A three-dimensional Gaussian splash-based Internet of Things visual monitoring method, system and computer device

Abstract

The present application relates to the technical field of monitoring, in particular to a kind of Internet of Things visual monitoring method and system based on three-dimensional Gaussian splash.The present application reconstructs equipment model by three-dimensional Gaussian splash algorithm, realizes the spatial accurate presentation of industrial equipment, and this stereoscopic display helps to improve fault location and operation response efficiency, through the classification and integration of equipment physical properties, spatial position and multiple heterogeneous data sources, so that multidimensional data real-time monitoring can be carried out on a unified space carrier, according to the dynamic association of equipment model topological structure and time dimension data, the running state and change trend of equipment can be monitored in real time, through the depth combination of time dimension data and equipment model, not only improve the comprehensive monitoring ability of equipment running state, but also strengthen the accuracy and reaction speed of equipment fault diagnosis, promote the realization of intelligent and accurate management of industrial plant.

Description

Technical Field

[0001] This invention relates to the field of monitoring technology, and in particular to an IoT visualization monitoring method, system, and computer device based on three-dimensional Gaussian splashing. Background Technology

[0002] As the global manufacturing industry accelerates its transformation towards intelligence and digitalization, industrial plants, as the core unit of the manufacturing system, rely heavily on efficient equipment monitoring systems for operational safety, equipment stability, and production efficiency. Against this backdrop, emerging technologies such as the Internet of Things, digital twins, 3D modeling, and visualization are gradually being introduced into the field of industrial monitoring, driving the evolution of traditional systems towards intelligence and visualization.

[0003] Currently, mainstream industrial monitoring systems generally use two-dimensional planar charts to display equipment status. Although this method was practical in the early stages of industrial informatization, with the rapid increase in the number of devices in industrial scenarios, the complexity of equipment structures, and the multi-layered spatial layout of operating environments, traditional two-dimensional monitoring can no longer meet the requirements of modern smart factories for spatial perception and data fusion. Specifically, modern industrial plants generally adopt a three-dimensional layout, including multiple production lines, pipeline systems, logistics equipment, etc. Two-dimensional projection views cannot accurately reflect the actual position, proximity relationships, and operating logic of equipment in three-dimensional space, which seriously restricts the speed of fault location and maintenance response. At the same time, because industrial equipment in modern industrial plants generates a large amount of heterogeneous data (such as temperature, humidity, energy consumption, electrical status, etc.) during real-time operation, most existing systems currently use a single data source to present data separately, lacking the ability to integrate cross-dimensional and cross-system data. In actual operation, many equipment faults involve the linkage between multiple variables, but because existing systems lack a unified spatial carrier, it is difficult to analyze them simultaneously from multiple perspectives. Summary of the Invention

[0004] The main objective of this invention is to provide an IoT visualization monitoring method based on three-dimensional Gaussian splashing, aiming to solve the technical problems in the prior art.

[0005] This invention proposes an IoT visualization monitoring method based on three-dimensional Gaussian splashing, comprising:

[0006] The physical attribute information, spatial location coordinates, and heterogeneous data source set of each target industrial equipment in the industrial plant are obtained, and the corresponding heterogeneous data source set is classified according to each physical attribute information to obtain classified data sources, wherein the classified data sources include core data sources, auxiliary data sources, and edge data sources;

[0007] Based on the three-dimensional Gaussian splashing algorithm, the physical attribute information and spatial position coordinates of all target industrial equipment are used to reconstruct three-dimensional models to generate three-dimensional equipment models, and the model topology and surface point cloud data of the three-dimensional equipment models are obtained.

[0008] Assign a unique equipment identifier to the corresponding target industrial equipment based on each category data source, and associate and bind each unique equipment identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish an equipment-model spatial association binding table;

[0009] Based on the device-model space association binding table, obtain the real-time collected data and data collection timestamp of each category data source, and add time dimension labels to the corresponding real-time collected data according to each data collection timestamp to obtain time dimension data;

[0010] The three-dimensional device model is divided into nodes according to the model topology to obtain multiple device model nodes. Each device model node is dynamically associated with the corresponding time dimension data to obtain a data-model association mapping relationship.

[0011] Real-time monitoring of the rate of change of values ​​for each category data source, and determination of whether the rate of change of values ​​is greater than a preset change threshold;

[0012] If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback.

[0013] Preferably, the step of classifying the corresponding heterogeneous data source set according to each of the physical attribute information to obtain the classified data source includes:

[0014] Each of the physical attribute information is converted into a quantifiable parameter value, and a corresponding device feature vector is constructed based on multiple parameter values;

[0015] Obtain the total association weight score between each parameter value and all data sources in the heterogeneous data source set, and divide the device feature vector into core feature dimensions and auxiliary feature dimensions according to the total association weight score. The core feature dimensions include device function type and device rated parameters, and the auxiliary feature dimensions include device volume features and device installation height.

[0016] Extract data features from all data sources in each heterogeneous data source set, and obtain a function matching coefficient based on each data feature and device function type;

[0017] Parameter adaptation coefficients are obtained based on each of the data features and device rated parameters, and core correlation is obtained based on each of the parameter adaptation coefficients and function matching coefficients.

[0018] The core correlation degree is corrected based on the device volume characteristics and device installation height to obtain the corrected correlation degree, and the relationship between the corrected correlation degree and the classification threshold range is analyzed.

[0019] If the corrected correlation is greater than the upper limit of the classification threshold range, then the data source is determined to be a core data source.

[0020] If the corrected correlation degree is within the classification threshold range, then the data source is determined to be an auxiliary data source;

[0021] If the corrected correlation is less than the lower limit of the classification threshold range, the data source is determined to be an edge data source.

[0022] Preferably, the step of generating a 3D equipment model by reconstructing the physical attribute information and spatial location coordinates of all target industrial equipment based on the 3D Gaussian splashing algorithm includes:

[0023] The equipment density, length, width, and height are obtained based on the physical attribute information of the target industrial equipment, and the three-dimensional reconstruction bounding box is determined based on the spatial location coordinates, equipment length, width, and height.

[0024] Obtain the maximum size and preset sampling accuracy of the target industrial equipment, and obtain the number of grid division dimensions based on the maximum size and preset sampling accuracy;

[0025] The 3D reconstructed bounding box is divided into multiple 3D mesh units according to the number of mesh division dimensions, and the first center coordinates of each 3D mesh unit are obtained;

[0026] Obtain the maximum density, maximum length, maximum width, and maximum height of all target industrial equipment, and determine the density weight ratio based on the maximum density and equipment density;

[0027] The length weight percentage is obtained based on the maximum length and the device length, and the width weight percentage is obtained based on the maximum width and the device width.

[0028] The height weight ratio is obtained based on the maximum height and the device height. The total splash value of the corresponding 3D mesh cell is calculated based on the height weight ratio, density weight ratio, length weight ratio, width weight ratio, the first center coordinate of each 3D mesh cell, and the coordinates of all spatial points. The calculation formula is as follows:

[0029] ;

[0030] Where Z(PJ) represents the total splash value, M(DQ) represents the density weight percentage, C(DQ) represents the length weight percentage, K(DQ) represents the width weight percentage, G(DQ) represents the height weight percentage, x represents the x-coordinate of the center of the 3D mesh cell, y represents the y-coordinate of the center of the 3D mesh cell, and z represents the y-coordinate of the center of the 3D mesh cell. i Y represents the x-coordinate of the i-th spatial point. i Z represents the vertical coordinate of the i-th spatial point. i Let represent the ordinate of the i-th spatial point, where i represents the index of the spatial point and N represents the number of spatial points;

[0031] Determine whether the total splash value is less than a preset splash value;

[0032] If the total splash value is less than the preset splash value, the three-dimensional mesh cell corresponding to the total splash value is determined to be an invalid three-dimensional mesh cell;

[0033] If the total splash value is not less than the preset splash value, the three-dimensional mesh unit corresponding to the total splash value is determined to be a valid three-dimensional mesh unit, and all valid three-dimensional mesh units are reconstructed using the moving cube algorithm to generate a three-dimensional device model.

[0034] Preferably, the step of assigning a unique equipment identifier to the corresponding target industrial equipment based on each category data source, and associating and binding each unique equipment identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish an equipment-model spatial association binding table includes:

[0035] Based on the data source classification, obtain the device function type, device installation time, and device spatial coordinates, and obtain the device type code based on the device function type;

[0036] The installation time code is obtained based on the equipment installation time, and the spatial coordinate code is obtained based on the equipment spatial coordinates;

[0037] The device type code, installation time code, and spatial coordinate code are concatenated in chronological order to obtain a unique device identifier;

[0038] Obtain the second center coordinates of the 3D device model, and obtain the coordinate deviation value based on the second center coordinates and the spatial position coordinates;

[0039] The unique device identifier is bound to the spatial coordinates of the target industrial equipment based on the coordinate deviation value;

[0040] Based on the three-dimensional coordinates of each spatial point in the surface point cloud data, the spatial distance between each three-dimensional coordinate and the spatial position coordinate is obtained;

[0041] Based on the spatial distance, the associated spatial points are bound to unique device identifiers to obtain the device-model spatial association binding table.

[0042] Preferably, the step of dividing the 3D device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associating each device model node with corresponding time dimension data to obtain a data-model association mapping relationship, includes:

[0043] Extract all triangular facets in the model's topology and obtain the centroid coordinates of each triangular facet;

[0044] Obtain the equipment function type of the target industrial equipment corresponding to the model topology, and obtain the core functional area based on the equipment function type;

[0045] Triangular facets whose centroid coordinates are located within the core functional area are classified as core facets, and triangular facets whose centroid coordinates are not located within the core functional area are classified as non-core facets.

[0046] Obtain the shortest path distance between each non-core surface element and the core surface element, and determine whether the shortest path distance is greater than a preset distance;

[0047] If the shortest path distance is greater than the preset distance, then the non-core surface element is classified as an auxiliary surface element;

[0048] If the shortest path distance is not greater than the preset distance, then the non-core surface element is divided into edge surface elements, and the corresponding device model node is obtained based on the edge surface elements, auxiliary surface elements and core surface elements.

[0049] Obtain the data type for each time dimension, wherein the data type is one of core data, auxiliary data, and peripheral data;

[0050] The core data, auxiliary data, and edge data are dynamically associated one-to-one with the core surface elements, edge surface elements, and auxiliary surface elements in the device model node, respectively, to obtain the data-model association mapping relationship.

[0051] Preferably, the step of generating spatial visualization feedback based on the dynamic rendering and abnormal visual feedback includes:

[0052] Obtain the visual attribute data and time dimension data of the dynamic rendering, and render the visual attribute data on the corresponding device model node;

[0053] Based on the time dimension data, the visual attribute data rendered on the corresponding device model node will be dynamically updated in real time to obtain dynamic visual rendering data.

[0054] Obtain the abnormal node identifier, abnormal attributes, and abnormal visual parameters of the abnormal visual feedback;

[0055] The abnormal node position in the 3D device model is determined based on the abnormal node identifier, and the abnormal node position is rendered based on the abnormal visual parameters to obtain abnormal node rendering data.

[0056] Obtain the associated device nodes of the abnormal attribute, and connect the associated device nodes with the abnormal node location as the center to obtain the abnormal propagation path;

[0057] The abnormal transmission path, dynamic visual rendering data, and abnormal node rendering data are presented as a spatially visualized distribution feedback in the form of a heatmap.

[0058] This application also provides an IoT visualization monitoring system based on three-dimensional Gaussian splashing, including:

[0059] The first acquisition module is used to acquire the physical attribute information, spatial location coordinates and heterogeneous data source set of each target industrial equipment in the industrial plant, and classify the corresponding heterogeneous data source set according to each physical attribute information to obtain classified data source, wherein the classified data source includes core data source, auxiliary data source and edge data source;

[0060] The reconstruction module is used to reconstruct three-dimensional models of all target industrial equipment based on the physical attribute information and spatial position coordinates of the three-dimensional Gaussian splashing algorithm, and to obtain the model topology and surface point cloud data of the three-dimensional equipment model.

[0061] The binding module is used to assign a unique device identifier to the corresponding target industrial equipment based on each category data source, and to associate and bind each unique device identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish a device-model spatial association binding table.

[0062] The second acquisition module is used to acquire the real-time collected data and data collection timestamp of each category data source based on the device-model space association binding table, and add time dimension labels to the corresponding real-time collected data according to each data collection timestamp to obtain time dimension data;

[0063] The association module is used to divide the three-dimensional device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associate each device model node with the corresponding time dimension data to obtain a data-model association mapping relationship;

[0064] The judgment module is used to monitor the rate of change of the values ​​of each category data source in real time and determine whether the rate of change of the values ​​is greater than a preset change threshold.

[0065] If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback.

[0066] Preferably, the binding module includes:

[0067] The first acquisition unit is used to acquire the device function type, device installation time and device spatial coordinates according to the classification data source, and to acquire the device type code according to the device function type;

[0068] The second acquisition unit is used to acquire an installation time code based on the device installation time and to acquire a spatial coordinate code based on the device spatial coordinates.

[0069] The splicing unit is used to splice the device type code, installation time code, and spatial coordinate code in chronological order to obtain a unique device identifier;

[0070] The third acquisition unit is used to acquire the second center coordinates of the three-dimensional device model, and to acquire the coordinate deviation value based on the second center coordinates and the spatial position coordinates;

[0071] The first binding unit is used to bind a unique device identifier to the spatial coordinates of the target industrial equipment according to the coordinate deviation value;

[0072] The fourth acquisition unit is used to acquire the spatial distance between each three-dimensional coordinate and the spatial position coordinate based on the three-dimensional coordinate of each spatial point in the surface point cloud data.

[0073] The second binding unit is used to bind the associated spatial point to the unique device identifier according to the spatial distance, so as to obtain the device-model spatial association binding table.

[0074] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described IoT visualization monitoring method based on three-dimensional Gaussian splashing.

[0075] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described IoT visualization monitoring method based on three-dimensional Gaussian splashing.

[0076] The beneficial effects of this invention are as follows: This invention reconstructs equipment models using a three-dimensional Gaussian splashing algorithm, overcoming the limitations of traditional two-dimensional projection maps that cannot accurately reflect equipment positions and interrelationships. It achieves precise spatial presentation of industrial equipment, and this three-dimensional display helps improve fault location and maintenance response efficiency. By classifying and integrating equipment physical attributes, spatial locations, and multiple heterogeneous data sources, it enables real-time monitoring of multi-dimensional data on a unified spatial carrier, solving the shortcomings of existing systems in handling multi-variable linkages. Based on the dynamic correlation between the equipment model topology and time-dimensional data, it can monitor the equipment's operating status and trends in real time. When data changes exceed preset thresholds, it can quickly trigger abnormal feedback and dynamic rendering, providing an intuitive and immediate visual response to help quickly locate faults and take measures. Through the deep integration of time-dimensional data and equipment models, it not only improves the comprehensive monitoring capability of equipment operating status but also enhances the accuracy and response speed of equipment fault diagnosis, promoting the realization of intelligent and precise management of industrial plants. Attached Figure Description

[0077] Figure 1 This is a schematic diagram of a method flow according to an embodiment of the present invention.

[0078] Figure 2 This is a schematic diagram of the system structure according to an embodiment of the present invention.

[0079] Figure 3 This is a schematic diagram of the internal structure of a computer device according to an embodiment of this application.

[0080] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0081] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0082] like Figure 1 As shown, this application provides an IoT visualization monitoring method based on three-dimensional Gaussian splashing, including:

[0083] S1. Obtain the physical attribute information, spatial location coordinates, and heterogeneous data source set of each target industrial equipment in the industrial plant, and classify the corresponding heterogeneous data source set according to each physical attribute information to obtain classified data sources, wherein the classified data sources include core data sources, auxiliary data sources, and edge data sources;

[0084] S2. Based on the three-dimensional Gaussian splashing algorithm, the physical attribute information and spatial position coordinates of all target industrial equipment are used to reconstruct the three-dimensional equipment model, and the model topology and surface point cloud data of the three-dimensional equipment model are obtained.

[0085] S3. Assign a unique equipment identifier to the corresponding target industrial equipment based on each category data source, and associate and bind each unique equipment identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish an equipment-model spatial association binding table.

[0086] S4. Based on the device-model space association binding table, obtain the real-time collected data and data collection timestamp of each category data source. Through data middleware, perform format conversion and noise filtering on the real-time collected data to obtain standardized real-time data. Add time dimension labels to the corresponding standardized real-time data according to each data collection timestamp to obtain time dimension data.

[0087] S5. Divide the three-dimensional device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associate each device model node with the corresponding time dimension data to obtain a data-model association mapping relationship.

[0088] S6. Monitor the rate of change of values ​​for each category data source in real time, and determine whether the rate of change of values ​​is greater than a preset change threshold.

[0089] If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback.

[0090] As described in steps S1-S6 above, this invention acquires the physical attribute information, spatial coordinates, and heterogeneous data source set of each target industrial device within an industrial plant. This ensures a comprehensive understanding of the device, whereas existing technologies typically focus only on a single parameter or data source, neglecting the device's actual spatial distribution, physical attributes, and the correlation of multi-dimensional data. Comprehensive data collection accurately reflects the overall status of the device, including heterogeneous data such as temperature, humidity, and energy consumption. This provides a comprehensive foundation for subsequent efficient fault detection and predictive maintenance, compared to existing solutions that rely on a single data source or lack spatial information. This approach provides multi-dimensional monitoring and analysis for industrial equipment, significantly improving the accuracy and efficiency of fault location and maintenance response. The combination of spatial information and physical attributes allows for precise equipment location in complex, multi-layered environments, avoiding the shortcomings of existing two-dimensional displays that cannot accurately represent the true location of equipment. Furthermore, it categorizes the corresponding heterogeneous data sources based on each physical attribute, resulting in categorized data sources. These categorized data sources include core data sources, auxiliary data sources, and edge data sources. Classifying heterogeneous data sources into core, auxiliary, and edge categories allows for more efficient management and processing of large amounts of real-time data from different sources. Different categories of data have different impacts on the operating status of the equipment; core data sources may be related to the equipment's critical functions. The primary data sources are directly related to the equipment, while auxiliary and edge data sources may be related to environmental factors or secondary functions. Classification allows for refined management of the priority, real-time performance, and sensitivity of fault warnings for each data source. Compared to existing technologies that treat all data sources equally, this invention effectively improves data processing efficiency and accuracy through classification management. Classified data more accurately corresponds to the equipment's operating status, avoiding unnecessary data interference, optimizing information transmission and processing, reducing system load, and improving the timeliness of fault warnings. Based on a 3D Gaussian splashing algorithm, 3D models are generated by reconstructing the physical attribute information and spatial coordinates of all target industrial equipment. The Gaussian splashing algorithm reconstructs a 3D model of equipment based on its physical properties and spatial coordinates. It realistically reproduces the equipment's position, shape, and relative position to other equipment in space. Through efficient mathematical modeling and calculation methods, the 3D Gaussian splashing algorithm accurately represents the geometric features and physical state of equipment. It is particularly suitable for complex equipment layouts and multi-layered spatial relationships in industrial plants. Unlike traditional 2D or simplified 3D display methods, the 3D Gaussian splashing algorithm better simulates the spatial distribution and operational logic of equipment, providing a more intuitive and accurate spatial view. This is especially important for multi-layered industrial plants, helping engineers accurately locate equipment in space and perform more refined fault analysis and maintenance.This invention acquires the topology and surface point cloud data of a 3D equipment model. This not only provides geometric and spatial information for the 3D model but also enhances its detail and accuracy. Extracting the topology helps reveal the relationships and dependencies between devices, while the surface point cloud data reflects the actual physical characteristics and details of the equipment, providing more accurate information for subsequent data analysis and fault prediction. Many existing models may not accurately reproduce the spatial relationships and surface details between devices. This invention, through the combination of topology and point cloud data, provides a more refined model structure, enabling more precise equipment monitoring and fault prediction. In particular, the point cloud data provides rich information about the equipment surface, facilitating high-precision visual analysis and fault diagnosis.

[0091] By assigning a unique device identifier to each target industrial device from each categorized data source, and associating each unique device identifier with the spatial coordinates and surface point cloud data of the corresponding target industrial device, a device-model spatial association binding table is established. This table effectively tracks the location, status, and associated data of each device within the system, clearly recording the entire lifecycle status of the device. This facilitates accurate fault tracking and historical data analysis. Existing systems often lack spatial association records between devices and their models, making it difficult to quickly locate relevant data sources and their spatial associations when devices malfunction. The device-model spatial association binding table established by this invention addresses this issue. The model space association binding table helps users quickly locate equipment and its operating status, significantly improving troubleshooting efficiency and system response speed. Based on the equipment-model space association binding table, real-time collected data and data collection timestamps for each category data source are obtained. Time dimension labels are added to the corresponding real-time collected data according to each data collection timestamp, resulting in time-dimensional data. Adding time dimension labels to each real-time data point allows for data tracing and analysis according to time series, providing more accurate dynamic changes. This enables users to intuitively understand the real-time operating status and trends of the equipment, providing valuable time-dimensional information for predictive maintenance and fault detection. Compared with static data storage in traditional methods, this invention... Introducing a time dimension not only allows us to view the current status of equipment but also analyze its historical trends. This provides stronger time-based prediction capabilities for fault early warning and maintenance decisions, helping engineers make more accurate judgments and optimization strategies. By dividing the 3D equipment model into nodes through a model topology, multiple equipment model nodes are obtained. Each equipment model node is dynamically associated with corresponding time-dimensional data, establishing a data-model association mapping relationship. Node division through model topology clearly represents each functional module or sub-device in the equipment model, and the dynamic association between time-dimensional data and nodes allows for precise tracking of the state changes of each node. This is crucial for real-time monitoring of equipment status and early warning. Detecting equipment failures is crucial. Unlike existing technologies that monitor equipment as a whole, this invention uses node-level monitoring to acquire real-time status changes for each module or component. This refined management not only improves monitoring accuracy but also identifies potential problems before equipment failure occurs, thereby reducing downtime and maintenance costs. By monitoring the rate of change of values ​​from each category of data source in real time and determining whether the rate of change exceeds a preset threshold, if the rate of change exceeds the preset threshold, anomaly visual feedback and dynamic rendering are triggered for the corresponding equipment model node based on the data-model association mapping relationship. By monitoring the data change rate in real time and determining whether it exceeds the preset threshold, potential problems can be detected in time before equipment anomalies occur.Visual feedback and dynamic rendering provide operators with intuitive fault information, thereby shortening reaction and decision-making times. Existing technologies often rely on periodic inspections or manual input, which may fail to capture subtle changes in equipment in real time. This invention, through automated real-time monitoring and feedback mechanisms, can respond rapidly when equipment malfunctions, significantly improving the timeliness of fault handling and reducing equipment downtime and losses. Furthermore, it generates spatial visualization feedback based on dynamic rendering and visual feedback, allowing operators to more intuitively see the spatial distribution and impact range of equipment faults. This visualization effectively improves the efficiency of equipment management and maintenance, especially in complex industrial environments, enabling rapid fault location and action. Compared to traditional text or tabular data displays, spatial visualization feedback provides operators with intuitive changes in equipment status, making the spatial relationships of equipment, fault impacts, and operational logic clearer. This feedback not only helps technicians locate problems faster but also assists decision-makers in making scientific and reasonable maintenance and scheduling decisions.

[0092] In one embodiment, step S1, which involves classifying the corresponding heterogeneous data source set based on each physical attribute information to obtain the classified data source, includes:

[0093] S11. Convert each of the physical attribute information into a quantifiable parameter value, and construct a corresponding device feature vector based on the multiple parameter values;

[0094] S12. Obtain the total association weight score between each parameter value and all data sources in the heterogeneous data source set, and divide the device feature vector into core feature dimension and auxiliary feature dimension according to the total association weight score. The core feature dimension includes device function type and device rated parameters, and the auxiliary feature dimension includes device volume feature and device installation height.

[0095] S13. Extract the data features of all data sources in each heterogeneous data source set, and obtain the function matching coefficient based on each data feature and device function type;

[0096] S14. Obtain parameter adaptation coefficients based on each of the data features and device rated parameters, and obtain core correlation based on each of the parameter adaptation coefficients and function matching coefficients;

[0097] S15. The core correlation degree is corrected based on the equipment volume characteristics and equipment installation height to obtain the corrected correlation degree;

[0098] S16. Analyze the relationship between the modified correlation degree and the classification threshold range;

[0099] If the corrected correlation is greater than the upper limit of the classification threshold range, then the data source is determined to be a core data source.

[0100] If the corrected correlation degree is within the classification threshold range, then the data source is determined to be an auxiliary data source;

[0101] If the corrected correlation is less than the lower limit of the classification threshold range, the data source is determined to be an edge data source.

[0102] As described in steps S11-S16 above, the specific steps for correcting the core correlation degree based on the equipment volume characteristics and equipment installation height are as follows: First, determine whether the equipment volume characteristics are greater than a preset volume. If the equipment volume characteristics are greater than the preset volume, then add a first correction value to the core correlation degree (the first correction value is the weight of the influence of volume on the necessity of data source monitoring); otherwise, do not add it. Simultaneously, determine whether the equipment installation height is greater than a preset installation height. If the equipment installation height is greater than the preset installation height, then add a second correction value to the core correlation degree (the second correction value is the weight of the influence of installation height on the necessity of data source monitoring); otherwise, do not add it. This yields the corrected correlation degree. The total correlation weight score is the sum of the correlation weight scores of each parameter value and all data sources in the heterogeneous data source set. The correlation weight score refers to the ratio between each parameter value and the maximum value corresponding to that parameter value among all target industrial equipment.

[0103] This invention converts each physical attribute information into quantifiable parameter values ​​and constructs corresponding device feature vectors based on multiple parameter values. It obtains the total correlation weight score between each parameter value and all data sources in a heterogeneous data source set, and divides the device feature vector into core feature dimensions and auxiliary feature dimensions based on the total correlation weight score. The core feature dimensions include device function type and rated parameters, while the auxiliary feature dimensions include device volume characteristics and installation height. Converting the physical attributes of the target industrial equipment into quantifiable parameter values ​​helps standardize device feature information, allowing the characteristics of different devices to be processed in a unified manner. The feature vector is an abstract way of representing the multi-dimensional and comprehensive characteristics of the device. Constructing a feature vector from multiple parameter values ​​provides a comprehensive representation of the status of each device, reflecting its performance indicators and operational status more fully. By calculating the correlation between each parameter value and other data sources, data from different devices, systems, and sensors is effectively integrated. This not only enhances data diversity but also helps identify and assess the interrelationships between different data sources and their impact on device status. By distinguishing between core and auxiliary features, the system can prioritize processing key functions and parameters of the device, such as function type and rated parameters. This decomposition method helps focus on the most critical aspects of device performance during analysis, avoiding interference from irrelevant information. Through feature decomposition, the system can target... By selectively using different feature dimensions for different devices and data needs, the flexibility and adaptability of the analysis are improved. Especially under different environments or requirements, auxiliary feature dimensions can be adjusted and optimized according to specific circumstances. By extracting data features from all data sources in each heterogeneous data source set, data features are extracted from heterogeneous data sources, and data is aggregated from different dimensions. Existing technologies often process different data sources separately, lacking cross-system data integration capabilities. By extracting data features from heterogeneous data sources, this invention can comprehensively analyze device status within a unified framework, breaking the limitations of traditional methods. Through the integration of features from multiple data sources, device operation and maintenance analysis is no longer limited to a single data point. It can comprehensively evaluate equipment from multiple dimensions, thereby providing more comprehensive and accurate decision support. It obtains a function matching coefficient based on each data feature and equipment function type, calculates the matching coefficient between each data feature and equipment function type, and assesses the applicability of data features to equipment functions. The function type of equipment determines its role in production lines, pipelines, and other systems. By calculating the function matching coefficient, precise matching between data features and equipment functions can be achieved, ensuring that equipment behavior is closely linked to actual application scenarios. The calculation of the function matching coefficient can effectively eliminate data that is meaningless for specific equipment functions, thereby improving the quality and efficiency of data analysis. It also obtains a parameter adaptation coefficient based on each data feature and equipment rated parameters.By comparing the equipment's rated parameters (such as rated voltage and rated power) with data characteristics, a parameter matching coefficient is derived. The rated parameters of equipment often have strict requirements in actual use. Calculating the parameter matching coefficient ensures that the parameters used by the equipment during operation conform to its rated design standards, effectively avoiding equipment failures caused by parameter mismatch. This allows the system to monitor the equipment's operating status in real time, adjust operating parameters promptly, and ensure the equipment's reliability and stability. Furthermore, the core correlation degree is obtained based on each parameter matching coefficient and functional matching coefficient. Combining the functional matching coefficient and parameter matching coefficient yields the core correlation degree, which assesses the correlation between data characteristics and the equipment's core performance. The calculation of correlation degree comprehensively considers the functions and parameters of the equipment, providing a more comprehensive evaluation of equipment performance. Compared with the single-dimensional analysis in traditional technologies, core correlation degree can evaluate the equipment status and performance from multiple perspectives. By comprehensively considering the influence of multiple factors on the equipment, it can more accurately locate the root cause of equipment failure, reducing the limitations of traditional methods that can only analyze from a single perspective. By correcting the core correlation degree through equipment volume characteristics and installation height, a corrected correlation degree is obtained. The volume and installation height of the equipment directly affect the actual position and functional performance of the equipment in three-dimensional space. By correcting the correlation degree, a more accurate assessment can be achieved. To accurately reflect the impact of these spatial factors on equipment operation, the corrected correlation degree helps the system accurately locate the operating status of equipment in complex three-dimensional space, especially in industrial plants with three-dimensional layouts. This avoids the shortcomings of traditional two-dimensional views that cannot accurately reflect the location and proximity of equipment. The system analyzes the relationship between the corrected correlation degree and the classification threshold range. If the corrected correlation degree is greater than the upper limit of the classification threshold range, the data source is determined to be a core data source; if the corrected correlation degree is within the classification threshold range, the data source is determined to be an auxiliary data source; and if the corrected correlation degree is less than the lower limit of the classification threshold range, the data source is determined to be a marginal data source. The corrected correlation degree is compared with the set classification threshold range. This invention determines whether a data source is a core, auxiliary, or peripheral source, automatically classifying it based on relevance. This simplifies data processing and analysis, improving efficiency and reducing human error compared to manual judgment. By classifying data sources, the system intelligently selects the most relevant and valuable ones for further analysis, making equipment fault diagnosis and operational optimization more accurate and efficient. Compared to existing technologies, this invention significantly improves the accuracy of fault diagnosis, equipment performance optimization, and maintenance response speed by introducing cross-dimensional and cross-system data integration and analysis methods, effectively solving the challenges of fault location and maintenance response in industrial equipment.

[0104] In one embodiment, step S2, which involves reconstructing a 3D equipment model from the physical attribute information and spatial coordinates of all target industrial equipment using a 3D Gaussian splashing algorithm, includes:

[0105] S21. Obtain the equipment density, equipment length, equipment width, and equipment height based on the physical attribute information of the target industrial equipment, and determine the three-dimensional reconstruction bounding box based on the spatial location coordinates, equipment length, equipment width, and equipment height;

[0106] S22. Obtain the maximum size and preset sampling accuracy of the target industrial equipment, and obtain the number of grid division dimensions based on the maximum size and preset sampling accuracy;

[0107] S23. Divide the 3D reconstructed bounding box into multiple 3D mesh units according to the number of mesh division dimensions, and obtain the first center coordinates of each 3D mesh unit;

[0108] S24. Obtain the maximum density, maximum length, maximum width, and maximum height of all target industrial equipment, and obtain the density weight ratio based on the maximum density and equipment density;

[0109] S25. Obtain the length weight ratio based on the maximum length and the device length, and obtain the width weight ratio based on the maximum width and the device width;

[0110] S26. Obtain the height weight ratio based on the maximum height and equipment height, and calculate the total splash value of the corresponding three-dimensional mesh unit based on the height weight ratio, density weight ratio, length weight ratio, width weight ratio, the first center coordinate of each three-dimensional mesh unit, and the coordinates of all spatial points. The calculation formula is as follows:

[0111] ;

[0112] Where Z(PJ) represents the total splash value, M(DQ) represents the density weight percentage, C(DQ) represents the length weight percentage, K(DQ) represents the width weight percentage, G(DQ) represents the height weight percentage, x represents the x-coordinate of the center of the 3D mesh cell, y represents the y-coordinate of the center of the 3D mesh cell, and z represents the y-coordinate of the center of the 3D mesh cell. i Y represents the x-coordinate of the i-th spatial point. i Z represents the vertical coordinate of the i-th spatial point. i Let represent the ordinate of the i-th spatial point, where i represents the index of the spatial point and N represents the number of spatial points;

[0113] S27. Determine whether the total splash value is less than the preset splash value;

[0114] If the total splash value is less than the preset splash value, the three-dimensional mesh cell corresponding to the total splash value is determined to be an invalid three-dimensional mesh cell;

[0115] If the total splash value is not less than the preset splash value, the three-dimensional mesh unit corresponding to the total splash value is determined to be a valid three-dimensional mesh unit, and all valid three-dimensional mesh units are reconstructed using the moving cube algorithm to generate a three-dimensional device model.

[0116] As described in steps S21-S27 above, the calculation formula for the total splash value requires normalization of the calculation parameters to eliminate dimensional differences between different variables. This ensures all variables are on the same order of magnitude, making the calculation more stable and efficient. The maximum size of the equipment refers to the maximum value among the three dimensions of the equipment's length, width, and height, which are specific to a single target industrial device. This invention obtains the equipment density, length, width, and height from the physical attribute information of the target industrial device, and determines the 3D reconstruction bounding box based on the spatial coordinates, length, width, and height. By using the physical dimensions and density of the target industrial device, accurate spatial... Layout information enables subsequent 3D reconstruction to accurately reflect the spatial distribution of equipment. Existing technologies often fail to accurately obtain detailed information about the equipment, especially its actual density. By acquiring these physical attributes, precise positioning and reconstruction of equipment in 3D space can be achieved, thereby improving the efficiency of fault location and maintenance response. By combining the spatial coordinates and physical dimensions (such as length, width, and height) of the equipment to determine the 3D reconstruction bounding box, the position and shape of all equipment in space can be accurately represented, avoiding the spatial projection errors common in existing technologies. This ensures that the equipment model conforms to the actual layout from every angle in 3D space. Compared with traditional 2D projection methods, the introduction of 3D bounding boxes can better reflect the relationships between equipment. By analyzing the proximity relationships, space occupancy, and dynamic changes of the equipment, the realism of the equipment layout is improved, enabling more effective fault diagnosis and response during operation and maintenance. By obtaining the maximum size and preset sampling accuracy of the target industrial equipment, and then determining the mesh dimension based on these parameters, the 3D reconstructed bounding box is divided into multiple 3D mesh units. Using the maximum size and preset sampling accuracy to determine the mesh dimension allows for reasonable mesh unit division while ensuring computational accuracy, thus balancing model complexity and computational efficiency. Traditional meshing methods may ignore differences in equipment size, while this method based on the maximum equipment size ensures accuracy for equipment of different sizes. The partitioning avoids wasting computational resources. The introduction of preset sampling precision allows for dynamic adjustment of mesh precision based on actual needs, further optimizing computational efficiency and model accuracy. Dividing the 3D reconstruction bounding box into multiple mesh units is a common discretization method in 3D modeling. Each 3D mesh unit can be calculated and analyzed independently, enabling more detailed capture of the spatial characteristics of each unit in subsequent calculations. The first center coordinates of each 3D mesh unit are obtained. The maximum density, maximum length, maximum width, and maximum height of all target industrial equipment are obtained to calculate the relative importance of each device within the entire industrial scene.The use of maximum values ​​provides a basis for subsequent weight allocation, ensuring that devices of different sizes are reasonably weighted according to their importance and space occupation during modeling. By combining maximum values ​​with the physical properties of the devices, not only is the accuracy of calculation improved, but the spatial distribution of the model is also optimized, avoiding the impact of deviations in the size of a single device on the overall modeling. Density weight percentages are obtained based on maximum density and device density. The calculation of density weight percentages considers the "weight" distribution of devices in space, reflecting the degree of influence of devices on the entire system. By introducing density as a parameter, the physical behavior of devices can be effectively simulated, especially when dealing with device collisions, physical compression, or space contention; density weight calculations provide more detailed information in these cases. Unlike traditional methods that rely solely on device size to assess impact, density weighting offers more granularity, aiding in optimized spatial planning. Especially in complex environments, it provides higher operational reliability. Length weighting is derived from the maximum length and device length, width weighting from the maximum width and device width, and height weighting from the maximum height and device height. This length weighting calculation ensures appropriate placement of devices of different lengths in the 3D model. The length weighting is closely related to device function and space requirements, effectively adjusting the device's spatial proportion within the model and preventing uneven space allocation caused by excessively long or short devices. Existing technologies often neglect the impact of device length on spatial planning, thus... To further improve the rationality of equipment layout and ensure no unreasonable interference or wasted space between devices, width weighting is another key parameter. It ensures that the lateral distribution of equipment meets actual requirements. In most industrial equipment, the width directly affects its occupancy of surrounding space; excessively wide equipment may affect the operation of other equipment or space utilization. By introducing width weighting, the lateral layout of equipment can be effectively optimized, avoiding collisions or wasted space, enabling more accurate equipment space simulation, and reducing unnecessary interference between devices. The weighting of equipment height also plays a crucial role, especially in multi-layered spatial layouts, where height directly affects the vertical distribution of equipment. This is achieved by calculating the height based on the equipment height. Weighting can rationally plan the vertical space of equipment, thereby ensuring coordination between equipment layout and spatial height constraints. Based on the weighting percentages for height, density, length, and width, as well as the first center coordinates of each 3D mesh cell and the coordinates of all spatial points, the total splash value for each 3D mesh cell is calculated. This calculation comprehensively considers the spatial weights of the equipment and their distribution within the 3D mesh, ensuring that the equipment's space occupation matches its impact within the overall industrial environment. Compared to existing technologies, this splash value calculation allows for dynamic adjustment of the equipment's weight distribution in 3D space, resulting in a more realistic and accurate final equipment model and preventing certain equipment from occupying too much or too little space in the model.This improves the rationality and accuracy of modeling. By judging whether the total splash value is less than a preset splash value, if the total splash value is less than the preset splash value, the corresponding 3D mesh element is determined to be an invalid 3D mesh element; if the total splash value is not less than the preset splash value, the corresponding 3D mesh element is determined to be a valid 3D mesh element. By judging whether the total splash value is less than a preset threshold, invalid 3D mesh elements can be effectively filtered out. Invalid mesh elements may be caused by improper occupation of equipment in space or interference with other equipment. This filtering process can eliminate these influences, ensuring the accuracy and reliability of the model. In traditional techniques, it is often impossible to effectively remove these invalid meshes, leading to... Decreased model accuracy and low operational efficiency can be addressed through this judgment step, which significantly improves the quality of the final model. The Moving Cubes algorithm reconstructs all effective 3D mesh cells to generate a 3D equipment model. Using the Moving Cubes algorithm to reconstruct 3D mesh cells produces a highly refined 3D equipment model. Compared to traditional modeling methods, the Moving Cubes algorithm enables more flexible spatial construction, especially when dealing with complex equipment or interactions between adjacent devices, allowing for a more accurate reconstruction of the equipment's 3D shape. This invention, through continuous small-block adjustments and optimizations, ensures that the final 3D equipment model is not only accurate but also efficient, adaptable to the needs of different industrial environments, and greatly improves modeling quality and practical application effectiveness.

[0117] In one embodiment, step S3, which involves assigning a unique device identifier to the corresponding target industrial equipment based on each classification data source and associating each unique device identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish a device-model spatial association binding table, includes:

[0118] S31. Obtain the device function type, device installation time, and device spatial coordinates according to the classification data source, and obtain the device type code according to the device function type;

[0119] S32. Obtain the installation time code based on the equipment installation time, and obtain the spatial coordinate code based on the equipment spatial coordinates;

[0120] S33. The equipment type code, installation time code, and spatial coordinate code are concatenated in chronological order to obtain a unique equipment identifier;

[0121] S34. Obtain the second center coordinates of the three-dimensional device model, and obtain the coordinate deviation value based on the second center coordinates and the spatial position coordinates;

[0122] S35. Determine whether the coordinate deviation value is greater than the preset deviation value;

[0123] If the coordinate deviation value is greater than the preset deviation value, the three-dimensional device model is translated until the coordinate deviation value is no greater than the preset deviation value.

[0124] If the coordinate deviation value is not greater than the preset deviation value, then the unique device identifier is bound to the spatial coordinates of the target industrial equipment;

[0125] S36. Based on the three-dimensional coordinates of each spatial point in the surface point cloud data, obtain the spatial distance between each three-dimensional coordinate and the spatial position coordinate;

[0126] S37. Determine whether the spatial distance is greater than a preset distance;

[0127] If the spatial distance is greater than the preset distance, the spatial point is determined to be a non-associated spatial point;

[0128] If the spatial distance is not greater than the preset distance, the spatial point is determined to be an associated spatial point, and the associated spatial point is bound to a unique device identifier to obtain a device-model spatial association binding table.

[0129] As described in steps S31-S37 above, the equipment type code is determined based on the equipment function type corresponding to the classification data source (the classification data source includes temperature, humidity, energy consumption, etc., whose type is directly related to the equipment function, such as energy consumption data of motor equipment and vibration data of pump equipment). The installation time code is based on the actual installation date of the equipment, used to distinguish different batches of equipment of the same type and in the same location (to avoid duplicate identifiers due to equipment replacement). The spatial coordinate code is based on the spatial coordinates of the equipment, used to associate the physical location of the equipment with the location of the 3D model. After generating a unique equipment identifier, it needs to be bound to the core information of the equipment to ensure that the identifier is valid. The correspondence between devices and data is as follows: bound to spatial location coordinates, bound to surface point cloud data, and bound to classification data sources. Binding to spatial location coordinates means storing the identifier and the physical location coordinates of the device in an associated database. The device installation location can be directly queried through the identifier. Binding to surface point cloud data means adding identifier tags to the surface point cloud data of the 3D model (by calculating the distance between the point cloud points and the physical center point of the device, filtering the point cloud points belonging to the device and associating them with IDs). Binding to classification data sources means mapping the identifier to the heterogeneous data source associated with the device (such as temperature sensor ID, energy consumption sensor ID), ensuring that the data can be matched to the corresponding device through the identifier after collection. For example, taking "pump equipment installed on May 10, 2023 at coordinates (horizontal coordinate 12.34m, vertical coordinate 8.76m, vertical coordinate 3.00m)" as an example, the equipment type code is "PU" (pump equipment, using the industry-standard equipment type abbreviation), the installation time code is "20230510" (corresponding to the installation date of May 10, 2023), and the spatial coordinate code is "123087" (corresponding to spatial coordinates of horizontal coordinate 12.34m, vertical coordinate 8.76m, vertical coordinate 3.00m). Therefore, the final unique equipment identifier is PU20230510123087.

[0130] The translation transformation of the 3D device model involves translating the x-coordinate, y-coordinate, and s-coordinate of the second center coordinate of the 3D device model (specifically, the translation is based on the actual situation; if the deviation of the x-coordinate in the coordinate deviation value is greater than the preset deviation value, then the x-coordinate of the second center coordinate of the 3D device model is translated). The translation transformation formula is: P(YJ)=Z(BZ)+α*[K(BZ)-Z(BZ)]; where P(YJ) represents the translation distance, Z(BZ) represents the second center coordinate value, K(BZ) represents the spatial position coordinate value (corresponding to the second center coordinate value, i.e., the second center coordinate value refers to the x-coordinate of the 3D device model, then the spatial position coordinate value refers to the x-coordinate of the spatial position coordinate), and α represents the iteration coefficient. The formula for calculating the spatial distance between each 3D coordinate and the spatial position coordinate is: Wherein, K(JL) represents spatial distance, S(HZ) represents three-dimensional horizontal coordinate, K(HZ) represents spatial position horizontal coordinate, S(SZ) represents three-dimensional vertical coordinate, K(SZ) represents spatial position vertical coordinate, S(ZZ) represents three-dimensional vertical coordinate, and K(ZZ) represents spatial position vertical coordinate.

[0131] This invention obtains equipment function type, installation time, and spatial coordinates from categorized data sources. By acquiring these information, it accurately provides independent identification information for each device from multiple dimensions. Based on the effective integration of multi-source data, this invention overcomes the limitations of single data sources commonly found in existing technologies. The device's function type identifies its specific purpose, the installation time helps track changes in the device's lifecycle, and the spatial coordinates clearly define the device's actual location within the factory, ensuring the comprehensiveness and accuracy of the data. Compared to existing systems that typically rely on a single data source such as device model or serial number, using a combination of diverse data sources effectively avoids issues such as duplication, errors, or lack of real-time data. This approach improves the accuracy of equipment traceability by obtaining equipment type codes based on equipment function type, installation time codes based on equipment installation time, and spatial coordinate codes based on equipment spatial coordinates. The equipment function type coding further refines equipment classification and identification. Function type coding clarifies the role and function of equipment, facilitating subsequent analysis of relationships between equipment. The introduction of installation time coding accurately records the installation history of equipment. Spatial coordinate coding not only provides the specific location of equipment in three-dimensional space but also lays the foundation for spatial relationship and proximity analysis between equipment. Traditional two-dimensional planar diagrams cannot reflect the three-dimensional layout of equipment; spatial coordinate coding effectively solves this problem, ensuring the accuracy of equipment traceability. Spatial relationships are precisely recorded and analyzed, providing spatial perspective support for fault diagnosis and maintenance. By concatenating equipment type code, installation time code, and spatial coordinate code in chronological order, a unique equipment identifier is obtained. This unique identifier integrates all key equipment information, forming a unique identity for the equipment. This avoids the potential for duplicate or missing equipment identifiers in traditional technologies. By concatenating codes from different dimensions, this identifier not only includes the equipment's function, installation time, and spatial location, but can also be extended to other equipment attributes, ensuring the uniqueness and integrity of the equipment. This allows maintenance personnel to quickly identify and locate equipment information, significantly improving efficiency. This method improves equipment management efficiency by obtaining the second center coordinates of the 3D equipment model and calculating the coordinate deviation value based on these coordinates and the spatial position coordinates. By comparing the center coordinates of the 3D equipment model with the spatial position coordinates, precise equipment position calibration is performed. Compared to traditional equipment position calibration methods, this method, based on comparing the second center coordinates of the 3D equipment model with the spatial position coordinates, ensures accurate equipment positioning in complex 3D space. The spatial deviation value directly reflects the difference between the actual and theoretical positions of the equipment, allowing for necessary adjustments when the deviation is large. This not only improves the accuracy of equipment positioning but also enhances the rationality of equipment layout. The system checks whether the coordinate deviation value exceeds a preset deviation value; if the coordinate deviation value exceeds the preset deviation value...The 3D equipment model is then translated until the coordinate deviation is no greater than a preset deviation value. If the coordinate deviation is no greater than the preset deviation value, the unique equipment identifier is bound to the spatial coordinates of the target industrial equipment. By judging and controlling the coordinate deviation value, the accuracy of the equipment's 3D spatial positioning can be effectively ensured. The introduction of the preset deviation value provides a tolerance range for equipment position adjustment, thereby avoiding the impact of minor equipment deviations on overall management. The translation and transformation operation is more intelligent than traditional manual correction and can automatically adjust the equipment's position, ensuring that the equipment always remains within the predetermined spatial position, avoiding the impact of deviations on equipment operation or fault diagnosis. Binding the unique equipment identifier to the equipment's spatial coordinates is a key step in information integration. After binding, the equipment can not only be identified by the unique identifier but also directly linked to its spatial position and related operation and maintenance data. Compared with the separate processing methods in existing technologies, this has significant advantages, enabling a close integration of the equipment's spatial position and its functional status, providing real-time data support for future equipment monitoring, analysis, and fault diagnosis. By using the 3D coordinates of each spatial point in the surface point cloud data and obtaining the spatial distance between each 3D coordinate and the spatial position coordinates, the accuracy of the 3D spatial positioning of the equipment can be judged. If the spatial distance is greater than a preset distance, the spatial point is determined to be an unrelated spatial point; if the spatial distance is not greater than the preset distance, the spatial point is determined to be an associated spatial point. The associated spatial point is then bound to a unique device identifier, resulting in a device-model spatial association binding table. The spatial distance from point cloud data determines the association between the device and the target spatial point. The introduction of point cloud data makes the connection between the device and the space more concrete and intuitive, providing more dimensions for analysis of the spatial relationships, proximity, and operational logic of the device. Judging the spatial distance helps to filter out the true associated points between the device and the environment and effectively eliminate irrelevant points, thereby improving the matching accuracy between the device and the space. Traditional technologies often rely on a single data source, making accurate matching in complex 3D spaces difficult. The combination of point cloud data greatly improves the comprehensiveness and accuracy of the analysis. By creating a device-space association binding table, the relationship between the device and its space can be systematically organized. Compared to the scattered processing of devices and spaces in existing technologies, the establishment of the device-model spatial association binding table allows all device-space relationships to be managed on a unified platform, greatly improving information integration and system efficiency.

[0132] In one embodiment, step S5, which involves dividing the 3D device model into nodes based on the model topology to obtain multiple device model nodes, and dynamically associating each device model node with corresponding time dimension data to obtain a data-model association mapping relationship, includes:

[0133] S51. Extract all triangular facets in the model topology and obtain the centroid coordinates of each triangular facet.

[0134] S52. Obtain the equipment function type of the target industrial equipment corresponding to the model topology, and obtain the core functional area according to the equipment function type;

[0135] S53. The triangular facets whose centroid coordinates are located within the core functional area are classified as core facets, and the triangular facets whose centroid coordinates are not located within the core functional area are classified as non-core facets.

[0136] S54. Obtain the shortest path distance between each non-core surface element and the core surface element, and determine whether the shortest path distance is greater than a preset distance;

[0137] If the shortest path distance is greater than the preset distance, then the non-core surface element is classified as an auxiliary surface element;

[0138] If the shortest path distance is not greater than the preset distance, then the non-core surface element is divided into edge surface elements, and the corresponding device model node is obtained based on the edge surface elements, auxiliary surface elements and core surface elements.

[0139] S55. Obtain the data type of each time dimension, wherein the data type is one of core data, auxiliary data, and peripheral data;

[0140] S56. Dynamically associate the core data, auxiliary data, and edge data with the core surface elements, edge surface elements, and auxiliary surface elements in the device model node to obtain the data-model association mapping relationship.

[0141] As described in steps S51-S56 above, the equipment model nodes include edge functional nodes, auxiliary functional nodes, and core functional nodes. Edge facets, auxiliary facets, and core facets correspond to the edge functional nodes, auxiliary functional nodes, and core functional nodes in the equipment model nodes, respectively. The step of obtaining the core functional area based on the equipment function type involves locating the model area corresponding to the core component in the 3D equipment model primarily through geometric features, positional features, and motion-related features. For geometric features, the core component typically has a specific shape (e.g., a cylindrical motor rotor or radial blades for an impeller). The shape, size, and connection relationships of the facets in the model's topology are used to identify the core functional area. For location features, the core components are mostly located at the physical center of the equipment or on the power / energy input / output path (e.g., the motor rotor is located at the center of the stator, and the pump impeller is located at the midpoint of the line connecting the inlet and outlet). For motion-related features, for moving parts (e.g., rotor, impeller), there is a dynamic gap between the model area and adjacent parts (e.g., bearing, housing), which is identified by the difference in the spacing between surface elements and the direction of the normal vector. The shortest path distance is obtained by traversing the adjacency relationship of surface elements. Through spatial coordinate calibration and time axis alignment, the dual matching of nodes and data in spatial position and time axis is ensured, including spatial coordinate dynamic calibration and data attribution determination, and time axis alignment and data interpolation completion.

[0142] This invention extracts all triangular facets from the model's topology and obtains the centroid coordinates of each facet. By precisely extracting each triangular facet and calculating its centroid coordinates, it ensures a detailed description of the equipment model's geometry and spatial distribution. Unlike existing technologies, many systems typically rely on simple meshes or nodes, failing to accurately define each triangular facet, which may lead to error accumulation or information loss. Obtaining the centroid coordinates of each facet provides a more accurate geometric basis for subsequent segmentation and analysis, thereby improving the precise matching between the model and the equipment's functional regions. Furthermore, by obtaining the equipment function type corresponding to the target industrial equipment within the model's topology and acquiring the core [data / data] based on the equipment function type, this invention further enhances the accuracy of the model's matching with the equipment's functional regions. Determining the core functional area based on the device's functional type is a crucial step in ensuring accurate mapping of critical areas within the system. In existing technologies, many devices define their functional areas solely through hardware parameters, lacking a deep understanding and dynamic partitioning of device functions. By dynamically identifying the core functional area based on function type, the device's functions can be precisely combined with its geometric model, making the functional division of each part of the device more aligned with actual application needs. This provides higher operational accuracy and decision support for subsequent operations. Triangular elements whose centroid coordinates lie within the core functional area are designated as core elements, while those whose centroid coordinates do not lie within the core functional area are designated as non-core elements. This process is further refined based on the centroid coordinates... The relationship between core and non-core areas is used to classify facets into core facets and non-core facets, effectively distinguishing the key and peripheral areas of the equipment model. Traditional methods often treat the entire equipment model uniformly, lacking differentiated processing for different functional areas. The division method of this invention clearly defines the distinction between core and non-core areas, providing more scientific spatial structure support for subsequent data mapping, thereby improving the efficiency and accuracy of data analysis and fault diagnosis. By obtaining the shortest path distance between each non-core facet and a core facet, and determining whether the shortest path distance is greater than a preset distance, if the shortest path distance is greater than the preset distance, the non-core facet is classified as an auxiliary facet; if the shortest path distance is not greater than the preset distance, it is classified as an auxiliary facet. The non-core facets are divided into edge facets, and corresponding device model nodes are obtained based on the edge facets, auxiliary facets, and core facets. By calculating the shortest path distance between non-core facets and core facets, the edge regions in the device model can be better identified, and the distance relationship provides a basis for subsequent facet division. Unlike existing technologies that rely solely on geometric centers or certain simplification methods, calculating the shortest path distance can more accurately reflect the actual spatial relationship between facets. Especially for complex device structures, this invention can avoid the inaccuracy of spatial layout, thereby improving the model's accurate description of the actual operation of the device. By setting a preset distance, auxiliary facets and edge facets can be flexibly divided according to the spatial relationship between facets in the device model.Compared to traditional static partitioning, this method offers greater flexibility and adaptability. Dynamic partitioning not only adapts flexibly to different equipment structures but also adjusts the classification of facets in real-time based on the working status of different areas during equipment operation. This provides higher accuracy for subsequent data analysis, especially when equipment operating conditions change significantly, reflecting the actual working environment in real time. By combining different facet types of the equipment model, corresponding equipment model nodes can be dynamically generated, avoiding the shortcomings of traditional methods that rely solely on fixed node structures. This improves the correlation between equipment model nodes and equipment functions. By dynamically acquiring equipment model nodes, the system can update equipment model node information in real-time based on different equipment operating states, providing flexible data support for subsequent data mapping, fault detection, and performance optimization. The data type for each time dimension is obtained, which can be one of core data, auxiliary data, or edge data. The core data, auxiliary data, and edge data are then compared with the core facets, edge facets, and auxiliary facets in the equipment model nodes, respectively. This invention employs a dynamic, one-to-one mapping between elements to establish a data-model association. By associating core, auxiliary, and peripheral data with corresponding element types, it enables efficient data integration and classification management. In existing technologies, data is often treated as isolated and independent, lacking real-time correlation with equipment models. This makes it difficult to fully utilize spatial information during analysis and decision-making. However, this invention, through dynamic association of data and model nodes, comprehensively improves data utilization, ensuring greater accuracy and real-time performance in equipment fault diagnosis and performance optimization. It provides a novel solution, particularly for cross-system integration of heterogeneous data. By dynamically associating data and models, a precise mapping relationship is established. Unlike the static mapping methods commonly used in existing technologies, dynamic data-model association can adapt to changes during equipment operation in real time, allowing the system to continuously optimize data processing. It effectively matches heterogeneous data with the actual operating conditions of the equipment model, providing cross-dimensional and cross-system data integration capabilities and offering more efficient solutions for intelligent equipment monitoring, fault prediction, and performance evaluation.

[0143] In one embodiment, step S6, which generates spatial visualization feedback based on the dynamic rendering and abnormal visual feedback, includes:

[0144] S61. Obtain the visual attribute data and time dimension data of the dynamic rendering, and render the visual attribute data on the corresponding device model node;

[0145] S62. Based on the time dimension data, the visual attribute data rendered on the corresponding device model node is dynamically updated in real time to obtain dynamic visual rendering data.

[0146] S63. Obtain the abnormal node identifier, abnormal attributes, and abnormal visual parameters of the abnormal visual feedback;

[0147] S64. Determine the location of the abnormal node in the 3D device model based on the abnormal node identifier, and render the abnormal node location based on the abnormal visual parameters to obtain abnormal node rendering data.

[0148] S65. Obtain the associated device node of the abnormal attribute, and connect the associated device nodes with the abnormal node position as the center to obtain the abnormal propagation path;

[0149] S66. Present the abnormal transmission path, dynamic visual rendering data, and abnormal node rendering data in the form of a heatmap to visualize the spatial distribution feedback.

[0150] As described in steps S61-S66 above, before real-time monitoring of the numerical change rate of each category data source, it is necessary to establish a data-visual attribute mapping rule based on the numerical range of standardized real-time data and the device operating threshold (wherein, visual attributes include model node color, transparency, vibration animation frequency, and luminous intensity). Based on a 3D Gaussian splashing rendering engine, the standardized real-time data is dynamically rendered on the corresponding device model node through visual attributes. By real-time monitoring of the numerical change rate of the standardized real-time data and the degree of deviation from the operating threshold, if the numerical change rate exceeds a preset change threshold or the deviation reaches a warning threshold, abnormal visual feedback is triggered (abnormal visual feedback includes the model node flashing frequency increasing to a preset multiple of the base frequency, the color switching to a preset warning color, and the transparency decreasing to a preset multiple of the base transparency). In the spatial region corresponding to the surface point cloud data of the device model node, colors are rendered according to the HSV color formula, so that the colors at different locations in space intuitively reflect the data values. (e.g., high-temperature areas are red, low-temperature areas are green). For transparency and luminous intensity, transparency can be calculated based on data reliability. In space, the transparency of nodes reflects data reliability. Combined with luminous intensity, light radiates outward from the node in three-dimensional space, with intensity increasing as luminous intensity increases, highlighting abnormal risk areas. For the spatial expression of vibration animation, vibration effects can be achieved by shifting vertices along the normal vector direction at the three-dimensional spatial position of the device model node based on vibration frequency. This allows the dynamic movement of nodes in space to intuitively reflect the speed of data change. In time-stamp order, visual attributes (e.g., color changes in real time with temperature) are updated at the corresponding spatial position for each frame, allowing the visual state of nodes in space to dynamically flow over time, intuitively presenting the temporal changes of data. At the spatial position of abnormal nodes, abnormal visual parameters are rendered, including increasing the flicker frequency to a preset value and forming a visual focus in space through periodic changes in brightness; transparency is reduced to a preset baseline value, making abnormal nodes easier to distinguish in space.

[0151] This invention acquires dynamically rendered visual attribute data and temporal dimension data, and renders the visual attribute data on the corresponding device model nodes. The purpose of acquiring dynamically rendered visual attribute data and temporal dimension data is to obtain real-time, dynamic device operating status information. The dynamically rendered data can accurately reflect the physical state of the device at the current moment (such as temperature, humidity, pressure, current, etc.), and these attributes change in real time. The temporal dimension data provides the historical trajectory and trend of the device's state evolution over time. Unlike traditional static monitoring systems, it can capture the complete picture of the device's state changes over time, providing the necessary data foundation for fault prediction and anomaly monitoring. Compared with existing technologies, this method of acquiring dynamic data overcomes the limitations of relying solely on static data. Overcoming the limitations of traditional two-dimensional monitoring, this method comprehensively and in real-time reflects the actual operating status of equipment, avoiding delayed responses to faults caused by the inability to detect changes in equipment promptly. Rendering visual attribute data onto corresponding equipment model nodes binds the real equipment status to the 3D model, allowing users to directly observe the real-time status of the equipment in 3D space. This avoids the loss of spatial information found in traditional 2D monitoring interfaces. For example, information such as equipment temperature, humidity, and energy consumption is intuitively displayed on the equipment nodes through color changes and brightness adjustments. The biggest difference from existing technologies is that it can accurately reflect the location and proximity of equipment through the 3D model, not only improving the visualization of equipment status but also providing clear spatial references to help users better understand the equipment. Compared to traditional single-data source presentation methods, 3D models provide a more comprehensive and three-dimensional information display of the system's operational status, enhancing its operability and intuitiveness. By dynamically updating the rendered visual attribute data on the corresponding device model nodes in real time using time-dimensional data, dynamic visual rendering data is obtained. This time-dimensional data represents the device's operational status and trends at different times. By dynamically updating the device model's visual attributes, changes in device status can be presented in real time, significantly improving the system's real-time monitoring capabilities. This allows the monitoring platform to not only display the current status but also reflect the evolution of the device's historical status, further enhancing its effectiveness. This invention provides early warnings of equipment operating trends and potential faults. Compared to existing static or periodically updated monitoring systems, it offers more accurate and timely information to maintenance personnel through real-time dynamic data updates. This reduces equipment fault diagnosis and response time, and improves the system's predictive and early warning capabilities. By acquiring abnormal node identifiers, abnormal attributes, and abnormal visual parameters from abnormal visual feedback, the system determines the location of abnormal nodes in the 3D equipment model using the abnormal node identifiers and renders these locations based on the abnormal visual parameters, obtaining abnormal node rendering data. By identifying abnormal nodes and their corresponding attribute changes, the system can accurately locate the equipment where faults or potential faults have occurred. The abnormal visual parameters define how these anomalies are presented.For example, using specific colors, shapes, or flashing effects to emphasize problem areas, existing technologies often cannot accurately distinguish between normal and abnormal equipment states and lack clear feedback mechanisms. However, this invention, by combining abnormal attributes with visual feedback parameters, can accurately and intuitively present abnormal areas and attributes to users, improving the efficiency of fault location and quickly guiding maintenance personnel to address the problem, avoiding equipment damage or production stoppages caused by slow response. By identifying abnormal nodes and determining their specific locations in the 3D equipment model, the problematic equipment can be accurately mapped into the actual 3D space. This not only improves the accuracy of fault location but also allows maintenance personnel to intuitively see the spatial location of the fault point by rendering the abnormal node position. By incorporating abnormal visual parameters (such as color changes and flashing effects), the intuitiveness and alertness of the feedback are further enhanced. Compared with existing two-dimensional display methods, this invention provides a clearer display of fault information through three-dimensional visualization, which is particularly suitable for complex industrial plant layouts. It can help maintenance personnel quickly identify the problem area and its impact range, thereby accelerating fault handling. By obtaining the associated device nodes with abnormal attributes and connecting the associated device nodes with the abnormal node location as the center, the abnormal propagation path can be obtained. By establishing the relationship between the abnormal node and the associated device nodes, the propagation path of the abnormal problem in the device network can be revealed. For example, if device A fails, it may cause interconnected problems with connected devices B, C, D, etc. By identifying these associated devices and establishing the propagation path, the chain of fault propagation can be accurately depicted, helping maintenance personnel identify devices that may be affected by the chain reaction. The main difference from existing technologies is that it can provide spatial correlation of fault propagation between devices, avoiding the shortcomings of traditional systems that cannot accurately display the scope and impact of fault spread, and helping maintenance personnel to anticipate the problem. To prevent potential cascading failures and improve system fault tolerance and operational efficiency, this invention presents spatial visualization feedback in the form of heatmaps, displaying anomaly propagation paths, dynamic visual rendering data, and anomaly node rendering data. This allows for a direct visualization of the abnormal states and trends of various devices within the space. The heatmap clearly highlights abnormal hotspots and uses color or brightness changes to indicate the severity of faults or potential risks in different areas. This spatial visualization feedback provides maintenance personnel with an intuitive and easy-to-understand way to assess equipment operating status and anomaly propagation paths. Compared to traditional charts or single-view displays, heatmaps combine dynamic, anomaly, and propagation path information, providing users with a more comprehensive and in-depth spatial data analysis perspective, leading to more accurate decision-making and faster responses. This invention, by introducing multi-dimensional technologies such as dynamic rendering, time dimension, anomaly feedback, associated devices, and heatmaps, provides a comprehensive, 3D visualization-based industrial equipment monitoring method.This significantly improves the efficiency and accuracy of equipment monitoring, fault location, anomaly early warning, data integration, and fault response.

[0152] like Figure 2 As shown, this application also provides an IoT visualization monitoring system based on three-dimensional Gaussian splashing, comprising:

[0153] The first acquisition module is used to acquire the physical attribute information, spatial location coordinates and heterogeneous data source set of each target industrial equipment in the industrial plant, and classify the corresponding heterogeneous data source set according to each physical attribute information to obtain classified data source, wherein the classified data source includes core data source, auxiliary data source and edge data source;

[0154] The reconstruction module is used to reconstruct three-dimensional models of all target industrial equipment based on the physical attribute information and spatial position coordinates of the three-dimensional Gaussian splashing algorithm, and to obtain the model topology and surface point cloud data of the three-dimensional equipment model.

[0155] The binding module is used to assign a unique device identifier to the corresponding target industrial equipment based on each category data source, and to associate and bind each unique device identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish a device-model spatial association binding table.

[0156] The second acquisition module is used to acquire the real-time collected data and data collection timestamp of each category data source based on the device-model space association binding table, and add time dimension labels to the corresponding real-time collected data according to each data collection timestamp to obtain time dimension data;

[0157] The association module is used to divide the three-dimensional device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associate each device model node with the corresponding time dimension data to obtain a data-model association mapping relationship;

[0158] The judgment module is used to monitor the rate of change of the values ​​of each category data source in real time and determine whether the rate of change of the values ​​is greater than a preset change threshold.

[0159] If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback.

[0160] In one embodiment, the binding module includes:

[0161] The first acquisition unit is used to acquire the device function type, device installation time and device spatial coordinates according to the classification data source, and to acquire the device type code according to the device function type;

[0162] The second acquisition unit is used to acquire an installation time code based on the device installation time and to acquire a spatial coordinate code based on the device spatial coordinates.

[0163] The splicing unit is used to splice the device type code, installation time code, and spatial coordinate code in chronological order to obtain a unique device identifier;

[0164] The third acquisition unit is used to acquire the second center coordinates of the three-dimensional device model, and to acquire the coordinate deviation value based on the second center coordinates and the spatial position coordinates;

[0165] The first binding unit is used to bind a unique device identifier to the spatial coordinates of the target industrial equipment according to the coordinate deviation value;

[0166] The fourth acquisition unit is used to acquire the spatial distance between each three-dimensional coordinate and the spatial position coordinate based on the three-dimensional coordinate of each spatial point in the surface point cloud data.

[0167] The second binding unit is used to bind the associated spatial point to the unique device identifier according to the spatial distance, so as to obtain the device-model spatial association binding table.

[0168] It should be noted that each module and unit in the IoT visualization monitoring system based on 3D Gaussian splashing corresponds one-to-one with the steps in the IoT visualization monitoring method based on 3D Gaussian splashing.

[0169] like Figure 3 As shown, this application also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores all data required for the process of the IoT visualization and monitoring method based on 3D Gaussian splashing. The network interface is used for communication with external terminals via network connection. When the computer program is executed by the processor, it implements the IoT visualization and monitoring method based on 3D Gaussian splashing.

[0170] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment on which the present application is applied.

[0171] An embodiment of this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the above-described IoT visualization and monitoring methods based on three-dimensional Gaussian splashing.

[0172] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0173] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0174] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A visualization and monitoring method for the Internet of Things based on three-dimensional Gaussian splashing, characterized in that, include: The physical attribute information, spatial location coordinates, and heterogeneous data source set of each target industrial equipment in the industrial plant are obtained, and the corresponding heterogeneous data source set is classified according to each physical attribute information to obtain classified data sources, wherein the classified data sources include core data sources, auxiliary data sources, and edge data sources; Based on the three-dimensional Gaussian splashing algorithm, the physical attribute information and spatial position coordinates of all target industrial equipment are used to reconstruct three-dimensional models to generate three-dimensional equipment models, and the model topology and surface point cloud data of the three-dimensional equipment models are obtained. Assign a unique equipment identifier to the corresponding target industrial equipment based on each category data source, and associate and bind each unique equipment identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish an equipment-model spatial association binding table; Based on the device-model space association binding table, obtain the real-time collected data and data collection timestamp of each category data source, and add time dimension labels to the corresponding real-time collected data according to each data collection timestamp to obtain time dimension data; The three-dimensional device model is divided into nodes according to the model topology to obtain multiple device model nodes. Each device model node is dynamically associated with the corresponding time dimension data to obtain a data-model association mapping relationship. Real-time monitoring of the rate of change of values ​​for each category data source, and determination of whether the rate of change of values ​​is greater than a preset change threshold; If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback. The step of generating a 3D equipment model by reconstructing the physical attribute information and spatial location coordinates of all target industrial equipment based on the 3D Gaussian splashing algorithm includes: The equipment density, length, width, and height are obtained based on the physical attribute information of the target industrial equipment, and the three-dimensional reconstruction bounding box is determined based on the spatial location coordinates, equipment length, width, and height. Obtain the maximum size and preset sampling accuracy of the target industrial equipment, and obtain the number of grid division dimensions based on the maximum size and preset sampling accuracy; The 3D reconstructed bounding box is divided into multiple 3D mesh units according to the number of mesh division dimensions, and the first center coordinates of each 3D mesh unit are obtained; Obtain the maximum density, maximum length, maximum width, and maximum height of all target industrial equipment, and determine the density weight ratio based on the maximum density and equipment density; The length weight percentage is obtained based on the maximum length and the device length, and the width weight percentage is obtained based on the maximum width and the device width. The height weight ratio is obtained based on the maximum height and the device height. The total splash value of the corresponding 3D mesh cell is calculated based on the height weight ratio, density weight ratio, length weight ratio, width weight ratio, the first center coordinate of each 3D mesh cell, and the coordinates of all spatial points. The calculation formula is as follows: ； Where Z(PJ) represents the total splash value, M(DQ) represents the density weight percentage, C(DQ) represents the length weight percentage, K(DQ) represents the width weight percentage, G(DQ) represents the height weight percentage, x represents the x-coordinate of the center of the 3D mesh cell, y represents the y-coordinate of the center of the 3D mesh cell, and z represents the y-coordinate of the center of the 3D mesh cell. i Y represents the x-coordinate of the i-th spatial point. i Z represents the vertical coordinate of the i-th spatial point. i Let represent the ordinate of the i-th spatial point, where i represents the index of the spatial point and N represents the number of spatial points; Determine whether the total splash value is less than a preset splash value; If the total splash value is less than the preset splash value, the three-dimensional mesh cell corresponding to the total splash value is determined to be an invalid three-dimensional mesh cell; If the total splash value is not less than the preset splash value, the three-dimensional mesh unit corresponding to the total splash value is determined to be a valid three-dimensional mesh unit, and all valid three-dimensional mesh units are reconstructed using the moving cube algorithm to generate a three-dimensional device model.

2. The IoT visualization monitoring method based on three-dimensional Gaussian splashing according to claim 1, characterized in that, The step of classifying the corresponding heterogeneous data source set according to each physical attribute information to obtain classified data sources includes: Each of the physical attribute information is converted into a quantifiable parameter value, and a corresponding device feature vector is constructed based on multiple parameter values; Obtain the total association weight score between each parameter value and all data sources in the heterogeneous data source set, and divide the device feature vector into core feature dimensions and auxiliary feature dimensions according to the total association weight score. The core feature dimensions include device function type and device rated parameters, and the auxiliary feature dimensions include device volume features and device installation height. Extract data features from all data sources in each heterogeneous data source set, and obtain a function matching coefficient based on each data feature and device function type; Parameter adaptation coefficients are obtained based on each of the data features and device rated parameters, and core correlation is obtained based on each of the parameter adaptation coefficients and function matching coefficients. The core correlation degree is corrected based on the device volume characteristics and device installation height to obtain the corrected correlation degree, and the relationship between the corrected correlation degree and the classification threshold range is analyzed. If the corrected correlation is greater than the upper limit of the classification threshold range, then the data source is determined to be a core data source. If the corrected correlation degree is within the classification threshold range, then the data source is determined to be an auxiliary data source; If the corrected correlation is less than the lower limit of the classification threshold range, the data source is determined to be an edge data source.

3. The IoT visualization monitoring method based on three-dimensional Gaussian splashing according to claim 1, characterized in that, The steps of assigning a unique equipment identifier to the corresponding target industrial equipment based on each category data source, and associating and binding each unique equipment identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish an equipment-model spatial association binding table include: Based on the data source classification, obtain the device function type, device installation time, and device spatial coordinates, and obtain the device type code based on the device function type; The installation time code is obtained based on the equipment installation time, and the spatial coordinate code is obtained based on the equipment spatial coordinates; The device type code, installation time code, and spatial coordinate code are concatenated in chronological order to obtain a unique device identifier; Obtain the second center coordinates of the 3D device model, and obtain the coordinate deviation value based on the second center coordinates and the spatial position coordinates; The unique device identifier is bound to the spatial coordinates of the target industrial equipment based on the coordinate deviation value; Based on the three-dimensional coordinates of each spatial point in the surface point cloud data, the spatial distance between each three-dimensional coordinate and the spatial position coordinate is obtained; Based on the spatial distance, the associated spatial points are bound to unique device identifiers to obtain the device-model spatial association binding table.

4. The IoT visualization monitoring method based on three-dimensional Gaussian splashing according to claim 1, characterized in that, The step of dividing the 3D device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associating each device model node with corresponding time dimension data to obtain a data-model association mapping relationship includes: Extract all triangular facets in the model's topology and obtain the centroid coordinates of each triangular facet; Obtain the equipment function type of the target industrial equipment corresponding to the model topology, and obtain the core functional area based on the equipment function type; Triangular facets whose centroid coordinates are located within the core functional area are classified as core facets, and triangular facets whose centroid coordinates are not located within the core functional area are classified as non-core facets. Obtain the shortest path distance between each non-core surface element and the core surface element, and determine whether the shortest path distance is greater than a preset distance; If the shortest path distance is greater than the preset distance, then the non-core surface element is classified as an auxiliary surface element; If the shortest path distance is not greater than the preset distance, then the non-core surface element is divided into edge surface elements, and the corresponding device model node is obtained based on the edge surface elements, auxiliary surface elements and core surface elements. Obtain the data type for each time dimension, wherein the data type is one of core data, auxiliary data, and peripheral data; The core data, auxiliary data, and edge data are dynamically associated one-to-one with the core surface elements, edge surface elements, and auxiliary surface elements in the device model node, respectively, to obtain the data-model association mapping relationship.

5. The IoT visualization monitoring method based on three-dimensional Gaussian splashing according to claim 1, characterized in that, The step of generating spatial visualization feedback based on the dynamic rendering and abnormal visual feedback includes: Obtain the visual attribute data and time dimension data of the dynamic rendering, and render the visual attribute data on the corresponding device model node; Based on the time dimension data, the visual attribute data rendered on the corresponding device model node will be dynamically updated in real time to obtain dynamic visual rendering data. Obtain the abnormal node identifier, abnormal attributes, and abnormal visual parameters of the abnormal visual feedback; The abnormal node position in the 3D device model is determined based on the abnormal node identifier, and the abnormal node position is rendered based on the abnormal visual parameters to obtain abnormal node rendering data. Obtain the associated device nodes of the abnormal attribute, and connect the associated device nodes with the abnormal node location as the center to obtain the abnormal propagation path; The abnormal transmission path, dynamic visual rendering data, and abnormal node rendering data are presented as a spatially visualized distribution feedback in the form of a heatmap.

6. An IoT visualization monitoring system based on three-dimensional Gaussian splashing, characterized in that, include: The first acquisition module is used to acquire the physical attribute information, spatial location coordinates and heterogeneous data source set of each target industrial equipment in the industrial plant, and classify the corresponding heterogeneous data source set according to each physical attribute information to obtain classified data source, wherein the classified data source includes core data source, auxiliary data source and edge data source; The reconstruction module is used to reconstruct three-dimensional models of all target industrial equipment based on the physical attribute information and spatial position coordinates of the three-dimensional Gaussian splashing algorithm, and to obtain the model topology and surface point cloud data of the three-dimensional equipment model. The binding module is used to assign a unique device identifier to the corresponding target industrial equipment based on each category data source, and to associate and bind each unique device identifier with the spatial location coordinates and surface point cloud data of the corresponding target industrial equipment to establish a device-model spatial association binding table. The second acquisition module is used to acquire the real-time collected data and data collection timestamp of each category data source based on the device-model space association binding table, and add time dimension labels to the corresponding real-time collected data according to each data collection timestamp to obtain time dimension data; The association module is used to divide the three-dimensional device model into nodes according to the model topology to obtain multiple device model nodes, and dynamically associate each device model node with the corresponding time dimension data to obtain a data-model association mapping relationship; The judgment module is used to monitor the rate of change of the values ​​of each category data source in real time and determine whether the rate of change of the values ​​is greater than a preset change threshold. If the rate of change of the numerical value is greater than the preset change threshold, then abnormal visual feedback and dynamic rendering are triggered on the corresponding device model node according to the data-model association mapping relationship, and spatial visualization feedback is generated according to the dynamic rendering and abnormal visual feedback. The step of generating a 3D equipment model by reconstructing the physical attribute information and spatial location coordinates of all target industrial equipment based on the 3D Gaussian splashing algorithm includes: The equipment density, length, width, and height are obtained based on the physical attribute information of the target industrial equipment, and the three-dimensional reconstruction bounding box is determined based on the spatial location coordinates, equipment length, width, and height. Obtain the maximum size and preset sampling accuracy of the target industrial equipment, and obtain the number of grid division dimensions based on the maximum size and preset sampling accuracy; The 3D reconstructed bounding box is divided into multiple 3D mesh units according to the number of mesh division dimensions, and the first center coordinates of each 3D mesh unit are obtained; Obtain the maximum density, maximum length, maximum width, and maximum height of all target industrial equipment, and determine the density weight ratio based on the maximum density and equipment density; The length weight percentage is obtained based on the maximum length and the device length, and the width weight percentage is obtained based on the maximum width and the device width. The height weight ratio is obtained based on the maximum height and the device height. The total splash value of the corresponding 3D mesh cell is calculated based on the height weight ratio, density weight ratio, length weight ratio, width weight ratio, the first center coordinate of each 3D mesh cell, and the coordinates of all spatial points. The calculation formula is as follows: ； Where Z(PJ) represents the total splash value, M(DQ) represents the density weight percentage, C(DQ) represents the length weight percentage, K(DQ) represents the width weight percentage, G(DQ) represents the height weight percentage, x represents the x-coordinate of the center of the 3D mesh cell, y represents the y-coordinate of the center of the 3D mesh cell, and z represents the y-coordinate of the center of the 3D mesh cell. i Y represents the x-coordinate of the i-th spatial point. i Z represents the vertical coordinate of the i-th spatial point. i Let represent the ordinate of the i-th spatial point, where i represents the index of the spatial point and N represents the number of spatial points; Determine whether the total splash value is less than a preset splash value; If the total splash value is less than the preset splash value, the three-dimensional mesh cell corresponding to the total splash value is determined to be an invalid three-dimensional mesh cell; If the total splash value is not less than the preset splash value, the three-dimensional mesh unit corresponding to the total splash value is determined to be a valid three-dimensional mesh unit, and all valid three-dimensional mesh units are reconstructed using the moving cube algorithm to generate a three-dimensional device model.

7. The IoT visualization monitoring system based on three-dimensional Gaussian splashing according to claim 6, characterized in that, The binding module includes: The first acquisition unit is used to acquire the device function type, device installation time and device spatial coordinates according to the classification data source, and to acquire the device type code according to the device function type; The second acquisition unit is used to acquire an installation time code based on the device installation time and to acquire a spatial coordinate code based on the device spatial coordinates. The splicing unit is used to splice the device type code, installation time code, and spatial coordinate code in chronological order to obtain a unique device identifier; The third acquisition unit is used to acquire the second center coordinates of the three-dimensional device model, and to acquire the coordinate deviation value based on the second center coordinates and the spatial position coordinates; The first binding unit is used to bind a unique device identifier to the spatial coordinates of the target industrial equipment according to the coordinate deviation value; The fourth acquisition unit is used to acquire the spatial distance between each three-dimensional coordinate and the spatial position coordinate based on the three-dimensional coordinate of each spatial point in the surface point cloud data. The second binding unit is used to bind the associated spatial point to the unique device identifier according to the spatial distance, so as to obtain the device-model spatial association binding table.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images