 1 and a concentration threshold 2 according to the operating parameters of the hydrogen storage tank production line; Comparing the hydrogen concentration data with the reference concentration distribution, and calculating a weighted value based on historical hydrogen leakage conditions; The hydrogen leakage risk rate is calculated according to the comparison result and the weighted value, wherein the calculation formula of the hydrogen leakage risk rate is: Where L represents the hydrogen leakage risk rate, j represents a single historical hydrogen leakage, n represents the total number of historical hydrogen leakages, η represents the hydrogen leakage safety factor at the collection node, and L j represents the hydrogen leakage risk rate of the jth historical hydrogen leakage, G k represents the mean hydrogen concentration of the acquisition node within the acquisition time range, M1 represents the concentration threshold 1, M2 represents the concentration threshold 2, max(·) represents the maximum value function, ΔG represents the rate of change of the hydrogen concentration of the acquisition node within the acquisition time range, G t Indicates the concentration change threshold.

8. The 3D digital factory equipment management method according to claim 7 is characterized in that: The specific steps of the optimization strategy are as follows: Use the correlation matrix to analyze the relationship between devices and calculate the mutual impact of multiple devices working together; Based on the results of multi-equipment correlation analysis, provide production process optimization suggestions, including optimizing welding equipment temperature to improve welding quality, adjusting cleaning equipment flow to improve cleaning effect, and optimizing servo motor output pressure to improve the overall working efficiency and safety of the equipment; The diffusion process of hydrogen leakage is simulated by a fluid mechanics diffusion model, and the diffusion path of hydrogen in the production line is predicted based on the location of the leakage source, the concentration change rate and the equipment layout; The diffusion analysis results are mapped into a 3D visualization model, using color gradients to show the leakage concentration levels in different areas.

9. A 3D digital factory equipment management system, used to implement the 3D digital factory equipment management method according to any one of claims 1 to 8, characterized in that: The system includes a three-dimensional modeling module, a data monitoring module, a data analysis module, an anomaly detection module and an operation optimization module; The three-dimensional modeling module is configured with a production line modeling strategy, and the production line modeling strategy is used to generate a three-dimensional visualization model of the hydrogen storage tank production line; The data monitoring module is used to collect and monitor various equipment data of the hydrogen storage tank production line in real time; The data analysis module is configured with a data analysis strategy, which is used to perform dynamic feature extraction and association analysis on the collected data; The abnormality detection module is used to detect abnormal conditions of the equipment and generate an early warning signal; The operation optimization module provides optimization suggestions for operation parameters based on the equipment operation status analysis results.

Images