A visual rapid deployment method and system for intelligent operation and maintenance scenarios

Abstract

This application relates to the field of intelligent operation and maintenance technology, and discloses a visual rapid deployment method and system for intelligent operation and maintenance scenarios. The method includes data access, diagnostic verification, fine-tuning and release, and operation scheduling steps. It parses multi-source heterogeneous data to generate device instances and constructs a business logic topology map. It calculates an adaptive physical allowable distance threshold to filter abnormal device instances, uses virtual force and simulated annealing algorithms to deduce recommended topology anchor points, performs coordinate rigid adsorption based on the dynamic magnetic adsorption range in the 3D viewport, generates a topology association loading weight matrix, calculates a comprehensive score by combining the basic loading priority weight and the topology association weight, and performs cross-line-of-sight high-precision model pre-assembly. This invention achieves spatial layout, business logic, and anomaly correction by calculating physical distance thresholds and force calculations; improves layout positioning accuracy through dynamic magnetic adsorption; and solves the problem of lagging loading of associated device models in emergency alarm states through a cross-line-of-sight scheduling mechanism.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology, specifically to a visual and rapid deployment method and system for intelligent operation and maintenance scenarios. Background Technology

[0002] In the construction of industrial 3D visualization and intelligent operation and maintenance systems, scene construction typically requires associating and spatially arranging business system data with 3D models. Existing 3D scene deployments mostly employ manual static layout methods, lacking a verification mechanism between the physical spatial characteristics of equipment and the business logic relationships. When faced with a large number of devices and complex process connections, this method cannot automatically identify anomalies such as model overlap or process link deviations, resulting in a significant workload for scene verification.

[0003] When adjusting the position of abnormal equipment models, current methods mostly rely on operators visually viewing and manually dragging them in a 3D perspective. In areas with dense equipment, due to limitations in viewing angle and operational precision, manual dragging makes it difficult to accurately place the model at coordinate positions that meet the requirements of physical spacing and logical connectivity, often requiring repeated verification and correction afterward.

[0004] During system operation, conventional 3D rendering engines primarily allocate rendering resources based on camera view distance and spatial occlusion. This scheduling method only focuses on spatial visibility and is not related to the underlying business logic status. When equipment alarms occur in the industrial field, the system cannot overcome physical view distance limitations to pre-execute model assembly for critical equipment that is out of view but has business relevance. When operators shift their perspective to conduct emergency troubleshooting, the high-precision models of related equipment need to be calculated and loaded temporarily, resulting in screen display lag and reducing the overall system response efficiency. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a visual and rapid deployment method and system for intelligent operation and maintenance scenarios. It solves the problems of large workload for manual layout and verification in existing 3D scene construction, low accuracy of manual drag-and-drop positioning, and delayed loading of critical equipment models in emergency situations due to the rendering scheduling mechanism being out of touch with business logic.

[0006] To achieve the above objectives, the first aspect of this invention provides a visual and rapid deployment method for intelligent operation and maintenance scenarios, the method comprising the following steps: Data access steps: Connect to external business systems in the industrial field to obtain multi-source heterogeneous data, parse and map according to standardized templates of equipment attributes and behaviors, generate a standardized set of equipment instance objects, and establish model binding relationships with 3D model assets; parse the process connectivity relationships in the data of external business systems to extract graph node and connectivity edge data, construct a scene business logic topology graph and write it into the data carrier.

[0007] Diagnostic verification steps: Read the standardized set of equipment instance objects from the data carrier, perform data mapping between equipment instance objects and measurement point parameters, infer the three-dimensional initial physical coordinates of the equipment instance objects, and extract the corresponding spatial bounding box scale and logical space region attributes; based on the three-dimensional initial physical coordinates, spatial bounding box scale, and connectivity in the scene business logic topology diagram, calculate the adaptive physical allowable distance threshold and the actual physical Euclidean distance between nodes; compare the distance parameters and filter out abnormal equipment instance objects that trigger spatial logic deviation anomalies; move the abnormal equipment instance objects into the abnormal isolation area, extract stable adjacent nodes to construct a virtual force model, apply the simulated annealing algorithm to iteratively calculate the convergence coordinates, and output the topology recommended anchor point.

[0008] Fine-tuning and publishing steps: Draw the abnormal device instance object and its corresponding hidden magnetic adsorption range in the visualization editing auxiliary layer of the 3D rendering viewport; receive drag coordinates, and when the capture radius condition corresponding to the magnetic adsorption range is entered, trigger the rigid adsorption command to update the coordinates to the topology recommended anchor point; after completing the coordinate correction of the abnormal object and passing the legality review, generate the topology association loading weight matrix according to the shortest logical connectivity hop count, write the updated data into the data carrier, and execute the publishing action.

[0009] The operation scheduling steps are as follows: Receive and parse the published data carrier; calculate the basic loading priority weight based on camera line of sight and occlusion culling algorithm during normal operation; after listening for and intercepting device alarm events, extract the topology-related loading weight from the topology-related loading weight matrix; calculate the comprehensive scheduling priority score by combining the basic loading priority weight and the topology-related loading weight; reconstruct the differential update queue based on the score; prioritize the extraction of high-precision 3D model resources for objects whose scores meet the threshold and perform background pre-assembly; and perform visible rendering instantiation when the object meets the visibility conditions.

[0010] Preferably, the standardized template for equipment attributes and behaviors includes a built-in equipment type coding dictionary and attribute field mapping rules. In the data access step, the conversion coefficients for data with different units of measurement are adjusted, and the system default values ​​are extracted and filled for missing fields of key attributes. The connected edges in the scenario business logic topology diagram are attached with logical attribute data, which includes process medium type, medium flow direction characteristics, process area codes of the two end nodes, and the corresponding basic connection theory allowable distance index.

[0011] Preferably, the data mapping between the device instance object and the measurement point parameter adopts a three-level mapping mechanism including rule matching, semantic matching and strong data type validation: the device tag number and the measurement point label are compared by regular expression to extract the initial matching pairs, the semantic similarity score of the Chinese description text is calculated based on the string similarity algorithm to filter out interfering mapping items, the data protocol type and internal attribute type returned by the measurement point channel are read and strongly validated to eliminate invalid associations, and the corresponding data mapping confidence is generated based on the three-level filtering results.

[0012] Preferably, the process of inferring the initial three-dimensional physical coordinates of the device instance object is as follows: parsing the region hierarchy description and querying the preset spatial reference table to obtain the basic origin coordinates; using regular matching rules to extract the relative offset values ​​of the attached direction description and distance units and convert them into coordinate axis numerical increments; performing three-dimensional vector addition operation on the basic origin coordinates and numerical increments to obtain the inferred initial three-dimensional physical coordinates; the spatial bounding box scale is the diagonal length scale calculated by constructing the circumscribed three-dimensional axial bounding box that encloses the three-dimensional model.

[0013] Preferably, the adaptive physical allowable distance threshold between nodes is calculated by: obtaining the basic theoretical allowable distance of the process medium corresponding to the connected link; extracting the spatial bounding box scale of the nodes at both ends of the link, selecting the maximum value and multiplying it by a preset model scale compensation coefficient to obtain the scale compensation increment; extracting the logical space region attributes of the nodes at both ends of the link, and querying a preset cross-region compensation matrix to obtain the tolerance increment value when it is determined that different physical regions are crossed; and adding the basic theoretical allowable distance, the scale compensation increment, and the tolerance increment value to generate the adaptive physical allowable distance threshold.

[0014] Preferably, the process of generating the preset cross-region compensation matrix is ​​as follows: extract the three-dimensional boundary features of all independent logical space regions in the scene; for any two different physical regions, use computational geometry retrieval algorithm to find the two boundary anchor points with the closest spatial distance on the outer surface; calculate the three-dimensional Euclidean distance between the two anchor points and superimpose the preset building barrier margin parameter to derive the shortest mapping distance; use the shortest mapping distance as the tolerance increment value at the intersection of the row and column; and uniformly fix the diagonal elements to zero.

[0015] Preferably, the method for screening abnormal device instance objects that trigger spatial logical deviation anomalies includes: determining whether the actual physical Euclidean distance is greater than the adaptive physical allowable distance threshold, or calling the regional compatibility rule table to determine whether there are mutual exclusion restrictions on the logical spatial region attributes of the two-end nodes; when any deviation condition is met, the nodes at both ends of the connected link are written into the candidate anomaly set; by comprehensively comparing the data mapping confidence of the nodes, the integrity of the regional attributes, the number of participating abnormal edges, and the consistency of the distance with stable adjacent nodes, the abnormal device instance objects that are moved into the anomaly isolation zone are finally determined.

[0016] Preferably, a virtual force model is constructed by extracting stable adjacent nodes, specifically including: tracing stable adjacent nodes that are directly or indirectly connected to the abnormal device instance object and have not experienced any abnormalities in the scene's business logic topology graph as fixed anchor points; calculating the resultant force vector of spring attraction from the abnormal node to each stable adjacent node based on the deformation principle of Hooke's Law and the deviation between the actual physical distance and the ideal arrangement physical distance; traversing the bounding box of the abnormal node to retrieve other entity nodes within its search radius, and when it is determined that bounding box overlap has occurred or the distance is lower than the safe distance, generating an anti-overlap repulsive force vector with a magnitude proportional to the overlap depth along the opposite direction of the center connection line; and adding the resultant force vector of spring attraction to all anti-overlap repulsive force vectors to form the total force vector.

[0017] Preferably, the process of iteratively calculating the convergent coordinates using the simulated annealing algorithm is as follows: the coordinate update step size is controlled by a temperature function that decays exponentially with the number of iterations; the previous generation of iterated coordinates is added to the total force vector adjusted by the decaying temperature function to obtain the new generation of coordinates; the iteration stops when the current temperature is lower than the freezing temperature threshold, or the magnitude of the difference between the position coordinate vectors of two consecutive iterations is less than the minimum displacement threshold; anti-overlap detection is performed on the final coordinates when the iteration stops, and when no overlap occurs, it is output as the topology recommended anchor point.

[0018] Preferably, when drawing the hidden magnetic adsorption range, the interactive capture radius is dynamically calculated: the physical distance from the virtual camera in the current rendering viewport to the topology recommended anchor point is obtained, the distance is multiplied by the preset interactive ratio coefficient, and the amplitude is truncated between the preset minimum capture radius and the maximum capture radius. The amplitude truncating result is used as the final capture radius of the invisible spherical trigger area in the current frame.

[0019] Preferably, the legality verification process after the interactive correction and locking of coordinates is as follows: re-execute distance threshold verification, regional compatibility verification, and bounding box overlap verification. If any verification fails, the current rigid coordinate locking state is released, the corresponding abnormal device instance object is returned to the abnormal isolation area, and a re-correction request carrying the current state is sent to the internal interface to trigger the anchor point recalculation process.

[0020] Preferably, the method for generating the topology association loading weight matrix is ​​as follows: use the breadth-first search algorithm to traverse the scenario business logic topology graph to calculate the shortest logical connectivity hop count between any two nodes, combine the preset business attenuation constant, use the exponential decay mapping algorithm to calculate the topology association loading weight score between nodes and fill it into the matrix; for node pairs that belong to disconnected isolated subgraphs, set their shortest logical connectivity hop count to infinity and assign the corresponding weight score to zero.

[0021] Preferably, the method for calculating the basic loading priority weight includes: obtaining the three-dimensional coordinate physical view distance between the virtual camera and the device instance object, and determining the visibility under the view frustum culling and occlusion culling mechanisms; when the camera is visible and the physical view distance is less than the global maximum visible view distance threshold, calculating the basic loading priority weight based on the ratio difference between the physical view distance and the global maximum visible view distance threshold; when the camera is invisible or the physical view distance is greater than or equal to the threshold, forcibly assigning the basic loading priority weight to zero.

[0022] Preferably, the process of calculating the comprehensive scheduling priority score by combining the basic loading priority weight and the topology-related loading weight is as follows: a dynamic weight mixing factor driven by business status is introduced, and a linear calculation method is adopted to proportionally superimpose the basic loading priority weight and the topology-related loading weight according to the mixing factor; during normal operation, the topology-related loading weight is forced to be calculated as zero and the mixing factor is set to an extreme value to maintain the line-of-sight-dominated strategy; when an alarm event is intercepted, the alarm node is locked to extract the associated row vector, and the value of the mixing factor is simultaneously reduced to increase the proportion of the topology-related loading weight in the comprehensive score.

[0023] Preferably, high-precision 3D model resources are extracted first for objects whose scores meet the threshold. The forced loading threshold is generated by a dynamic memory measurement mechanism: the available memory ratio of the current graphics processor is monitored in real time. When it is higher than the preset upper limit of the available ratio, the forced loading threshold is reduced. When it is lower than the preset lower limit of the available ratio, the forced loading threshold is increased. Adaptive scaling limits are applied to the number of high-precision models assembled across the line of sight.

[0024] Preferably, the cross-view distance intervention and pre-assembly logic includes: exceeding the limitations of the camera's physical view distance and layer detail rendering, directly retrieving the highest precision mesh data and high-resolution texture of the object that meets the score requirements, performing memory decoding and caching, maintaining the invisible attribute in the background until the physical occlusion state of the rendering viewport is lifted, and maintaining a low precision placeholder rendering state for device objects whose comprehensive scheduling priority score is lower than the forced loading threshold.

[0025] The second aspect of this invention provides a visualized rapid deployment system for intelligent operation and maintenance scenarios, comprising: a data carrier, which acts as a distributed node for the entire data chain and is used to persistently store business data, topological relationships, parameter matrices, and spatial features during the scenario construction process; a data access module, which connects to external sources to obtain multi-source heterogeneous data mapping to generate a standardized set of equipment instance objects, extracts process connectivity relationships to construct a scenario business logic topology diagram; a diagnostic verification module, which performs data mapping pre-binding and initial physical coordinate inference, calculates adaptive physical distance thresholds to perform deviation verification, and uses virtual force and simulated annealing algorithms to automatically correct the deviation of isolated objects and outputs topology recommended anchor points; a fine-tuning and publishing module, which performs drag-and-drop blocking fine-tuning calculation of the topology association loading weight matrix based on the hidden magnetic attraction range in the three-dimensional viewport and broadcasts the calculation; and a running scheduling module, which parses the data carrier, calculates a comprehensive score based on the line-of-sight base weight and the topology association weight triggered by alarms, and performs cross-line-of-sight dimension pre-assembly scheduling intervention for video memory resources.

[0026] This invention provides a visual and rapid deployment method and system for intelligent operation and maintenance scenarios. It has the following beneficial effects: 1. This invention achieves automated cross-validation of physical space coordinates and business logic connectivity by extracting the 3D bounding box scale and logical space region attributes of devices and calculating adaptive physical allowable distance thresholds in conjunction with the topological connectivity of scene business logic. Furthermore, by extracting stable adjacent nodes to construct a virtual force model and applying simulated annealing algorithm for coordinate convergence calculation, it can automatically identify and correct abnormal device layouts with spatial overlap or logical deviations, reducing the workload of manual verification during the 3D scene construction phase.

[0027] 2. This invention introduces a dynamic magnetic adsorption fine-tuning mechanism into the 3D rendering viewport, dynamically calculating the interactive capture radius based on the physical distance from the virtual camera to the topologically recommended anchor point. When an operator drags an abnormal device instance into this trigger area, the system automatically executes a rigid adsorption command to update the device coordinates to the algorithm-deduced recommended position, and simultaneously performs a legality check on distance thresholds, regional compatibility, and spatial overlap, avoiding spatial deviations caused by manual visual dragging and positioning, and improving the accuracy of high-density model layout.

[0028] 3. This invention calculates the line-of-sight-based loading priority weight and topology-related loading weight, and introduces a dynamic hybrid factor to generate a comprehensive scheduling score. When the system intercepts a device alarm event, it can adaptively adjust the hybrid factor to amplify the topology-related weight, prioritizing background pre-assembly for critical devices that are outside the physical line of sight but closely related to business logic. Simultaneously, it works in conjunction with a dynamically generated forced loading threshold based on available video memory, improving the instantiation and rendering speed of high-precision models in emergency maintenance situations while preventing video memory overflow. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram illustrating the data access and business logic topology extraction principle of the present invention. Figure 4 This is a schematic diagram illustrating the data mapping and initial spatial feature extraction principle of the present invention. Figure 5 This is a schematic diagram of the spatial logic cross constraint diagnosis and adaptive correction calculation principle of the present invention; Figure 6 This is a schematic diagram illustrating the principle of generating the visual magnetic fine-tuning and topological correlation loading weight matrix of the present invention. Figure 7 This is a schematic diagram of the topology-driven runtime dimension reduction intervention and dynamic assembly principle of the present invention. Figure 8 This is a three-dimensional spatial correction trajectory diagram based on virtual force and simulated annealing according to the present invention; Figure 9 This is a dual Y-axis trend graph showing the temperature decay and displacement convergence of the simulated annealing algorithm of this invention. Figure 10 The above is a time series graph comparing rendering performance under the sudden alarm state of the present invention. (a) is a graph comparing the ready rate of the high-precision model associated in the video memory of the two methods from 10s to 30s, and (b) is a graph comparing the scene rendering frame rate of the two methods.

[0030] Among them, 10 is the data access module; 20 is the diagnostic verification module; 30 is the fine-tuning and release module; 40 is the operation scheduling module; and 50 is the data carrier. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] See attached document Figure 1 , Figure 1 This is a schematic diagram of the system architecture of a visual rapid deployment system for intelligent operation and maintenance scenarios provided by an embodiment of the present invention. The present invention provides a visual rapid deployment system for intelligent operation and maintenance scenarios, the system comprising: The data access module 10 is used to read multi-source heterogeneous business data, equipment ledger data, real-time location lists, and 3D model assets input from external business systems. It generates a standardized set of equipment instance objects based on standardized templates for equipment attributes and behaviors, and parses process connectivity relationships to construct a scenario business logic topology diagram. As the data aggregation entry point for the underlying system, this module is responsible for parsing multi-source heterogeneous industrial asset files and initializing a unified format data carrier in memory, thereby providing structured basic data support for subsequent spatial mapping and physical logic cross-validation.

[0033] The diagnostic verification module 20 is used to perform data mapping pre-binding between device instance objects and external measurement point parameters, and infers the initial 3D physical coordinates of the device instance objects based on the parsed regional hierarchical description. To prevent the front-end 3D spatial layout from deviating from the underlying process logic, this module calculates the adaptive physical allowable distance threshold between nodes based on the spatial bounding box scale of the device and the business connectivity relationship, and thereby filters out abnormal device nodes that trigger spatial logic deviations. For abnormal nodes in an isolated state, the diagnostic verification module 20 constructs a virtual force model containing the resultant force of spring attraction and spatial anti-overlap repulsive force, and uses a simulated annealing algorithm for iterative solution, outputting physically feasible topology recommendation anchor points.

[0034] The fine-tuning and publishing module 30 is used to draw abnormal device instances and their corresponding hidden magnetic adsorption ranges in the 3D rendering viewport. As a preferred method, this module captures the dynamic drag coordinates input by the user in real time, blocks the input flow and performs rigid adsorption when spatial intersection conditions are met, forcibly locking the device coordinates to the topology-recommended anchor point to complete interactive correction. After the spatial poses of all scene entities meet the validity verification, the fine-tuning and publishing module 30 generates a topology association loading weight matrix based on the shortest logical connectivity hop count between nodes, and finally serializes the updated coordinate data, mapping relationships, and weight matrix into the data carrier and broadcasts it to the bus.

[0035] The runtime scheduling module 40 parses the published data carrier at the runtime end and generates basic loading priority weights in real time based on the camera's physical line-of-sight and spatial occlusion status. When a device alarm event is detected, this module extracts the topological association loading weights of the corresponding alarm node from the data carrier and, combined with a dynamic weight mixing factor driven by business status, calculates the comprehensive scheduling priority score for each node in the scene. Based on this calculation, the runtime scheduling module 40 reorders the differential update queue of video memory resources in descending order, forces cross-line-of-sight resource dimensionality reduction intervention for associated objects whose scores meet the threshold, and directly extracts their high-precision 3D models for priority assembly and rendering.

[0036] Data carrier 50 connects to various modules, acting as a distributed data node across the entire data chain. Data carrier 50 stores 3D spatial physical attribute data, business logic topology data, model asset binding relationships, measurement point mapping relationships, topology association loading weight matrix, and underlying rendering resource scheduling parameters. Specifically, data carrier 50 includes a high-speed memory cache component for temporarily storing frequently read / written real-time measurement point parameters, dynamic coordinates during interactive drag-and-drop processes, the video memory differential update queue during runtime, and transient scheduling weights when alarms are triggered, thereby ensuring low-latency data interaction between various computational modules. The graph database component is used to structure and persistently store the set of device instance nodes and process connection edges in the scenario business logic topology graph, as well as the tree-like mounting relationship between devices and 3D model levels, thereby supporting graph structured queries and efficient traversal such as breadth-first search algorithm and shortest logical connection hop count calculation. Persistent documents and relational storage components are used to systematically store device ledger details, standardized template rules, basic distance rule tables, cross-region compensation matrices, and serialized topology association loading weight matrices, serving as the underlying persistent foundation for the entire system's data recovery, version tracing, and asset comparison. See attached document Figure 2 , Figure 2 This is a flowchart illustrating a visual rapid deployment method for an intelligent operation and maintenance scenario provided by an embodiment of the present invention. The present invention provides a visual rapid deployment method for an intelligent operation and maintenance scenario, which includes the following steps: S10, the data access module 10 connects to the external business system of the industrial site and acquires multi-source heterogeneous data. According to the standardized template of equipment attributes and behaviors, it parses and maps the multi-source heterogeneous data, generates a standardized set of equipment instance objects, and establishes the model binding relationship between the standardized equipment instance objects and the 3D model assets. It parses the process connectivity relationship in the data of the external business system, extracts node and edge data to construct the scene business logic topology diagram, and writes the standardized set of equipment instance objects, the model binding relationship and the scene business logic topology diagram into the data carrier 50. S20, the diagnostic verification module 20 reads the standardized equipment instance object set from the data carrier 50, performs data mapping between the equipment instance objects and the measurement point parameters, infers the three-dimensional initial physical coordinates of the equipment instance objects, and extracts the corresponding spatial bounding box scale and logical space region attributes. S30, the diagnostic verification module 20 calculates the adaptive physical allowable distance threshold and the actual physical Euclidean distance between nodes based on the initial 3D physical coordinates, the spatial bounding box scale, and the connectivity in the scene business logic topology diagram; compares the adaptive physical allowable distance threshold and the actual physical Euclidean distance to filter out abnormal device instance objects that trigger spatial logic deviation anomalies; moves the abnormal device instance objects into the abnormal isolation area, extracts stable adjacent nodes to construct a virtual force model, applies the simulated annealing algorithm to calculate convergence coordinates, and outputs topology recommended anchor points; S40, the fine-tuning release module 30 draws the abnormal device instance object and its corresponding hidden magnetic adsorption range in the visualization editing auxiliary layer of the 3D rendering viewport; it receives the drag coordinates input from the 3D rendering viewport, and when the drag coordinates enter the capture radius condition corresponding to the magnetic adsorption range, it executes the rigid adsorption command to update the 3D initial physical coordinates of the abnormal device instance object; after completing the coordinate correction of the abnormal object and passing the legality review, it generates a topology association loading weight matrix based on the shortest logical connection hop count of each device instance object in the scene business logic topology diagram, writes the updated coordinate data, mapping relationship and topology association loading weight matrix into the data carrier 50 and executes the release action; S50, the operation scheduling module 40 receives and parses the published data carrier 50. During normal operation, it calculates the basic loading priority weight based on the camera's line of sight and occlusion culling algorithm, and executes basic 3D rendering scheduling. It listens for device alarm events in the bus, intercepts the alarm events, and reads the topology-related loading weight matrix in the data carrier 50. It combines the basic loading priority weight and the topology-related loading weight matrix to calculate the comprehensive scheduling priority score of each device instance in the scene. Based on the comprehensive scheduling priority score, it reconstructs the differential update queue, adjusts the network resource download order, and prioritizes the extraction of high-precision 3D model resources for device instance objects whose scores meet the threshold, and performs downloading, decoding, and video memory pre-assembly. When the device instance objects meet the visibility conditions, it performs visible rendering instantiation.

[0037] To further clarify the implementation of each technical aspect of the present invention, the following will provide a detailed description of the implementation of each functional module involved above and its internal processing flow.

[0038] See attached document Figure 3 After acquiring the underlying data from the industrial site, the data access module 10 needs to organize the multi-source, heterogeneous, discrete data into structured data that can be uniformly scheduled at each level of the system, and establish a mathematical correlation model of physical entities in business processes. This provides a basic rule input and reference system for subsequent spatial logic cross-diagnosis. Based on this processing logic, step S10 may include the following sub-steps: S101, the data access module 10 connects to an external business system via an industrial communication gateway. In this embodiment, the aforementioned external business system may include an enterprise resource planning system, a data acquisition and monitoring control system, and a manufacturing execution system. This data may include equipment asset ledgers, model specifications, and full lifecycle maintenance records from the enterprise resource planning system; real-time sensor telemetry values, dynamic operating conditions, equipment start-up and shutdown status, and underlying alarm event data from the data acquisition and monitoring control system; and process flow directions, production scheduling indicators, and pipeline / cable connectivity logic data from the manufacturing execution system. These multi-source data collectively provide a comprehensive data foundation for the construction of intelligent operation and maintenance scenarios. Subsequently, the system acquires multi-source heterogeneous data including structured data tables, real-time location lists, and various equipment ledger files, and reads a pre-set standardized template for equipment attributes and behaviors. As a preferred approach, this standardized template for equipment attributes and behaviors is configured with a unified equipment type coding dictionary and attribute field mapping rules. The data access module 10 performs field alignment and data cleaning on multi-source heterogeneous data according to the attribute field mapping rules. Specifically, when it is found that the unit of measurement in the input data is inconsistent with the template dictionary, the system will retrieve the conversion coefficient to adjust the value. For missing fields of key attributes, the system default value will be automatically extracted for safe filling.

[0039] Through the above processing operations, the device naming, units, and data types that differ across different business systems are uniformly converted into a standard format, and a standardized device instance object corresponding to each device entity is instantiated based on the standard format. Simultaneously, the data access module 10 matches the corresponding model asset identifier from the 3D model asset library based on the device type code, device specifications, or external asset identifier, and writes the model asset identifier into the model binding field of the standardized device instance object; if no corresponding model asset is matched, a preset general model asset identifier is written to ensure that the subsequent spatial feature extraction process has an executable data entry point.

[0040] To facilitate global addressing and scheduling of the system, the data access module 10 integrates all standardized device instance objects and generates a set of standardized device instance objects. A collection of standardized equipment instance objects. Represented as ,in Representing the A specific standardized equipment instance object, The total number of device instance objects in the system, which is an integer greater than or equal to 1.

[0041] After instantiating isolated devices, the system needs to further obtain the process flow relationships of each device in actual industrial production in order to support the root cause tracing and impact analysis of faults in intelligent operation and maintenance scenarios.

[0042] S102, the data access module 10 parses the process connectivity relationships in the data from the external business system. Typically, these process connectivity relationships originate from industrial piping and instrumentation flow diagrams or electrical single-line diagrams output by the external business system. Based on the extracted process connectivity relationships, the data access module 10 extracts node and edge data and constructs a scenario business logic topology diagram in memory. In the scenario business logic topology diagram In this context, the node set is the collection of standardized device instance objects that have been constructed. Edge set Represented as This represents the business connectivity status between different node objects. In the specific mapping process, when determining standardized device instance objects... and When there are direct physical connections or control links between entities in the on-site process flow, such as after matching the start node identifier and end node identifier in the instrumentation flow diagram association table with the corresponding entities, the data access module 10 generates the corresponding connected edges in the diagram. ,and Meanwhile, the data access module 10 generates each connected edge The corresponding logical attribute data is attached, including the process medium type, process medium flow direction characteristics, process zone codes of the two end nodes, and the basic connectivity theoretical allowable distance index corresponding to the process medium type. Regarding the specific storage method and conventional traversal operations of the graph data structure in computer memory using adjacency matrices or adjacency lists, those skilled in the art can implement it using conventional programming methods according to the project architecture requirements. The specific algorithms and code construction are well-known technologies in this field and will not be elaborated upon here.

[0043] At this point, the physical entities and business relationships at the industrial site have been completely transformed into the system's internal mathematical expression.

[0044] S103, Data access module 10 completes the standardized device instance object set With scene business logic topology diagram After the construction and calculation, a data serialization operation is performed. The data access module 10 converts the structured object feature data, model asset binding index, process connectivity edge attributes, basic connection distance index, and graph topology relationship data into a unified, persistent storage format, such as a conventional hierarchical markup language file. The data access module 10 initializes and writes the serialized full data into the data carrier 50. The data carrier 50, as the core data hub, maintains this basic data framework in system memory or a cloud-based distributed storage cluster, thereby providing a basic model reference and data write-back entry point for subsequent functional modules and upper-level intelligent operation and maintenance applications.

[0045] See attached document Figure 4 To achieve rapid deployment of visualized intelligent operation and maintenance scenarios, the diagnostic verification module 20 needs to link data instances from the business dimension with renderings from the physical space dimension, and determine the baseline spatial characteristics of each entity in the 3D scene, thereby providing a computational basis for subsequent spatial logic cross-diagnosis. Therefore, step S20 may include the following sub-steps: S201, the diagnostic verification module 20 reads a standardized set of equipment instance objects from the data carrier 50, and performs data mapping pre-binding between the equipment instance objects and external measurement point parameters. In this embodiment, the diagnostic verification module 20 introduces a three-level mapping mechanism consisting of rule matching, semantic matching, and strong data type validation. The system compares the equipment tag number of the equipment instance object with the tag code in the measurement point list using regular expressions to extract initial screening matching pairs with basic association attributes. Considering that naming conventions are often inconsistent in industrial settings, the system performs semantic matching calculations based on string similarity algorithms on the Chinese description text of the initial screening matching pairs.

[0046] As a preferred method, the system compares the calculated string similarity score with a preset similarity threshold. If the score is lower than the similarity threshold, the current initial screening match is determined to be an interference mapping item with similar names but substantial semantic differences and is removed. After completing semantic filtering, the system reads the data protocol type returned by the measurement point channel and performs a strong verification comparison with the internal attribute types supported by the device instance object, filtering out invalid associations such as incorrectly binding a Boolean switch quantity to a floating-point temperature attribute, thereby ensuring the accuracy and reliability of real-time data presentation during subsequent intelligent operation and maintenance monitoring. Through the above three-level filtering mechanism, the diagnostic verification module 20 establishes a data mapping relationship between the device instance object and the real-time measurement point parameters it needs to carry. At the same time, the diagnostic verification module 20 generates the corresponding data mapping confidence score based on the rule matching result, semantic matching score, and data type strong verification result, and writes the data mapping confidence score and data mapping relationship into the data carrier 50; among them, the higher the rule matching hit, the higher the semantic matching score, and the higher the data type strong verification, the higher the corresponding data mapping confidence score. For the specific program implementation of string similarity calculation, those skilled in the art can use conventional dynamic programming algorithms or edit distance algorithms. The code construction is a well-known technology in this field and will not be described in detail here.

[0047] S202, after establishing the data mapping relationship, the diagnostic verification module 20 infers the initial three-dimensional physical coordinates of the equipment instance object. The system parses the regional hierarchy description and text clues attached to the standardized equipment instance object. The system has a preset spatial reference table, which records the reference three-dimensional origin coordinates corresponding to the factory area, workshop, floor, and column grid number. After parsing the regional hierarchy label of the equipment, the system queries the spatial reference table to obtain the basic origin coordinates of its belonging hierarchy. For the parsing and processing of text clues, the system uses regular expression matching rules to extract the relative offset values ​​of the attached direction description and distance unit, and converts them into numerical increments on the corresponding coordinate axes according to the direction description. The system performs a three-dimensional vector addition operation on the basic origin coordinates and the extracted numerical increments to infer the initial three-dimensional physical coordinates of the equipment instance object. .in , , These represent the three orthogonal axis components of the device instance object in the three-dimensional world coordinate system. If the extracted text clues do not contain explicit relative offset values, the system will default to assigning the geometric center coordinates of the spatial reference area to the inferred three-dimensional initial physical coordinates to avoid logical dead zones where the initial coordinate inference process results in empty calculations.

[0048] S203, to reduce the computational overhead of subsequent physical distance diagnostic algorithms when dealing with complex and irregular models, the diagnostic verification module 20 parses the 3D model bound to the device instance object and extracts the corresponding spatial bounding box scale and logical space region attributes. Introducing the diagonal length of the bounding box as a feature scale simplifies the complex mesh collision problem to region overlap determination based on extreme radius. Specifically, the system loads the set of mesh vertices corresponding to the 3D model, traverses the spatial coordinates of all vertices in the set, and first transforms the vertex coordinates in the model's local coordinate system to a unified 3D world coordinate system or a unified scale coordinate system based on the model's scaling ratio, rotation attitude, and transformation matrix of the mounted nodes. It then obtains the extreme value data along each spatial axis and constructs a 3D axial bounding box that encloses the model based on this extreme value data. The system calculates the spatial bounding box scale based on the constructed 3D axial bounding box. The specific calculation process uses the following formula: ; In the formula, Indicates the first The diagonal length of the bounding box of a device instance object; and These represent the maximum and minimum coordinate values ​​of the mesh vertex set along the X-axis in the three-dimensional coordinate system, respectively. and These represent the maximum and minimum coordinate values ​​along the Y-axis, respectively. and These represent the maximum and minimum coordinate values ​​along the Z-axis, respectively. When an anomaly occurs, such as a 3D model asset loading failure or an empty mesh vertex set, the system assigns a preset default safety scale value to the device instance object, ensuring the diagnostic process is not interrupted by a single asset defect. The system extracts the corresponding logical spatial region attributes based on the hierarchical node classification of the 3D model asset in the scene tree, combined with the regional hierarchy description attached to the standardized device instance object and the spatial reference region where the initial 3D physical coordinates are located. When conflicts arise from the aforementioned sources, the system prioritizes the logical spatial region attributes recorded in the equipment ledger or regional hierarchy description, and writes the conflict status into the data carrier 50 for subsequent verification. The diagnostic verification module 20 uses the inferred initial three-dimensional physical coordinates, the calculated spatial bounding box scale, and the logical spatial region attributes as initial spatial feature parameters, updates and writes them into the data carrier 50.

[0049] See attached document Figure 5 After acquiring the initial spatial characteristics of each device, the diagnostic verification module 20 needs to establish a cross-verification system between physical distance and logical topology. When an unreasonable spatial location is detected, it automatically calculates a physically feasible corrected reference location to prevent spatial interference and logical conflicts in actual intelligent operation and maintenance inspections and equipment repair planning. In actual industrial 3D scenarios, the size differences between different device entities are huge, and using a static, uniform distance threshold can easily lead to false alarms. Furthermore, when constraint conflicts occur, conventional force-directed layout algorithms are prone to getting stuck in infinite loops in extremely dense areas. Based on the above processing requirements, step S30 may include the following sub-steps: S301, the diagnostic verification module 20 calculates the adaptive physical distance threshold between nodes based on the initial 3D physical coordinates, the spatial bounding box scale, and the connectivity relationships in the scene business logic topology graph. In this embodiment, the system traverses all connected edges in the scene business logic topology graph. For nodes and nodes For each connected link, the system retrieves the previously written connected edge attributes and basic connection distance index from the data carrier 50, or queries the theoretically permissible basic connection distance from the preset basic distance rule table according to the process medium type. This fundamental connection theory allows for the pre-calibration of distances based on the statistical average of standard pipeline laying lengths from past engineering drawings.

[0050] In one embodiment, the establishment of the basic distance rule table includes: acquiring historical engineering drawing data and actual pipeline survey data from the industrial site; extracting the actual pipeline laying lengths between equipment for different process media types (such as water, gas, chemical liquids, electrical cables, etc.); classifying and summarizing the pipeline laying lengths under the same process media type, and calculating the statistical mean of the standard pipeline laying lengths corresponding to each process media after removing extreme outliers using statistical methods; finally, using the process media type as the retrieval index and the corresponding statistical mean of the standard pipeline laying length as the basis, connecting the theoretical allowable distances to construct a structured data mapping table, which can be preset through the data dictionary configuration interface or offline configuration file import module provided by the system. Specifically, during the initial deployment phase of the system, implementers can import the pre-built basic distance rule table in a standard data format and persistently store it in the data carrier for initial preset; at the same time, during the subsequent operation phase of the system, maintenance personnel can also dynamically fine-tune and update the various basic theoretical distance parameters in the rule table through a visual backend management interface according to the actual modification of the on-site pipeline architecture or changes in process specifications and standards.

[0051] To mitigate the implicit impact of equipment volume differences on pipeline laying distance, the system uses the extracted spatial bounding box dimensions as a compensation parameter to calculate a dynamically adaptive allowable distance threshold. The specific calculation process uses the following formula: ; In the formula, Represents a node With nodes The dynamically generated adaptive maximum allowable physical distance between them. This indicates the distance allowed by the fundamental connectivity theory. This represents the model scale compensation coefficient, which is used to proportionally relax the distance constraint based on the physical volume of the equipment itself. Its value is usually set to a real number between 0.5 and 1.5. and Representing nodes respectively With nodes The diagonal length of the spatial bounding box. This represents the cross-regional tolerance compensation value, which the system uses to determine the node. Logical space region attributes With nodes Logical space region attributes If they are the same, and both belong to the same physical region, then assign a value. If the system crosses different physical areas, it will consult the preset cross-area compensation matrix and assign the corresponding tolerance increment value. The cross-area compensation matrix records the shortest mapping distance between the three-dimensional boundaries of different logical space areas to compensate for the objective physical span caused by crossing floors or factory buildings.

[0052] In one embodiment, the specific process of establishing the cross-region compensation matrix is ​​as follows: the system pre-extracts the 3D boundary features of all independent logical spatial regions in the scene. Specifically, the system traverses the 3D model asset library or scene space tree to obtain the 3D bounding boxes corresponding to each region. For the total number of regions in the scene... Each is an independent logical space region, and the system initializes a dimension in memory. A symmetric matrix. For the off-diagonal elements of this matrix, i.e., any two distinct physical regions... and The system extracts the outer surface data of the bounding boxes of both regions and uses a nearest-point retrieval algorithm in computational geometry to find the two boundary anchor points on the surfaces of the two bounding boxes that are spatially closest. The system calculates the three-dimensional Euclidean distance between these two boundary anchor points and superimposes a preset building barrier margin parameter. This parameter is mainly used to compensate for the increase in pipeline detour length caused by the objectively existing standard wall thickness or floor slab perforation, thereby deriving the shortest survey distance between the two regions and establishing it as the tolerance increment value. The system fills all the obtained tolerance increment values ​​into the corresponding row and column coordinate positions of the matrix, while fixing the diagonal elements of the matrix (representing compensation within the same region) to zero. Through the above global traversal and calculation operations, the system generates a complete cross-region compensation matrix and persistently stores it in the data carrier 50, thereby supporting the fast threshold call based on row and column indexes in the subsequent distance verification process.

[0053] S302, after obtaining the dynamic threshold, the system executes the cross-constraint deviation judgment logic. The diagnostic verification module 20 calculates the actual physical Euclidean distance between the two associated nodes. The calculation formula is: ; In the formula, , , For nodes The three-dimensional coordinate components, , , For nodes The system defines the three-dimensional coordinate components. In addition to physical distance verification, the system establishes a logical verification mechanism for regional attributes and defines a judgment function. The region compatibility rule table corresponding to this decision function is pre-stored in the data carrier 50. The region compatibility rule table uses logical space region attributes as row and column indices to record whether direct connectivity between different regions is allowed. and The function outputs 1 when direct pipeline connections are permitted under industrial regulations, and outputs 0 when physical isolation requirements or mutual exclusion restrictions exist. For example, when a direct pipeline connection between an explosion-proof isolation zone and a conventional hot work zone is detected, the function determines it as mutually exclusive and outputs 0. The system uses bidirectional comparison to filter out abnormal device instances that trigger spatial logic deviation anomalies. The specific anomaly determination boundary conditions are expressed as follows: ; When the above logical expression evaluates to true, the system first connects the edges. Mark as a spatial logical divergence edge, and set the node and The node is added to the candidate anomaly set. Then, based on the node's data mapping confidence, whether its regional attributes are missing, the number of edges it participates in, and its distance consistency with other stable neighboring nodes, the system determines the abnormal device instances that need correction. Stable neighboring nodes are those that have direct or indirect business connections with the current candidate anomaly node, are not added to the candidate anomaly set, or whose participating connected edges have not triggered spatial logic deviation conditions after verification. If a single-sided anomaly cannot be distinguished, both ends of the node are retained as anomaly candidates and proceed to the subsequent isolation solution process. As a preferred approach, if a node is isolated in the topology graph without any connected edges, the system will skip this constraint deviation calculation logic to avoid wasting computing power and creating logical dead zones.

[0054] S303, for objects determined to be abnormal, the system needs to suspend them and calculate corrected coordinates. The diagnostic verification module 20 intercepts the selected abnormal device instance objects and moves them into the abnormal isolation zone, preventing them from entering the regular rendering queue of the formal release scene, but retaining their display permissions in the visual editing auxiliary layer so that the subsequent fine-tuning release module 30 can perform abnormal highlighting, dragging, and snapping corrections. The system establishes a spatial recovery mechanism for abnormal device instance objects. The system extracts abnormal nodes within the isolation zone. Furthermore, it traces all stable adjacent nodes in the scenario's business logic topology graph and forms a set of adjacent nodes that have not experienced any anomalies. The system is based on a set The node coordinates in the data are fixed anchor points, indicating abnormal nodes. A virtual force model is constructed, incorporating the resultant force of spring gravity and the spatial anti-overlap repulsive force. The system calculates the resultant force vector of spring gravity based on Hooke's law and its deformation principle. When calculating the distance between an abnormal node and its stable neighboring nodes, if the coordinate distance between them is less than a preset lower limit, this preset lower limit is used instead of the actual distance in the calculation, and a normalized direction vector pointing from the abnormal node to the stable neighboring node is adopted. If the direction vector cannot be determined, a random three-dimensional unit vector is generated as the initial direction to avoid numerical singularities caused by division by zero or uncertain direction. Simultaneously, to prevent model penetration collisions between the correction coordinates and other already positioned equipment, the system introduces an anti-overlap repulsion vector. .

[0055] Specifically, the system iterates through the bounding box of the abnormal node to search for other entity nodes within its radius. When it determines that the bounding boxes of two nodes overlap or the distance between them is less than a preset safety margin, it generates a repulsive force vector in the opposite direction of the line connecting the centers of the two nodes. If it determines that the center coordinates of the two nodes completely coincide, the system generates a random three-dimensional unit vector as the direction of the repulsive force to break the coordinate symmetry deadlock. Multiple repulsive force vectors are superimposed to form... The sum of the two is the result of the first two. The total force vector at the next iteration : ; In the formula, This represents the preset spring stiffness coefficient, which controls the gravitational convergence strength between abnormal nodes and their stable adjacent nodes. To avoid severe oscillations in node positions or even model overshooting due to excessive gravitational feedback in the early stages of iterative calculations, the value of this coefficient is usually normalized based on the maximum spatial span of the scene's 3D bounding box. In this embodiment, its value is set to a positive real number between 0.1 and 0.5 to balance the proportional relationship between gravity and anti-overlap repulsion force, ensuring the smoothness of the virtual force model and the stability of its final spatial distribution during multiple iterations. Indicates stable adjacent nodes The three-dimensional coordinate vector in space. Indicates the first The three-dimensional coordinate vector of the abnormal node during the next iteration. This represents the ideal physical distance for arrangement. As a preferred method, this value is directly taken from the allowable distance in the aforementioned basic connection theory. If the system finds a set during tracing... If the value is empty, it means that all surrounding neighboring points are in an abnormal state. The system will extract the coordinates of the geometric center point of the region corresponding to the node as the input of the virtual unique gravity source to ensure that the vector has a calculation basis under extreme topology.

[0056] S304, based on virtual force, the diagnostic verification module 20 applies simulated annealing algorithm to calculate convergence coordinates and outputs topology recommendation anchor points. Pure force-guided algorithms are prone to local oscillations in complex node networks due to shifts in force equilibrium points. The system introduces a method that adjusts the algorithm based on the number of iterations. Decreasing system temperature function The step size is updated to control the coordinates. The formula for calculating the temperature function is: ; In the formula, The initial scalar temperature of the system is set, which determines the size of the search range in the initial stage. It is usually proportional to the maximum diagonal span of the 3D bounding box of the current industrial scene. This represents the cooling attenuation coefficient, which controls the rate of cooling convergence in the system. Its value is set between 0.90 and 0.99 to ensure a smooth transition in the annealing process. The base is a natural number. The system defines the iterative coordinate update equation for detached objects: ; In the formula, Indicates the first The three-dimensional coordinate vector of the abnormal node at the next iteration; As the number of iterations increases, the magnitude of position updates gradually decreases due to the suppression of the temperature function. The system has a set freezing temperature threshold. With minimum displacement threshold ,in This is a positive real constant approaching zero. The value of this constant primarily depends on the accuracy of the underlying spatial coordinates of the 3D rendering engine and the visual tolerance requirements for the interactive layout of industrial site models. In this embodiment, considering that large industrial 3D scenes typically use meters as the standard coordinate system unit, and in order to balance the convergence efficiency of the simulated annealing algorithm with the precision of the final anti-overlapping equipment layout, The value range is usually set to a specific value between 0.001 and 0.01 (i.e., representing 1 millimeter to 1 centimeter in actual physical space). When the coordinate displacement of the device node is lower than this range for two consecutive iterations, it means that the virtual force model has tended to a microscopic dynamic equilibrium state. Continuing to occupy computing power for iteration has little significance for correcting the macroscopic visual layout. Therefore, the system uses this value limit as a reasonable exit cutoff condition.

[0057] After each iteration, the system determines whether the condition is met. Or the magnitude of the difference between the position coordinate vectors of two consecutive iterations is less than .

[0058] When any stopping condition is met, the system forcibly terminates the iteration process, obtains the final 3D physical coordinates that meet the stopping conditions, and performs anti-overlap detection on these final 3D physical coordinates. If these coordinates still overlap with other entities, the coordinates from the most recent iteration that did not overlap are used as the final 3D physical coordinates. If there are no historical iteration coordinates that did not overlap, a gridded candidate point search is performed within a preset search radius and at a safe distance, using the geometric center point of the logical space region to which the abnormal node belongs as the reference. The candidate point closest to the topology recommendation direction and without bounding box overlap is used as the final 3D physical coordinate. The system defines this coordinate as the topology recommendation anchor point. The serialized information, along with the identification information of the abnormal device, is written into the data carrier 50.

[0059] See attached document Figure 6 In this embodiment, to balance the user's subjective scene layout intentions with the correctness of the system's underlying spatial logic, and to empower operation and maintenance personnel to quickly build a 3D monitoring interface that conforms to process specifications with minimal interaction, the system guides users to complete visual fine-tuning by constructing a hidden magnetic adsorption space. After confirming the final position, the system pre-calculates the topology association weight, thereby providing a preliminary strategy basis for resource scheduling at the runtime. Based on the above interaction and calculation mechanism, step S40 may include the following sub-steps: S401, the fine-tuning release module 30 draws the abnormal device instance object and its corresponding hidden magnetic adsorption range in the 3D rendering viewport. As a preferred method, the fine-tuning release module 30 extracts the 3D bounding box data of the abnormal device instance object and replaces its surface rendering material with a preset warning material to achieve highlighted display of the abnormal node. The system draws 3D Bézier curves between the abnormal device instance object and its stable adjacent nodes, forming topological anomaly connections. The topological recommended anchor points calculated previously are extracted. Then, the system uses the anchor point as the geometric center and a preset capture radius. Using the radius as the trigger region, an invisible spherical collider is constructed in the interactive collision detection layer of the 3D engine or the GPU pickup detection cache. (Regarding the capture radius...) The value of is dynamically calculated in this embodiment based on the distance from the camera to the anchor point in the current viewport. Specifically, it can be calculated according to... Confirmed, among which Recommend the distance of the anchor point from the camera to the topology. This is the interaction scaling factor, a dimensionless parameter, with a value ranging from 0.005 to 0.05. and These are the minimum and maximum capture radii, respectively, to ensure a consistent interactive experience for users across different scaling levels. This spherical trigger area does not participate in color rendering or final image output in the regular 3D rendering viewport; it serves only as an invisible interaction detection object for underlying spatial intersection detection.

[0060] S402, the fine-tuning publishing module 30 receives the drag coordinates input from the 3D rendering viewport. When it detects that the user is dragging an abnormal device instance object through the input device, the system uses camera ray detection technology to convert the screen's 2D coordinates into dynamic drag coordinates in the 3D scene in real time. The system continuously determines the dynamic drag coordinates. Topology recommended anchor points The Euclidean distance between them. When this distance enters the hidden magnetic adsorption range, the condition is met. At that time, the fine-tuning release module 30 executes the rigid adsorption command. Upon receiving this command, the system blocks the dynamic drag coordinate data stream of the input device, updates and locks the three-dimensional initial physical coordinates of the abnormal device instance object as the topology recommended anchor point. The coordinate parameters are then used to complete the interactive correction of the coordinates. After locking is completed, the system re-executes the same distance threshold verification, area compatibility verification, and bounding box overlap verification as in step S302; if the verification fails, the current locking state is released, and the abnormal device instance object is returned to the abnormal isolation area. The fine-tuning release module 30 sends a re-correction request to the diagnostic verification module 20, or calls the anchor point recalculation interface of the diagnostic verification module 20 to regenerate the topology recommended anchor points.

[0061] S403: After correcting the coordinates of the abnormal object and passing distance threshold verification, region compatibility verification, and bounding box overlap verification, the spatial position of the scene entity meets the logical verification standard, and the system enters the pre-calculation stage of the rendering scheduling strategy. The fine-tuning and release module 30 generates a topology association loading weight matrix based on the shortest logical connectivity hop count of each device instance object in the scene business logic topology diagram.

[0062] The specific calculation process is as follows: The fine-tuning release module traverses all nodes in the scenario business logic topology graph 30 times, and for any two nodes... and The breadth-first search algorithm is used to calculate the shortest logically connected hop count between the two. The hop count reflects the tightness of coupling between two devices in an industrial business process. The system combines the hop count with the attenuation constant to calculate the topology association loading weight matrix. Elements in the matrix The calculation formula is as follows: ; In the formula, Represents a node With nodes The topological association weight score between them, with a value between 0 and 1; This represents the business attenuation constant, used to control the degree to which the weight decreases as the number of hops increases. Its value is set according to the association depth requirements of the industrial scenario, and it usually ranges from 0.1 to 0.5. This represents the shortest logically connected hop count calculated using breadth-first search. If the node... With nodes If a subgraph is an isolated subgraph that is not connected in the topological graph, then its hop count is... Defined as infinity, the corresponding weight score The value is assigned to 0.

[0063] For the specific code implementation of the breadth-first search algorithm traversing the graph data structure and finding the shortest path in the unweighted graph, those skilled in the art can use a conventional queue data structure to program it. Its search logic is a well-known technology in this field and will not be elaborated here.

[0064] S404, after obtaining the three-dimensional spatial pose and topology scheduling strategy through the above steps, the fine-tuning release module 30 determines that all abnormal device instance objects have completed coordinate correction and passed the legality review; when the judgment result is passed, the fine-tuning release module 30 writes the updated coordinate data, mapping relationship and topology association loading weight matrix into the data carrier 50 and executes the release action; when the judgment result is failed, the fine-tuning release module 30 keeps the current scene in the visual editing state, prohibits the execution of the release action, and outputs the abnormal device instance object that failed the legality review and the corresponding reason for failure.

[0065] Specifically, the fine-tuning release module 30 integrates and packages the above information, serializes it into a unified format data stream, and writes it into the corresponding memory block of the data carrier 50. The system generates a globally unique timestamp version number and broadcasts a completion event signal to the system bus to trigger the listening action of the running end, thereby maintaining multi-terminal data synchronization in the network environment.

[0066] See attached document Figure 7Limited by the video memory capacity of computer graphics processors, conventional 3D scene rendering engines typically rely strictly on camera view distance and spatial occlusion when processing massive models to prioritize resource loading. However, in industrial monitoring and maintenance scenarios, once an equipment alarm occurs, personnel often need to quickly check the status of other nodes closely related to the equipment in the process topology for rapid fault diagnosis, impact assessment, and emergency isolation and troubleshooting. In this case, traditional rendering scheduling mechanisms based on physical view distance can easily lead to delayed loading of remote related equipment models, affecting maintenance response efficiency. To address these issues, this embodiment introduces a cross-view distance resource priority scheduling mechanism that combines logical topology weights and physical view distance weights. Therefore, step S50 may include the following sub-steps: S501, the runtime scheduling module 40 parses the published data carrier 50 and generates basic loading priority weights based on the camera's view distance and occlusion status. As the basic framework for regular scene rendering, the runtime scheduling module 40 obtains the spatial coordinates of the virtual camera in the 3D rendering viewport in real time and calculates the physical view distance between the virtual camera and the 3D coordinates of each device instance object in the scene. The system uses view frustum culling and occlusion culling algorithms to determine whether a device instance is visible in the current frame. For device instances that are visible, the system calculates their basic load priority weight. The calculation formula is as follows: ; In the formula, Indicates the first The basic loading priority weight of each device instance object, with a value ranging from 0 to 1; This indicates the physical line-of-sight distance between the virtual camera and the device instance object; This represents the system's preset global maximum visible viewing distance threshold, which is synchronously set based on the virtual camera's far clipping plane distance parameter. If the calculated physical viewing distance... Greater than or equal to If the device instance is determined to be invisible by the occlusion culling algorithm, then the value is directly assigned. For the specific spatial geometric intersection calculations of frustum culling and depth buffer occlusion culling, those skilled in the art can use conventional graphics collision box detection techniques. The processing logic is well-known in the field and will not be elaborated here.

[0067] In S502, the status of the industrial site is dynamically changing. The system monitors the equipment status data reported by the underlying industrial control network in real time via the event bus. When the operation scheduling module 40 intercepts a device alarm event, it parses the event message to extract the identifier of the source device that triggered the alarm. Based on the source device identifier, the system performs a reverse matching against the standardized set of device instance objects in the data carrier 50 to locate the corresponding alarm node. .

[0068] The execution scheduling module 40 reads the topology association loading weight matrix generated in the previous steps from the data carrier 50. And extract the corresponding alarm nodes from the matrix. The row vector containing the elements of that row vector. Represents all other nodes in the scene With the current alarm node The logical topological correlation between them. If alarm events from multiple different devices are intercepted within a very short time window, the system will extract the corresponding row vectors and analyze them for each node. Extract the maximum value from the corresponding column as the weight for loading the merged topology association to prevent weight conflicts in the case of multiple concurrent alarms.

[0069] In S503, to achieve a balance between spatial physical line-of-sight and business-related logic, the runtime scheduling module 40 introduces a dynamic weighting hybrid factor to calculate the comprehensive scheduling priority score of each device instance object in the scenario. As a preferred method, the system linearly superimposes the basic loading priority weight and the topology-related loading weight to obtain the comprehensive scheduling priority score. The specific mathematical formula is as follows: ; In the formula, Indicates the first The overall scheduling priority score of each device instance object; This represents a dynamic weighting blending factor, used to adjust the dominant ratio of physical spatial view distance and logical topology association in rendering scheduling; Represents a node With alarm nodes The topological association loading weights. In this embodiment, the dynamic weight mixing factor... The value is driven by the current business status of the system. In the normal inspection state without alarms, since no alarm source has been identified, the system loads weights based on the topology association of all nodes. All are assigned the value 0, and Forced to 1.0, this degrades the rendering scheduling strategy to rely entirely on the base load priority weight. Prioritizes maintaining view-distance-based rendering smoothness; when an alarm event is detected, the system will... The weight is dynamically reduced to 0.3, thereby amplifying the topology association loading weight. The weight of the overall score allows devices that are closely related to the alarm node logic to obtain higher scores.

[0070] S504, the scheduling module 40 runs the overall scheduling priority score calculated by the system. The resource differential update queue of the rendering engine is reordered in descending order. A mandatory high-precision resource loading threshold is set by the system. This threshold is dynamically calculated based on the available video memory capacity of the current graphics processor. Specifically, it is reduced when the available video memory ratio exceeds a preset upper limit. To expand the loading range of high-precision models, and improve performance when the available video memory ratio is below a preset lower limit. To limit the number of objects loaded for high-precision models, a default threshold is maintained when the available video memory ratio is in the middle range; this is typically configured as a real number between 0.6 and 0.8. This applies to the scores in the rearranged queue that meet the following criteria. For device instance objects, the scheduling module 40 executes cross-line-of-sight resource dimensionality reduction intervention actions. Based on this strategy, the system transcends conventional physical space line-of-sight limitations and hierarchical detail rendering rules, directly extracting the high-precision 3D model (such as LOD0 level mesh data) and high-resolution textures corresponding to the logically related object from the model asset library, and prioritizing their download, decoding, and assembly into the video memory cache. For objects that are still outside the view frustum or occluded in the current frame, the system performs background pre-assembly without forcibly destroying the view frustum culling and occlusion culling results, and performs visible rendering instantiation only after the object enters the visible state. For devices with a comprehensive scheduling priority score lower than the threshold, the system continues to use conventional low-precision models as placeholders or does not render them. Through the above intervention execution logic, the system ensures that critical devices on the alarm service link can be prioritized and pre-assembled despite physical distance under limited rendering computing power, and are quickly presented when the visibility conditions are met, thereby greatly shortening the waiting time for maintenance personnel to find related fault nodes in space, and making the emergency response experience of intelligent operation and maintenance scenarios more efficient.

[0071] To enable those skilled in the art to more clearly understand the purpose, technical solution, and advantages of this invention, the present invention will be further described in detail below with reference to specific application embodiments, real experimental test data, and corresponding drawings. It should be noted that the embodiments described herein are only for explaining the present invention and are not intended to limit the scope of protection of the present invention.

[0072] The following example demonstrates a specific industrial application scenario (a 3D visualization monitoring project for a large-scale petrochemical catalytic cracking unit), and provides corresponding experimental verification and effect comparison data.

[0073] I. Specific Application Examples In the specific application scenario of this embodiment, the target monitoring area is a petrochemical catalytic cracking unit containing more than 1,200 equipment nodes. The internal process pipelines of this unit are intricate and involve high-temperature and high-pressure hazardous areas.

[0074] During the system initialization phase (corresponding to steps S10 and S20), the data access module 10 reads the process piping and instrumentation diagrams and equipment ledgers of the external system. The system successfully parses out a direct process crude oil transport topology link between feed pump P-101 and heat exchanger E-102, with a hop count of 1. Simultaneously, by parsing the ledger text, the system extracts the initial coordinate label of feed pump P-101 as "Second-floor platform - East side" and maps it to three-dimensional initial physical coordinates.

[0075] Upon entering the spatial logic cross-constraint diagnosis stage (corresponding to step S30), due to an error by the personnel who entered the data into the ledger earlier, the area label of feed pump P-101 in the ledger was incorrectly recorded as the distant waste liquid treatment area. When calculating the adaptive physical allowable distance threshold, the diagnostic verification module 20 discovered that the actual Euclidean distance between the currently inferred physical coordinates of P-101 and the actual physical coordinates of the heat exchanger E-102 was as high as 150 meters, far exceeding the dynamic threshold (45 meters) after superimposing the bounding box scale and cross-region compensation. Simultaneously, the area compatibility rule table determined that a direct crude oil connection is not allowed between the waste liquid treatment area and the high-temperature heat exchange area. After triggering the deviation condition, the system immediately moved feed pump P-101 into the abnormal isolation area.

[0076] To correct this erroneous coordinate, the system uses the stable and error-free heat exchanger E-102 as the gravitational source to establish a virtual force model. Under the iterative simulation annealing algorithm, the coordinate point representing the location of P-101 is pulled towards E-102 in three-dimensional space by gravity, while simultaneously being pushed away by the anti-overlapping repulsive forces of the surrounding existing reactors. After dozens of temperature decay iterations, the algorithm converges at a spatial location that satisfies a safe distance and is approximately 15 meters from E-102, outputting the topology-recommended anchor point.

[0077] During the visualization fine-tuning and deployment phase (corresponding to step S40), the implementer sees the feed pump P-101 displayed in red warning light in the 3D rendering viewport. When the implementer drags it with the mouse to the vicinity of the topology recommended anchor point, the system's underlying dynamic calculation-based capture radius triggers rigid adsorption, and the device instantly and smoothly locks into the correct position. After the system verifies that everything is correct, it calculates the topology weights of all devices and executes the deployment.

[0078] Once the unit enters normal operation (corresponding to step S50), the virtual camera is focused on the plant gate (which is far from the catalytic cracking unit). At this time, the underlying industrial network reports an alarm event indicating that the feed pump P-101 shaft temperature is too high. After the system intercepts the alarm, the dynamic weighted mixing factor... The priority was instantly reduced from 1.0 to 0.3. Although heat exchanger E-102 was extremely far from the camera, and its basic loading priority weight was almost zero, it received a very high topology-related loading weight because its logical connection hop count with alarm source P-101 was only 1. The system ignored physical line-of-sight limitations, forcibly extracted the high-precision LOD0 model of E-102, and pre-assembled it into the video memory. When maintenance personnel quickly zoomed in on the process pipeline area following the alarm prompts, the high-precision model of the entire upstream and downstream business chain was already displayed, completely eliminating the white model occupation and texture blurring delay phenomena of the traditional line-of-sight loading mode.

[0079] II. Experimental Verification and Effect Comparison To verify the convergence stability of the system's automatic coordinate correction in real-world complex industrial scenarios, as well as its cross-line-of-sight 3D rendering scheduling performance under sudden alarm conditions, the applicant conducted specific performance tests using a test server (configured with an Intel Xeon Gold 6248R CPU, an NVIDIA RTX 4090 24GB graphics card, and 128GB of RAM). The test scenario uniformly used a large-scale petrochemical digital factory dataset containing 15,000 3D model asset nodes.

[0080] Comparison group settings: Experimental group: The system adopts a visualization-based rapid deployment system for intelligent operation and maintenance scenarios provided by this invention, enables an automatic correction mechanism for abnormal coordinates based on virtual forces, and activates a dimensionality reduction intervention and dynamic pre-assembly strategy driven by both logical topology and physical line of sight at the runtime end.

[0081] Control group: A traditional 3D visualization system was used, with automatic skew correction algorithms disabled (relying on purely manual layout), and a conventional rendering scheduling mechanism based on physical view distance and spatial occlusion culling was strictly adopted at the runtime.

[0082] The test results are as follows: According to the appendix Figure 8 As can be seen, in the spatial constraint logic deviation correction test, the three-dimensional spatial coordinate points (solid line trajectory) representing the abnormal equipment exhibit a smooth motion trajectory with effective obstacle avoidance capability under the drive of the virtual force model. Under the combined action of the traction of stable adjacent nodes (gravity sources) and the repulsion of obstacle nodes (repulsion sources), the equipment coordinates do not fall into physical dead ends or experience model penetration in the dense obstacle environment. Instead, they dynamically detour and eventually settle and converge to the topology recommended anchor point that meets the process safety distance limit.

[0083] According to the appendix Figure 9It can be seen that during the iterative process of solving the recommended coordinates based on the simulated annealing algorithm, the system temperature variable (solid black line, left Y-axis) exhibits a standard exponential smooth decay. Under this suppression, the oscillation amplitude of the three-dimensional coordinate displacement (dashed black line, right Y-axis) of the abnormal device in two consecutive iterations, after an initial rapid leap in the search, quickly converges and decreases. Within 60 iterations, the displacement difference successfully decreases and stabilizes below the preset minimum displacement threshold (dashed dot reference line), triggering a reasonable truncation of the algorithm. This proves that the algorithm of this invention overcomes the dead-loop oscillation defect that is easily generated in traditional pure force-guided layouts under complex topologies, and has both coordinate correction calculation efficiency and system convergence stability.

[0084] According to the appendix Figure 9 As shown in subgraph (a), in the resource scheduling performance test under sudden alarm conditions, when a device fault alarm is triggered at the 10th second, even though the fault source is 800 meters away from the current virtual camera, the experimental group (the solution of this invention) immediately reconstructs the differential update queue using a dynamic weighted mixing factor. Under the condition that the line-of-sight elimination rule is not originally loaded, the experimental group forces the pre-assembly of the high-precision model associated with the alarm link in the background, causing the memory readiness rate to quickly climb to over 95% within just 5 seconds. In contrast, the control group, due to the limitation of the pure physical line-of-sight rule, has its associated model readiness rate stagnant at a freezing point during this time window.

[0085] According to the appendix Figure 9 Subgraph (b) further shows that when the maintenance personnel instantly switched the view to the remote alarm area via a quick operation at the 15th second, the experimental group, benefiting from the pre-loading intervention of topology weights, exhibited extremely strong resilience in its 3D scene rendering frame rate (FPS), quickly stabilizing at a full-load smooth level of nearly 60 frames after only a brief minor adjustment. In contrast, the control group experienced I / O resource congestion and memory paging blockage at the instant of this sudden view change, resulting in a deep V-shaped drop in scene FPS (bottoming out at approximately 12 frames), producing extremely severe screen stuttering and model white-model placement delay. The significant difference between the two sets of data fully confirms that the topology-driven dimensionality reduction intervention and dynamic assembly strategy proposed in this invention can avoid view distance limitations under the constraints of massive industrial models.

[0086] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A visualized and rapid deployment method for intelligent operation and maintenance scenarios, characterized in that, Includes the following steps: Parse data from external business systems to generate a standardized set of device instance objects and a scenario business logic topology diagram, and write them into a data carrier; Infer the initial three-dimensional physical coordinates of each object in the standardized equipment instance object set, and extract the corresponding spatial feature parameters; Based on the three-dimensional initial physical coordinates, the spatial feature parameters, and the scene business logic topology diagram, cross-constraint diagnosis is performed to filter out abnormal device instance objects that cause spatial and logical discrepancies. Based on the scenario business logic topology diagram, a virtual force model is constructed for the abnormal device instance object, and the corresponding topology recommendation anchor point is calculated and output. Based on the topology-recommended anchor points, a hidden magnetic adsorption range is constructed in the 3D rendering viewport to perform rigid adsorption on the received dynamic drag coordinates, thereby completing the coordinate correction of the abnormal device instance object. After the coordinate correction legality verification is passed, a topology association loading weight matrix is ​​generated based on the scenario business logic topology diagram, and the updated data is published. The data carrier after publication is parsed to generate basic loading priority weights. When a device alarm event is detected, the comprehensive scheduling priority score is calculated by combining the basic loading priority weights with the topology-related loading weight matrix. Based on the comprehensive scheduling priority score, a high-precision 3D model is extracted to perform video memory pre-assembly.

2. The visualized rapid deployment method for intelligent operation and maintenance scenarios according to claim 1, characterized in that, Before inferring the initial three-dimensional physical coordinates of each object in the standardized set of device instance objects, the process further includes performing a data mapping pre-binding between the device instance objects and external measurement point parameters: The device tag number of the device instance object is compared with the tag code in the measurement point list by regular expression to extract the initial screening matching pairs; Calculate the string similarity score of the Chinese description text of the initial screening matching pair; if the string similarity score is lower than the preset similarity threshold, it is rejected. Read the data protocol type returned by the measurement point channel, and perform a strong data type validation comparison with the internal attribute types supported by the device instance object to establish a data mapping relationship and generate the corresponding data mapping confidence level.

3. The visualized rapid deployment method for intelligent operation and maintenance scenarios according to claim 1, characterized in that, The extraction of corresponding spatial feature parameters includes: extracting the corresponding spatial bounding box scale and logical spatial region attributes; The cross-constraint diagnosis includes: querying the theoretically allowed distance of the basic connection for the connected links formed by the first node and the second node in the business logic topology graph of the scenario; extracting the spatial bounding box scale of the first node and the second node as compensation parameters; determining whether the logical spatial region attributes of the first node and the logical spatial region attributes of the second node are the same; if they cross different physical regions, querying the preset cross-region compensation matrix to assign a tolerance increment value. The adaptive physical allowable distance threshold is generated by summing the product of the basic connectivity theory allowable distance, the maximum value of the spatial bounding box scale of the first node and the second node and the preset model scale compensation coefficient, and the tolerance increment value; the actual physical Euclidean distance between the first node and the second node is then calculated.

4. The visualized rapid deployment method for intelligent operation and maintenance scenarios according to claim 3, characterized in that, The process of filtering out abnormal device instances that exhibit spatial and logical discrepancies includes: Establish a region compatibility rule table and define a decision function to output the decision result on whether direct connectivity is allowed between different logical space region attributes; When the actual physical Euclidean distance is greater than the adaptive physical allowable distance threshold, or when the determination result is that direct connection is not allowed, the corresponding connected edge is marked as a spatial logical divergence edge, and the nodes at both ends of the connected edge are written into the candidate anomaly set. Based on the node's data mapping confidence, whether the regional attribute is missing, the number of participating spatial logical divergent edges, and the distance consistency of stable adjacent nodes, the abnormal device instance object is determined from the candidate anomaly set; the stable adjacent node is a device instance object that has a direct or indirect business connection with the abnormal device instance object and is not written into the candidate anomaly set.

5. The visualized rapid deployment method for intelligent operation and maintenance scenarios according to claim 4, characterized in that, The process of constructing a virtual force model for the abnormal device instance based on the scenario business logic topology graph, and calculating and outputting the corresponding topology recommendation anchor point, includes: using the coordinates of the stable adjacent nodes as fixed anchor points, constructing the virtual force model for the abnormal device instance containing a spring attraction resultant force vector and an anti-overlap repulsion force vector; calculating the spring attraction resultant force vector between the abnormal device instance and the stable adjacent nodes according to the deformation principle of Hooke's Law; traversing the bounding box of the abnormal device instance to search for other entity nodes within the radius, and when it is determined that the bounding box overlaps or the distance is less than a preset safety distance, generating the anti-overlap repulsion force vector in the opposite direction of the line connecting the centers of the two nodes; introducing a system temperature function that decays with the number of iterations to control the coordinate update step size, adding the spring attraction resultant force vector and the anti-overlap repulsion force vector to obtain the total force vector for coordinate iterative update; stopping the iteration when the iteration meets the freezing temperature threshold or the magnitude of the difference between the position coordinate vectors of two consecutive iterations is less than the minimum displacement threshold, and using the final three-dimensional physical coordinates that meet the stopping condition as the topology recommendation anchor point.

6. The visualized rapid deployment method for intelligent operation and maintenance scenarios according to claim 1, characterized in that, The step of constructing a hidden magnetic snapping range in the 3D rendering viewport based on the topology-recommended anchor points to perform rigid snapping on the received dynamic drag coordinates includes: Using the topology-recommended anchor point as the geometric center, the capture radius is dynamically calculated based on the distance from the camera to the topology-recommended anchor point. An invisible spherical collider is constructed in the collision detection layer of the 3D rendering viewport as the hidden magnetic adsorption range. The Euclidean distance between the dynamic drag coordinates and the topology recommended anchor point is continuously determined. When the Euclidean distance is less than or equal to the capture radius, the dynamic drag coordinate data stream of the input device is blocked, and the three-dimensional initial physical coordinates of the abnormal device instance object are locked to the coordinate parameters of the topology recommended anchor point to complete rigid adsorption.

7. The method for rapid deployment of visualized intelligent operation and maintenance scenarios according to claim 1, characterized in that, The step of generating a topology association loading weight matrix based on the scenario business logic topology graph includes: Traverse all nodes in the business logic topology graph of the scenario and use a breadth-first search algorithm to calculate the shortest logical connectivity hop count between any two nodes; if any two nodes belong to an isolated subgraph that is not connected, then the shortest logical connectivity hop count is defined as infinity. Combining the preset service attenuation constant with the shortest logical connectivity hop count, the negative exponent of the natural number base is calculated as the topology association weight score between two nodes, and the topology association weight scores of all nodes constitute the topology association loading weight matrix.

8. The method for rapid deployment of visualized intelligent operation and maintenance scenarios according to claim 1, characterized in that, The generation of basic loading priority weights includes: obtaining the physical view distance between the virtual camera and each device instance object's 3D coordinates in the 3D rendering viewport, and generating the basic loading priority weights based on the ratio between the physical view distance and the global maximum visible view distance threshold; the calculation of the comprehensive scheduling priority score by combining the basic loading priority weights with the topology-related loading weight matrix includes: parsing the device alarm events to extract the source device identifier, locking the corresponding alarm node based on the source device identifier; extracting the row vector corresponding to the alarm node in the topology-related loading weight matrix; introducing a dynamic weight mixing factor driven by the current business state, linearly superimposing the basic loading priority weights with the topology-related weight scores in the vector, and calculating the comprehensive scheduling priority score.

9. The method for rapid deployment of visualization in intelligent operation and maintenance scenarios according to claim 1, characterized in that, The step of extracting a high-precision 3D model based on the comprehensive scheduling priority score and performing video memory pre-assembly includes: The forced high-precision resource loading threshold is dynamically calculated based on the available video memory ratio of the current graphics processor; when the available video memory ratio is higher than the preset upper limit, the forced high-precision resource loading threshold is reduced, and when the available video memory ratio is lower than the preset lower limit, the forced high-precision resource loading threshold is increased. The resource differential update queue is reordered in descending order. For device instance objects in the reordered queue whose comprehensive scheduling priority score is greater than or equal to the forced high-precision resource loading threshold, the high-precision 3D model is directly extracted and downloaded, decoded, and pre-assembled in the background video memory.

10. A visual rapid deployment system for intelligent operation and maintenance scenarios, applied to a visual rapid deployment method for intelligent operation and maintenance scenarios as described in any one of claims 1-9, characterized in that, include: The data access module is used to parse data from external business systems to generate a standardized set of device instance objects and a scenario business logic topology diagram; The diagnostic verification module is used to infer the initial three-dimensional physical coordinates and extract spatial feature parameters, perform cross-constraint diagnostic screening of abnormal equipment instance objects, and use the virtual force model to calculate and output topology recommended anchor points. The fine-tuning release module is used to construct a hidden magnetic adsorption range and perform rigid adsorption based on dynamic drag coordinates. After the review is passed, it generates a topology association loading weight matrix based on the scenario business logic topology diagram and executes the release action. The runtime scheduling module is used to generate basic loading priority weights, and when a device alarm event is detected, it calculates a comprehensive scheduling priority score by combining the topology-related loading weight matrix. Based on the score, a high-precision 3D model is extracted to perform video memory pre-assembly. Data carrier, used to connect various modules for data caching, sharing and publishing.

Images