Building water supply and drainage system teaching demonstration method based on AR technology

Abstract

The invention relates to a building water supply and drainage system teaching demonstration method based on an AR technology. The method comprises the steps of performing mixed space registration processing according to an AR client camera image, a teaching scene plane graph and a reference coordinate to obtain a space transformation matrix; based on a pre-constructed AR resource packet, converting the AR resource packet into an enhanced view according to the spatial transformation matrix, and calling a scene engine and a scene trigger library according to a teaching target to perform matching linkage rule processing on the enhanced view to obtain an event scene; converting the operation event into a scene control command to obtain an interaction command stream; on the basis of a pre-constructed reduced-order simulation model and a scene pre-calculation library, performing hydraulic state solving according to the interaction command stream and the linkage control command to obtain a simulation state stream; based on a preset scoring rule, key performance indexes are extracted from the interaction command flow and the simulation state flow, and a learning evaluation report is obtained. According to the method, physical rationality and operation efficiency are considered through reduced-order simulation and pre-calculation.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of augmented reality, and particularly relates to a building water supply and drainage system teaching demonstration method based on AR technology. BACKGROUND

[0002] With the development of building engineering and information technology, building information modeling (BIM), augmented reality (AR) and visualization simulation combined technology appear, and the traditional building water supply and drainage and fire protection field teaching training transmits system principles through two-dimensional drawings, three-dimensional static models and on-site physical sample plates, engineering simulation is mostly offline batch processing, and output charts or offline animation; emergency training relies on physical drills, video teaching or independent VR platforms; on-site maintenance is through BIM models and tablets to view static information.

[0003] However, the above method is difficult to safely and controllably reproduce dangerous or rare emergency scenarios, physical drills are high in cost and risky, videos and offline animations lack interactivity and empathy, real-time scenes such as fire-triggered water supply and drainage system linkage cannot be intuitively displayed; simulation and on-site visualization are disconnected, high-fidelity simulation results are difficult to accurately superimpose on site, and cannot be updated in real time with teaching operations, affecting students' understanding of dynamic systems; spatial registration and interactive feedback are insufficient, virtual information and real space rely on manual comparison, which is prone to positional deviation, and existing means lack sufficient interaction depth, making it difficult for students to observe real-time changes such as water flow and pressure through operation; device computing power and teaching evaluation are limited, mobile devices are difficult to support complex transient simulation, either simplifying the performance or relying on offline processing, and there is a lack of automatic operation records, which is not conducive to personalized tutoring. SUMMARY

[0004] Therefore, it is necessary to provide a building water supply and drainage system teaching demonstration method based on AR technology, which can organically combine engineering simulation and on-site teaching to achieve higher teaching effectiveness.

[0005] In a first aspect, the application provides a building water supply and drainage system teaching demonstration method based on AR technology, comprising:

[0006] An AR client camera image of a teaching site, a teaching site plan and a reference coordinate are acquired, and a mixed space registration process is performed according to the AR client camera image, the teaching site plan and the reference coordinate to obtain a space transformation matrix; the space transformation matrix is used to establish a mapping relationship between a virtual model coordinate system and a real scene coordinate system;

[0007] Based on the pre-constructed AR resource package, the three-dimensional water supply and drainage model and the visualization resources in the AR resource package are converted into an augmented view superimposed on the teaching site according to a spatial transformation matrix, and the augmented view is processed according to matching linkage rules of a scenario engine and a scenario trigger library according to a teaching target, to obtain an event scenario; the event scenario includes a triggerable linkage control command;

[0008] In response to obtaining an operation event of a user on the event scenario, the operation event is converted into a scenario control command, to obtain an interactive command stream;

[0009] Based on a pre-constructed reduced-order simulation model and a scenario pre-computation library, hydraulic state solving is performed according to the interactive command stream and the linkage control command, to obtain a simulation state stream; the simulation state stream includes a node pressure vector, a pipe segment flow vector and device state information;

[0010] Based on a preset scoring rule, key performance indicators are extracted after the interactive command stream and the simulation state stream are aligned according to a time axis, to obtain a learning evaluation report and provide operation playback.

[0011] In one of the embodiments, a spatial transformation matrix is obtained through mixed space registration processing of an AR client camera image, a teaching site plan and a reference coordinate, including:

[0012] The AR client camera image is sequentially subjected to visual synchronization positioning and map construction, to obtain a local coordinate system and a feature point cloud;

[0013] The teaching site plan is subjected to drawing alignment processing based on the reference coordinate and the local coordinate system, and key features of the teaching site plan are extracted, to obtain feature reference points;

[0014] The feature reference points are matched with site features in the AR client camera image, and a global transformation matrix is estimated;

[0015] When the automatic matching accuracy is insufficient, identifiable markers are set for the AR client camera image and the teaching site plan, and the global transformation matrix is corrected through marker recognition, to obtain the spatial transformation matrix.

[0016] In one of the embodiments, the AR resource package is obtained through the following method:

[0017] Obtain an architectural design file and architectural water supply and drainage design parameters;

[0018] According to the architectural design file and the building water supply and drainage design parameter, semantic extraction is performed to obtain semantic information, and a hydraulic topology graph and corresponding 3D geometric data are established based on the semantic information; the hydraulic topology graph is a component node and a component connection edge structure, and corresponding 3D geometric data node attributes are stored in each component node; the semantic information includes drainage components and attributes thereof; the drainage components include pipes, nodes, valves, pumps and sprinklers;

[0019] Based on the hydraulic topology graph and the 3D geometric data, visual resources for AR rendering are generated, and a component attribute library is established to obtain an AR resource package.

[0020] In one of the embodiments, the reduced-order simulation model and the scenario pre-computation library are obtained by the following method:

[0021] Based on the engineering fluid mechanics conservation equation, the hydraulic topology graph is mapped to an engineering hydraulic equation set, and a reduced-order simulation model is obtained by performing reduced-order approximation on the flow of each pipe section according to the engineering hydraulic equation set; the reduced-order simulation model maintains the node flow conservation and energy approximation relationship; the engineering fluid mechanics conservation equation includes flow conservation and energy conservation;

[0022] For the interactive operation and fault scenarios required by the teaching target, high-fidelity transient simulation is used to calculate the time series samples under each scenario, and the time series samples are stored in the scenario pre-computation library according to the parameterized index.

[0023] In one of the embodiments, based on the pre-constructed reduced-order simulation model and the scenario pre-computation library, the hydraulic state is solved according to the interactive command stream and the linkage control command to obtain a simulation state stream, including:

[0024] When the operation event corresponding to the interactive command stream is a steady-state change, the reduced-order simulation model is called to perform steady-state engineering hydraulic equation solving to obtain the simulation state stream;

[0025] When the operation event corresponding to the interactive command stream is a transient event, the pre-computed time series sample matched with the operation event parameter is searched in the scenario pre-computation library, and the approximate transient response is reconstructed by sample interpolation to obtain the simulation state stream.

[0026] In one of the embodiments, according to the teaching target, the scenario engine and the scenario trigger library are called to perform matching linkage rule processing on the augmented view to obtain an event scenario, including:

[0027] The fire prevention rules and the sprinkler control logic are coded as an event-driven finite state machine or a rule engine to obtain a scenario trigger library; the scenario trigger library takes the fire prevention rules, the detector-sprinkler linkage relationship and the pump control rules as searchable trigger items;

[0028] The scenario engine retrieves a linkage rule matching the teaching target from a scenario trigger library, and generates an event scene based on the matched linkage rule; the event scene contains linkage control commands to be activated for the reduced-order simulation model and operation control; the teaching target corresponds to a fire source event or a sensor signal.

[0029] In one of the embodiments, the method further comprises:

[0030] The perspective and conversation information of each student AR client are collected and associated with the teacher AR client;

[0031] The teacher AR client is taken as a master node, state differences of each student AR client are broadcast based on a master-slave synchronization architecture, and each student AR client receives and renders a synchronization state stream published by the teacher master node based on conversation association management authority.

[0032] In a second aspect, the application further provides a building water supply and drainage system teaching demonstration device based on AR technology, comprising:

[0033] The registration module obtains an AR client camera image of a teaching site, a teaching site plan and a reference coordinate, and performs hybrid space registration processing on the AR client camera image, the teaching site plan and the reference coordinate to obtain a space transformation matrix; the space transformation matrix is used to establish a mapping relationship between a virtual model coordinate system and a real scene coordinate system;

[0034] The scenario view module is used to convert a three-dimensional water supply and drainage model and visual resources in an AR resource package into an augmented view superimposed on the teaching site based on the space transformation matrix and the pre-constructed AR resource package, and perform matching linkage rule processing on the augmented view based on a scenario engine and a scenario trigger library according to a teaching target to obtain an event scene; the event scene includes triggerable linkage control commands;

[0035] The interaction module is used to convert an operation event of a user on the event scene into a scene control command to obtain an interaction command stream in response to obtaining the operation event of the user on the event scene;

[0036] The simulation solving module is used to perform hydraulic state solving based on a pre-constructed reduced-order simulation model and a scenario pre-computation library according to the interaction command stream and the linkage control command to obtain a simulation state stream; the simulation state stream includes a node pressure vector, a pipe segment flow vector and device state information;

[0037] The evaluation module is used to extract a key performance indicator after aligning the interaction command stream and the simulation state stream on a time axis based on a pre-set scoring rule to obtain a learning evaluation report and provide an operation playback.

[0038] In a third aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any of the above-mentioned AR technology-based building water supply and drainage system teaching demonstration methods when executing the computer program.

[0039] In a fourth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the steps of any of the above-mentioned AR technology-based building water supply and drainage system teaching demonstration methods.

[0040] The above-mentioned AR technology-based building water supply and drainage system teaching demonstration method comprises the following steps: obtaining an AR client camera image, a teaching site plan and a reference coordinate, and performing hybrid space registration; calculating a space transformation matrix; and accurately superimposing a virtual water supply and drainage three-dimensional model in space to a teaching site to intuitively present system structure and flow conditions. A scenario engine is used to retrieve a scenario trigger library and generate an event scenario, so as to trigger a spray / pump / valve linkage demonstration according to teaching needs. User operation events on the event scenario are semantically converted into scenario control commands to form an interactive command stream to drive simulation. The interactive command stream and the linkage control command are solved based on a reduced-order simulation model and a scenario pre-computation library to output a simulation state stream containing node pressure, pipe segment flow and device state, and a physically interpretable hydraulic response is made under an interactive condition to drive visualization. The interactive command stream and the simulation state stream are aligned according to a time axis, and performance index features are extracted to automatically generate a learning evaluation report and provide operation playback, thereby supporting teaching feedback and assessment. Emergency linkage scenarios such as firefighting are simulated without endangering personnel and equipment, thereby facilitating teaching demonstration and training. Dynamic hydraulic information is presented in a superimposed visualization form in a real space, thereby improving students' understanding of complex system space-time behavior. Simulation and visualization driven by the interactive command stream enable students to instantly observe the influence of their operation on system state, thereby improving learning initiative and experiencing instant feedback. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort.

[0042] Figure 1 FIG. 1 is a flowchart of the AR technology-based building water supply and drainage system teaching demonstration method of the present application;

[0043] Figure 2 FIG. 3 is a flowchart of the sub-step of step S103;

[0044] Figure 3 A sub-step flowchart for step S102;

[0045] Figure 4 A composition structure diagram of the AR technology-based building water supply and drainage system teaching demonstration device. DETAILED DESCRIPTION

[0046] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0047] In one embodiment, as shown in Figure 1 An AR technology-based building water supply and drainage system teaching demonstration method is provided, and the present embodiment is exemplarily described by taking the method applied to a terminal. It should be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is realized through the interaction of the terminal and the server. In the present embodiment, the method includes the following steps:

[0048] S101, acquiring an AR client camera image of a teaching site, a teaching site plan and a reference coordinate, and performing mixed space registration processing according to the AR client camera image, the teaching site plan and the reference coordinate to obtain a space transformation matrix; the space transformation matrix is used to establish a mapping relationship between a virtual model coordinate system and a real scene coordinate system.

[0049] The teaching site refers to a physical environment for classroom or practical training, which can be an indoor classroom, a laboratory, or a floor or room in a real building. The AR client camera image refers to a sequence of images continuously captured by the camera of an AR device such as a tablet or AR glasses on the student side or the teacher side. The image can include an RGB frame, and when the device supports it, it can also include a depth map or inertial sensor data. The teaching site plan refers to planar building information corresponding to the teaching site, such as BIM plan, CAD / DWG, or vectorized plan. The reference coordinates refer to coordinate system information used to calibrate or define the positional relationship between the teaching site and the design model, such as local coordinates or local coordinate values of engineering reference points used in BIM. Further, a spatial transformation matrix is obtained by hybrid space registration, so that the virtual three-dimensional water supply and drainage model can be superimposed in the real scene with correct position, orientation and scale, thereby ensuring the spatial consistency and occlusion relationship between the visualization object and the real geometry. Specifically, visual synchronization positioning and map construction (SLAM) are performed on the AR client camera image to obtain the local coordinate system of the device in the site and the feature point cloud to be registered, which includes a set of key frames with timestamps, local feature descriptors, and dense or sparse point cloud representations. Key features such as wall corners, doors and windows, column positions, and stairwell boundaries are extracted from the teaching site plan under its coordinate reference, and a set of feature reference points are generated. If there is a BIM model, the geometric primitives of the BIM can be used as high-precision references. The feature reference points of the plan are matched with the site features obtained by SLAM through feature matching or least squares registration based on geometric constraints to estimate the global transformation matrix. When the automatic matching accuracy does not meet the preset threshold, identifiable markers can be introduced and placed at known positions in the site, and the accurate detection results of the markers are used to correct the aforementioned global transformation, and finally a spatial transformation matrix that meets the accuracy requirements is output. The matrix includes rotation and translation components, as well as scale factors, and can be updated slightly over time as the device moves, thereby supporting real-time tracking and maintaining model alignment.

[0050] S102, based on the pre-constructed AR resource package, the three-dimensional water supply and drainage model and the visualization resource in the AR resource package are converted into an augmented view superimposed on the teaching site according to the spatial transformation matrix, and the scenario engine and the scenario trigger library are called according to the teaching target to process the augmented view according to the matching linkage rule, to obtain an event scenario; the event scenario includes a triggerable linkage control command.

[0051] The AR resource package refers to a set of rendering resources and metadata that can be directly loaded by the client offline, including a three-dimensional water supply and drainage model with multi-level detail LOD, a component attribute library, i.e., the unique identification, pipe diameter, material, valve model, pump curve, etc. of each component, and visual texture and particle system templates for representing water flow and spray atomization effects, and mapping rules for driving simulation visualization. The three-dimensional water supply and drainage model and visualization resources in the AR resource package are converted into an augmented view superimposed on the teaching site, i.e., the model in the resource package is projected into the camera view cone of the device according to the spatial transformation matrix, and virtual information is superimposed on the camera image with appropriate rendering strategies to achieve visual fusion and interaction. Specifically, the spatial transformation matrix is read and each component in the AR resource package is subjected to coordinate transformation according to its geometry and metadata to ensure that the model is aligned with the real physical geometry. In the rendering stage, a semi-transparent or visible hierarchical presentation method is used to present the pipeline distribution and the internal structure of the building at the same time. If the device provides a depth map or estimates the depth by monocular estimation, the depth information is used to realize real occlusion processing, so that the pipeline is correctly hidden or displayed in front of the desktop or behind the object. For dynamic visualization such as water flow, the rendering module generates visual effects based on the simulation engine or an approximate view, such as particle rate and color mapping, and for teaching purposes, specific subsystems can also be highlighted, hidden or sectioned to help students observe the internal flow path.

[0052] Further, the scenario trigger library stores linkage rules or pre-defined scenario templates indexed by teaching goals, such as fire demonstration scenarios, local valve failure demonstration, etc. The scenario engine retrieves the corresponding rules and generates an event scenario according to the current teaching task or teacher trigger condition. The event scenario is an encapsulation body that describes a set of linkage control commands that can be triggered or suggested to be triggered in this teaching scenario, such as turning on the sprinkler in a certain area, switching a certain pump to standby, isolating a certain section of the pipe network, etc. The linkage control commands can then be issued to the simulation module or used to guide student interaction.

[0053] S103, in response to obtaining the operation event of the user on the event scenario, converting the operation event into a scene control command to obtain an interaction command stream.

[0054] Illustratively, the operation event refers to the interactive input generated by the user in the AR augmented view, which can be a direct touch click, long press, swipe, gesture recognition based on the camera, voice control command, or instructions sent through an external input device. The operation event usually has an operation subject, a target component identifier, such as a valve ID, a pump ID, a pipe section ID, and includes an operation type, such as opening, closing, adjusting, observing, and an operation parameter, such as an opening percentage, an adjustment rate, an observation angle, etc., and is accompanied by a timestamp and client session information.

[0055] Further, the original input is pre-processed for events, including de-bouncing, gesture recognition confidence evaluation, natural language parsing and object recognition, and the pre-processed events are semantized into scene control commands through a rule mapping table or an instruction compiler, which describes the actions to be performed in a standardized format. Exemplarily, the command structure includes a unique identifier of the target component, an action type, a parameter value, an initiator ID and a timestamp. Optionally, a legality check is performed before generating the command to avoid invalid or harmful operations, such as prohibiting direct disconnection of the main water supply in some teaching scenarios, and the permissions of teachers and students can be identified to determine whether the command is immediately issued to the simulation module or first approved by the teacher end. All generated scene control commands are organized into an interactive command stream in chronological order as the core data stream for driving simulation and recording evaluation.

[0056] In S104, based on the pre-constructed reduced-order simulation model and the scenario pre-computation library, the hydraulic state is solved according to the interactive command stream and the linkage control command to obtain a simulation state stream. The simulation state stream includes a node pressure vector, a pipe segment flow vector and device state information.

[0057] Exemplarily, the reduced-order simulation model refers to a simplified mathematical model established on the topological level of the water supply and drainage system based on the principles of engineering fluid mechanics, which has the characteristics of considering both physical rationality and computational efficiency. That is, through model reduction, linearization or using a grid graph theory solving method, the complex continuous medium problem is transformed into a discrete solving problem for nodes and pipe segments, thereby supporting quasi-real-time calculation in a mobile terminal or interactive environment. The scenario pre-computation library is a time series response sample calculated in advance in an offline high-fidelity simulation environment for a number of typical teaching scenarios or fault conditions, such as pressure waves caused by spray start-stop, transient response caused by rapid valve closure, etc. These samples are stored in a parameterized manner for quick retrieval and used for visual reconstruction or as compensation information for solving the reduced-order model during online teaching.

[0058] Specifically, when receiving the interactive command stream and the linkage control command generated by the scenario engine, the simulation coordination module acts as a hub to map the commands to the input boundary conditions and operation sequences of the reduced-order simulation model, and then drive the model to update the state. For example, the interactive command stream is parsed to convert the commands into boundary conditions such as valve opening, pump speed, source and sink flow, etc., the current system state vector is assembled and maintained, including node pressure reference, valve position vector, pump state set, etc., and the current system state vector is transmitted to the reduced-order simulation model for solving to generate a new simulation state. At the same time, based on the characteristics of the linkage control command or the teaching scenario, it can be searched in the scenario pre-computation library whether there is a pre-computed sample with similar parameters, and if a suitable sample is found, the sample can be used to quickly reconstruct the transient response or used as a correction to the output of the reduced-order model, so as to balance the speed and physical fidelity. The output of the simulation module is in the form of a simulation state stream, including node pressure vector, pipe segment flow vector, and discretized state information of various devices, and is synchronized with the interactive command stream through time stamping, for the rendering module to update the augmented view.

[0059] Optionally, the reduced-order simulation model can use a sparse matrix solver represented by graph theory, an iterative method based on node-ring, or a preconditioned linear solution technique to improve real-time performance; the index of the scenario pre-computation library needs to support fast matching and retrieval according to event type, parameter vector and initial state, in order to meet the low delay requirement in the teaching process.

[0060] S105, based on the preset scoring rules, the interactive command stream and the simulation state stream are aligned on the time axis to extract key performance indicators, obtain a learning evaluation report, and provide an operation playback.

[0061] Illustratively, the scoring rules refer to a set of rules and quantitative standards defined and configurable in advance by the teaching, for objectively evaluating the operation performance and decision quality of the student in the AR teaching process. The scoring rules can include operation-based rules, such as completing valve operations in the required order according to the specification, response time-based rules, such as the time threshold from the fire source trigger to taking the correct control action, system state-based rules, such as whether to avoid triggering low pressure alarms or pump overload, and safety-based rules, such as whether to prohibit dangerous operations.

[0062] Specifically, the interactive command stream and the simulation state stream are time-synchronized, the time stamp is corrected and reordered considering network delay and local cache time difference, and further, key performance indicators (KPIs) such as response time, operation accuracy, stability indicators, and resource utilization efficiency are extracted from the time-aligned data according to the scoring rules. Based on the KPIs, the evaluation engine calculates the comprehensive score of the student and generates diagnostic feedback, including operation playback, i.e., visualizing the operation sequence and the corresponding simulation state timeline, and specific improvement suggestions, such as pointing out when to isolate which section to reduce the impact on the key node pressure.

[0063] In the above-mentioned AR technology-based building water supply and drainage system teaching demonstration method, the space transformation matrix can accurately position and align the virtual three-dimensional water supply and drainage model in the teaching site, ensuring consistency in position, direction, and scale between the model and the real building geometry, thereby improving the accuracy of visualization and teaching credibility. The event scenario can automatically or according to the teaching trigger generate a standardized linkage control command sequence according to the preset rules, thereby demonstrating the expected fire / water supply linkage process, improving the consistency and controllability of teaching. The interactive command stream can directly associate the student's operation with the engineering physical quantity, generate an interpretable physical response, and be used for visualization, meeting the closed-loop demonstration requirement from operation to physical result in teaching. When dealing with steady-state or routine changes, a reduced-order model can be quickly called to give real-time or quasi-real-time results; when encountering complex transient scenarios, precomputed samples can be used to quickly reconstruct the response, thereby ensuring physical reasonableness while meeting the performance constraints of AR devices and meeting the latency requirements of interactive teaching. The teaching process can be accurately time-sequenced, supporting complete playback and post-analysis, facilitating teacher review, quantitative evaluation of student operations, and improvement of teaching design, and can automatically generate a quantitative evaluation report based on operation effect and system response, providing teachers with comparable learning performance and diagnostic recommendations, reducing evaluation subjectivity and supporting personalized tutoring. Using simulation instead of physical practice can repeatedly demonstrate dangerous situations in a zero-risk environment, thereby improving the repeatability and safety of teaching, and through visualized flow paths, pressure / flow presentations, and immediate feedback, students can more quickly establish a system causality awareness, thereby improving classroom teaching efficiency and the depth of learning effectiveness.

[0064] In one embodiment, a mixed space registration process is performed according to the AR client camera image, the teaching site plan, and the reference coordinates to obtain a space transformation matrix, including:

[0065] S11, visual synchronization positioning and map construction are performed in sequence according to the AR client camera image to obtain a local coordinate system and a feature point cloud.

[0066] Illustratively, visual simultaneous localization and mapping (SLAM) refers to the process of recovering the trajectory of a device in a scene and simultaneously generating a sparse or semi-dense map of the environment by tracking features between camera frames, estimating camera poses, and detecting loop closures. Specifically, feature detection and descriptor generation is performed on each frame of RGB images continuously captured by the AR client, and feature matching is performed between adjacent frames to obtain a preliminary estimate of the relative pose between frames. Further, camera pose optimization is performed using visual-inertial fusion or pure visual methods to reduce drift, and loop detection and closure are performed as necessary to correct for long-term accumulated errors. In parallel, a time-ordered feature point cloud is generated by aggregating successfully matched key points, which can be sparse or extended to semi-dense / dense point clouds on supported devices, resulting in a set of time-stamped camera pose sequences, a set of keyframe images, and a corresponding set of three-dimensional feature points, constituting a local coordinate system and feature point cloud.

[0067] S12, based on the reference coordinate and the local coordinate system, a floor plan of a teaching site is processed for drawing alignment, and key features of the floor plan of the teaching site are extracted to obtain feature reference points.

[0068] Illustratively, the floor plan of the teaching site can come from a BIM / IFC model, a CAD / DWG file, or a vectorized architectural plan image, and the reference coordinate refers to the coordinate reference associated with the floor plan, such as the project reference point in BIM, the coordinate label in architectural surveying, or the control point on the construction drawing. Specifically, if the floor plan is originally BIM / IFC, the room boundaries, door and window positions, wall lines, and structural axes can be directly extracted from the model as high-confidence geometric primitives, and the primitives are coordinate-transformed in their original reference coordinate system; if only CAD or bitmap is available, vectorization and semantic recognition of the drawing are required, such as identifying wall lines, door and window symbols, and room numbers through primitive analysis or image processing, and then extracting the two-dimensional coordinates of these key elements, mapping the coordinates of the plan elements to values consistent with the engineering reference according to the reference coordinate, such as converting CAD millimeter units to meters, and corresponding the origin of the model coordinate system to the site survey reference.

[0069] S13, matching the feature reference points with the site features in the camera image of the AR client, and estimating a global transformation matrix.

[0070] Illustratively, the global transformation is represented by a rigid or similarity transformation, including a rotation matrix, a translation vector, and a scale factor, and is generally represented by a spatial transformation matrix. Specifically, when both the floor plan and the camera image support matching based on local image descriptors, a descriptor can be generated for each feature reference point on the floor plan and searched for in the SLAM keyframes, or the geometric points of the floor plan and the SLAM point cloud can be matched through geometric constraints, i.e. using projection consistency or nearest neighbor search to find the closest point in the three-dimensional point cloud to the two-dimensional point on the floor plan.Figure Two corresponding 3D point candidates. Optionally, to enhance the robustness of matching, multi-modal constraints are used, such as geometric distance, normal consistency, angle difference, small region texture similarity, and pre-defined semantic labels, jointly as the matching score. After obtaining several pairs of candidate matching points, a robust estimation algorithm is used to eliminate false matches and estimate the optimal rigid transformation parameters. Specifically, the algorithm repeatedly samples the minimum sample set from the candidate matching set, calculates the transformation and verifies the re-projection error of other matching points to identify the inlier set, and performs weighted least squares fitting on the inlier set to obtain the rotation matrix, translation vector and optional scale factor, thereby constructing the global transformation matrix T.

[0071] S14, when the automatic matching accuracy is insufficient, setting identifiable markers for the AR client camera image and the teaching site plan, and correcting the global transformation matrix through marker recognition to obtain the spatial transformation matrix.

[0072] The insufficient automatic matching accuracy can be based on quality indicators, such as the inlier ratio being lower than a threshold, the re-projection RMSE exceeding a threshold, or the inlier distribution being uneven. When any criterion is triggered, several identifiable markers are automatically prompted to be arranged in the teaching site, which can be standardized two-dimensional codes, AprilTag, ArUco labels or target plates with known geometric size and position information. The positions of the markers should cover the spatial distribution of the teaching area as much as possible, and the reference coordinates of each marker on the teaching site plan are recorded. Through the AR client camera, the markers are detected and recognized to obtain their pixel coordinates in the camera image and the unique ID obtained from the marker code. Combined with the corresponding marker reference coordinates recorded in the plan in advance, several pairs of 2D-3D or 3D-3D corresponding points can be directly obtained, and based on the identifiable corresponding relationship, the least squares or Umeyama algorithm is used to directly solve the approximate rigid or similarity transformation parameters, thereby correcting and replacing the preliminary transformation, and finally outputting the corrected spatial transformation matrix T. Optionally, only the minimum number of markers are arranged for the correction of scale or rotation based on the preliminary automatic registration, or the markers are placed in key controlled positions, thereby reducing the site preparation cost as much as possible while ensuring the accuracy.

[0073] In one of the embodiments, the AR resource package is obtained by the following method:

[0074] S21, obtaining the architectural design file and the architectural water supply and drainage design parameters.

[0075] Illustratively, the original design data for constructing a virtual plumbing model in the teaching system needs to be collected. The architectural design files usually include BIM / IFC models, Revit export files, CAD / DWG files, vector or raster images of engineering drawings, and documents related to architectural locations, room numbers, and structural grid. The architectural plumbing design parameters refer to a set of data used to express the engineering properties of the plumbing system, such as pipe diameter, pipe material, slope, connector type, local resistance coefficient, pump performance curve, sprinkler layout table, valve model and location, detector and controller layout, design flow and design pressure, etc. At the same time, relevant specification parameters and engineering instructions need to be collected, such as national / industry plumbing design standards, sprinkler activation threshold, and minimum allowable pressure.

[0076] S22, semantic extraction is performed based on the architectural design files and the architectural plumbing design parameters to obtain semantic information, and a hydraulic topology graph and corresponding 3D geometric data are established based on the semantic information; the hydraulic topology graph is a structure of component nodes and component connection edges, and each component node stores corresponding 3D geometric data node attributes; the semantic information includes plumbing components and their attributes; the plumbing components include pipes, nodes, valves, pumps, and sprinklers.

[0077] Illustratively, semantic extraction refers to identifying and extracting component types in the plumbing system from BIM / CAD / drawings and parameter tables, such as pipes, joints, valves, pumps, water tanks, sprinkler heads, and detectors, and their attributes, such as internal diameter, material, length, flange type, valve opening range, pump characteristic curve, installation coordinates, etc., and standardizing these identification results into an internal semantic model. Specifically, for IFC / BIM, semantic tags are used to extract components and read their attribute dictionaries; for CAD vector drawings, symbol recognition and connectivity reasoning are performed based on a symbol library and topological relationships; for imaged drawings, image recognition and symbol classifiers are used, supplemented by text OCR to extract annotations. The identified components are assigned unique identifiers. Based on the identified components and their connection information, a hydraulic topology graph is constructed, where nodes represent fluid nodes, such as junctions, branches, sprinkler nodes, water tank interfaces, or device endpoints, and edges represent pipe segments with edge attributes, such as internal diameter, length, roughness, local resistance coefficient, etc. The topology construction process performs connectivity checking, loop detection, and ambiguity resolution, such as line segment disconnection or overlap commonly found in CAD, which requires the use of endpoint matching, topology arc merging, threshold distance merging, and other topology repair algorithms for cleaning. For pumps, valves, and other devices, not only are node / edge relationships established in the topology graph, but their control logic, such as valve type, action response time, pump curve parameters, etc., is also stored as additional attributes for simulation module invocation.

[0078] Further, generate or map corresponding 3D geometry data for each topological node or edge, which is derived from geometric bodies in BIM model or generated by parametric modeling, such as generating pipe segment geometry according to pipe diameter, elbow radius, support position and other parameters. In the case of only two-dimensional drawings, three-dimensional position information can be inferred by regularized height assumptions and construction specifications, and marked as estimated data to prompt manual confirmation. The 3D geometry data should include topological references, i.e. each geometric entity contains its node / edge ID in the hydraulic topology graph, material information and rendering related metadata such as normal, UV coordinate, LOD preset, etc.

[0079] S23, generate visualization resources for AR rendering based on the hydraulic topology graph and 3D geometry data and establish a component attribute library to obtain an AR resource package.

[0080] Illustratively, the output hydraulic topology graph and 3D geometry data are converted into an AR resource package that can be directly loaded, rendered and queried by the client, including multi-level detail geometry (LOD) for rendering, materials and textures, particle / shader templates for flow representation, component attribute library and manifest file for quick indexing and version control. The component attribute library is a structured database that stores the engineering attributes of each component, such as pipe diameter, material, length, roughness coefficient, valve type, pump curve parameters, component unique identifier, metadata source, maintenance record, etc., and provides a quick access interface for simulation modules and interaction modules. Specifically, mesh optimization and LOD generation are performed on the 3D geometry data. Mesh optimization uses mesh simplification algorithms to remove redundant vertices and generate models of different precision levels to switch between different device capabilities or different observation distances. Further, the appearance of pipes and equipment is parameterized and replaced, for example, pipes are represented by a cylindrical parameter and detailed by a map when needed to significantly reduce the number of surfaces, generate or specify the required rendering materials, transparency and section effect parameters, so that semi-transparent display, section viewing or highlight identification can be implemented as needed during teaching. For water flow and sprinkler effects, pre-set particle system templates or GPU-based streamline rendering shaders are mapped with simulation output variables such as flow rate, pressure, etc. to drive visualization intensity, such as faster particle speed with higher flow rate and more red color with higher pressure.

[0081] Further, the attributes of the nodes and edges of the hydraulic topology graph are sorted into the component attribute library, stored in a lightweight database or indexed JSON file, and each component records a set of required attributes, i.e. internal diameter, length, roughness coefficient, local resistance coefficient, material, installation coordinates, connection relationship, device control parameters, etc., and optional attributes such as maintenance period. At the same time, in order to facilitate the call of the simulation module, the component attribute library needs to provide a one-to-one mapping relationship with the fields of the reduced-order simulation model, and record the correspondence between each component and the simulation ID in the resource package.

[0082] In one embodiment, the reduced-order simulation model and the scenario pre-computation library are obtained through the following method:

[0083] S31. Based on the engineering fluid dynamics conservation equations, the hydraulic topology diagram is mapped to the engineering hydraulic equation set, and the flow rate of each pipe segment is approximated by a reduced order according to the engineering hydraulic equation set to obtain a reduced-order simulation model; the reduced-order simulation model maintains the nodal flow rate conservation and energy approximation relationship; the engineering fluid dynamics conservation equations include flow rate conservation and energy conservation.

[0084] The output of the reduced-order simulation model is a set of discretized state variables, such as pressure vectors at each node, flow vectors at each pipe segment, and updatable equipment state variables, such as valve opening and pump speed. An illustrative example is the hydraulic topology diagram. Each node Treating nodes as having conserved mass, establish the node flow conservation equation. ,in For connecting nodes With adjacent nodes Pipeline flow rate, This refers to external injection / extraction items such as water supply sources, drainage collection points, and sprinkler outlets. Regarding energy / head relationships, this includes pressure drop or head loss in the pipe section. It can be expressed using common engineering empirical formulas, such as the Darcy–Weisbach or Hazen–Williams form. Hazen–Williams is commonly used in water supply projects. This can be converted into a relationship between flow rate and head loss, where length ,diameter Roughness coefficients are coupled with flow rate to form a system of nonlinear equations. Furthermore, equipment such as pumps, valves, and storage tanks are represented as node or edge sub-models with attributes. Pumps are represented by pump characteristic curves. Control parameters, such as rotational speed and start / stop status, are used. Valves are represented by opening-flow resistance relationships or specific local loss coefficients, and can include action delay and action rate parameters. Spray nodes can serve as instantaneous injection boundaries under specific triggering conditions. To achieve quasi-real-time solutions, the above nonlinear equations are reduced in order or linearized. For example, around the current steady-state operating point... A first-order Taylor expansion of the nonlinear terms yields a linear time-varying model. or algebraic equations, and then use sparse linear solvers to solve them quickly. Alternatively, for large-scale networks, reduce the number of unknowns by merging nodes or eliminating external nodes, retain key nodes such as pumps, special control valves, spray junctions, and convert minor branches into equivalent impedances. Also, for transient response, use a small number of basis functions such as empirical orthogonal functions or pre-computed modes to represent the time response, thereby converting the time evolution problem into a low-dimensional coefficient evolution problem.

[0085] Convert the reduced-order equations into a sparse matrix solution problem, and use sparse LU, preconditioned conjugate gradient (PCG), or QR / least squares solvers with transactions to obtain the steady-state solution; when short-time dynamic response is required, use implicit / explicit time integration to integrate the simplified transient model.

[0086] S32, for the interactive operation and fault scenarios required by the teaching goal, use high-fidelity transient simulation to calculate the time series samples under each scenario, and store the time series samples according to the parameterized index in the scenario pre-computation library.

[0087] Illustratively, for transient phenomena that are often required to be demonstrated in teaching, such as pressure waves caused by rapid closing of valves, pump response caused by simultaneous activation of multiple sprinklers, local overpressure / low pressure caused by short circuit or blockage of the pipe network, etc. An offline high-fidelity time series sample library is constructed for the scenario pre-computation library, which contains time series response samples in a multi-dimensional parameter space, including initial operating conditions, event parameters such as trigger position, trigger intensity, action rate, etc., as well as device characteristics, etc. During online teaching, approximate transient responses can be quickly reconstructed by retrieving and interpolating these samples, thereby avoiding real-time high-cost simulation while maintaining physical credibility and interpretability. Specifically, according to the teaching syllabus and common teaching goals, representative scenarios are listed, such as single valve rapid closing (parameters: valve position, closing time constant, valve position relative to pipe segment position), regional fire triggering multiple sprinklers (parameters: trigger area, number of sprinklers simultaneously opened and closed, sprinkler opening), pump tripping (parameters: pump model, tripping time, standby pump switching strategy), etc.

[0088] Define a parameter vector for each type of scenario, and set a discrete sampling level for each parameter, such as {0.1s, 0.5s, 1s, 2s} for valve closing time, and key node set for trigger position, etc. In an offline high-performance computing environment, use professional transient simulation tools to run simulations with sufficient time step for each parameter combination to obtain time series data sets: node pressure, pipe segment flow, boundary device response, and several derived indicators such as maximum pressure difference. Each sample entry records metadata, including scenario category, parameter vector, initial operating condition identifier, timestamp, simulator version and numerical settings, storage location of compressed representation or original time series, and error estimate, resulting in a scenario pre-computation library.

[0089] In one embodiment, such as Figure 2 As shown, based on the pre-built reduced-order simulation model and scenario pre-calculation library, the hydraulic state is solved according to the interactive command flow and linkage control commands to obtain the simulation state flow, including:

[0090] S201. When the operation event corresponding to the interactive command flow is a steady-state change, the reduced-order simulation model is called to solve the steady-state engineering hydraulic equations to obtain the simulation state flow.

[0091] Indicatively, steady-state changes refer to operations in the interactive command stream that are relatively slow on a time scale. These can be approximated as the network reaching a new steady state after the operation is completed, such as gradually adjusting valve opening, slowly changing pump speed, or switching water supply sources. Specifically, instructions from the interactive command stream and linkage control commands are parsed, semantically representing each command as a change in simulation input boundary conditions or model parameters. For example, valve-related operations are mapped to changes in the corresponding valve opening parameters, pump-related operations are mapped to changes in pump speed or start / stop status, and switching water supply sources is mapped to changes in the node's external injection / extraction rate. The parsed parameters are merged with the current system state to form a new steady-state condition description, which is then fed into the steady-state solver of the reduced-order simulation model. The steady-state solver, based on the reduced-order simulation model, represents the node-pipeline network as a set of nonlinear algebraic equations, including node flow conservation equations and energy / head loss expressions based on empirical formulas. Since the reduced-order model has already employed reduction strategies such as linearization, node merging, or equivalent impedance reduction, the solver can use efficient solution algorithms within a sparse matrix framework, such as Newton-Raphson iteration combined with a sparse linear equation solver or a pre-conditional iteration method, to quickly obtain a steady-state solution. To ensure numerical stability and convergence, the solver needs to set reasonable initial guesses and monitor residuals and iteration convergence indices during iteration. When the iteration fails to converge within the preset number of iterations or time budget, a backoff strategy is adopted and an anomaly is reported to the instructor. After the solution is completed, the steady-state solver will output a simulation state vector labeled with timestamps and package it into a simulation state flow term through a unified data interface, typically including fields such as node pressure vectors, pipe flow vectors, and equipment state sets. For example, in a multi-branch water supply network, if a student adjusts the opening of a valve from 30% to 70%, the command will be mapped to a change in the valve resistance coefficient. Based on the current state as an initial guess, a reduced-order steady-state solver will be invoked, completing the iteration within tens to hundreds of milliseconds and outputting the new pressure and flow distribution. After receiving the simulation state stream, the client smoothly reflects the change at a rendering frame rate of 60Hz or 30Hz, so that the student can observe the impact of valve adjustment on the pressure of surrounding nodes in real time.

[0092] S202、When the operation event corresponding to the interactive command stream is a transient event, retrieve the pre-computed time series samples matching the operation event parameters in the scenario pre-computation library, and reconstruct the approximate transient response by sample interpolation to obtain the simulation state stream.

[0093] Illustratively, transient events refer to those operations that cause strong nonlinearity or rapid changes in a short time scale, such as instantaneous rapid closing of valves, sudden pump tripping, simultaneous activation of a large number of sprinklers, or sudden blockage of pipe sections, etc. Such events can produce pressure fluctuations, reflections, and instantaneous overpressure or low pressure phenomena, and require high spatiotemporal resolution transient solutions to obtain physically plausible responses. Specifically, the transient events in the interactive command stream are parameterized to form a parameter vector of the operation event, which includes but is not limited to event type, event occurrence location, i.e., component identifier, action duration / rate parameter, current initial operating condition, and related equipment attributes. With the parameter vector, the scenario retrieval module performs similarity retrieval in the scenario pre-computation library, and determines a number of most similar pre-computed samples based on the distance measure of the parameter vector and the sample metadata. The retrieval strategy can combine parameter space distance weighting, sample feature similarity evaluation, such as peak pressure, waveform energy, and initial operating condition matching degree, for comprehensive scoring to select the sample set that best represents the current event. After retrieving the samples, the system converts the offline samples into time series responses closer to the current parameters using interpolation or modal coefficient reconstruction, etc. Illustratively, linear interpolation based on sample weights, radial basis function (RBF) based parameter space interpolation, or weighting interpolation reconstruction of modal coefficients in the modal space of PCA / SVD compressed samples can be used. After interpolation reconstruction, necessary post-processing is performed to ensure numerical and physical reasonableness, such as smoothing filtering of the reconstructed time series to remove interpolation noise, and forced satisfaction of node flow conservation constraints, which can be projected back to the conservation subspace by local correction or projection methods, and the reconstruction error estimate based on sample interpolation residual or historical reconstruction error statistics is calculated.

[0094] In one embodiment, as shown in Figure 3 the scenario engine and the scenario trigger library are called according to the teaching objectives to perform matching and linkage rule processing on the augmented view to obtain an event scenario, including:

[0095] S301, encode the fire protection rules and sprinkler control logic into an event-driven finite state machine or rule engine to obtain a scenario trigger library; the scenario trigger library takes the fire protection rules, detector-sprinkler linkage relationships, and pump control rules as retrievable trigger items.

[0096] Illustratively, the scenario trigger library is organized in the form of event-driven finite state machine (FSM) or rule engine, whose entries describe triggering conditions, triggering actions, action priorities, constraint conditions, and optional delay / hysteresis parameters. The triggering entries in the scenario trigger library include typical fire protection rules, detector-sprinkler linkage, pump control logic, alarm linkage, and scenario templates related to teaching objectives. Specifically, the scenario trigger library adopts a structured rule description language or rule templates in JSON / YAML, each rule contains at least rule ID, rule category, triggering condition, list of linkage control commands to be activated, priority, delay / hysteresis parameters, effective condition, safety constraints. Further, for linkage logic that needs to be advanced in stages, such as multi-stage process of alarm confirmation-sprinkler delay-pump start-valve isolation, it is modeled as a FSM. The states of the FSM include initial state, detection state, confirmation state, release state, reset state, etc., the transition between states is triggered by events, and action commands or sub-rules can be output or called at each state transition.

[0097] S302, retrieve linkage rules matching the teaching objective from the scenario trigger library through the scenario engine, and generate an event scenario based on the matched linkage rules; the event scenario contains linkage control commands to be activated for the reduced-order simulation model and the running control; the teaching objective corresponds to a fire source event or a sensor signal.

[0098] Illustratively, the teacher expresses the instructional objectives as tags or semantic descriptions through the UI or a pre-planned lesson plan. The scenario engine maintains an index of instructional objectives to trigger groups, which can be stored in the rule base as metadata and support the teacher's custom extensions. The scenario engine continuously listens to multi-source inputs, such as real sensor signal streams, teacher's manual trigger commands, student operation events, and scheduler events. These inputs are pre-processed and formed into uniform event description objects, containing event type, parameter vector, source, and timestamp. The scenario engine retrieves matching rules in the scenario trigger library based on instructional objective tags, current event type, and real-time parameters. The matching adopts a multi-dimensional scoring mechanism, considering the semantic similarity of rules and instructional objectives, the parameter consistency of trigger conditions and input events, rule priority, rule current availability, and historical usage preferences. The scoring results are used to select a number of candidate rules and assign them matching weights. When multiple rules are related or complementary, the scenario engine combines the selected rule set and generates an encapsulation, i.e. an event scenario. The combination process needs to handle the dependency and conflict between rules, such as ensuring the timing constraints if rule A needs to be executed after rule B, or selectively merging or replacing actions according to priority or teacher-specified strategies if there are mutually exclusive actions between rules. The event scenario contains a sequence of linked control commands to be activated, including the parameters, expected execution time, delay, priority, and safety constraints of each command, and with metadata, i.e. source rule ID, matching confidence, instructional objective reference. When generating the event scenario, the scenario engine binds the real-time input parameters to the parameter slots in the trigger to form executable specific commands. For example, the trigger template may define to turn on the sprinkler in area X, and in actual generation, X is replaced by the specific area ID of the current trigger event. If the trigger contains configurable parameters, the scenario engine will automatically calculate or request the teacher to confirm these parameters based on the instructional objectives and system constraints.

[0099] In one of the embodiments, the method further comprises:

[0100] S41, collect the perspective and conversation information of each student AR client, and associate the perspective and conversation information with the teacher AR client.

[0101] The perspective and session information of each student AR client refers to a set of metadata collected and reported by the student terminal device in real time, which usually includes but is not limited to camera pose, current rendering frustum parameters, local timestamp, current loaded AR resource package version number, local rendering state, local cache of interaction command stream, network connection quality indicators and client identification information. The association with the teacher AR client refers to establishing a session management entry on the server or teacher master node, binding the perspective and session metadata reported by the student terminal to the corresponding student session record, so that the teacher can view, manage and uniformly control the perspective and state of the students if necessary. Specifically, the student AR client registers its session information to the teacher master node or cloud session management service with a session registration request when establishing a classroom session or joining a teacher session. The registration information includes student identification, device capability description, first frame pose and preferred rendering quality level. After successful registration, the student terminal reports the perspective and session information in a heartbeat or event-driven manner according to the preset strategy. The heartbeat period should be self-adaptive based on network conditions and classroom needs, and key operations should be reported immediately. The session management service associates the received data with the session record of the teacher AR client, and presents the student list, current pose thumbnail, network status and registration confidence in the management interface of the teacher terminal, so as to facilitate the teacher to quickly locate the students who need to be paid attention to.

[0102] S42, taking the teacher AR client as the master node, broadcasting the state difference to each student AR client according to the master-slave synchronization architecture, and enabling each student AR client to receive and render the synchronization state stream published by the teacher master node based on the session association management authority.

[0103] Illustratively, at the start of a classroom session, the teacher AR client is designated as the master node or connects to the teacher master node service in the cloud to take the master authority. The teacher master node maintains a global state snapshot that includes the current simulation master state, augmented view display settings, event scene metadata, and classroom control parameters. When the teacher makes an operation or the simulation produces a new state, the teacher master node computes the global state difference, i.e., which fields have changed and the changed values relative to the last broadcasted state, and serializes the difference into a lightweight message format to reduce network burden. The state difference message should include the difference sequence number, timestamp, changed content, trigger source, and authority tag. The teacher master node broadcasts the state difference to all associated student AR clients; the students apply the difference in order according to the local time correction information and sequence number with the teacher, and request a frame or snapshot resynchronization from the teacher master node if sequence loss or disorder is detected. To improve real-time performance and reduce latency, the students can use interpolation to display a smooth transition locally after receiving the difference, and then correct the residual error after receiving the subsequent difference. Illustratively, assuming that the teacher triggers a fire demonstration event in the classroom, the teacher master node generates a set of state differences, such as setting the fire source position, triggering the scene engine, and starting the initial parameters of the sprinkler animation, and the differences are broadcast to all student ends and immediately rendered by the students in an interpolated manner.

[0104] It should be understood that although the steps in the flowcharts involved in the embodiments described above are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the embodiments described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be alternately or alternately executed with at least some of the other steps or steps or stages in other steps.

[0105] Based on the same inventive concept, the embodiments of the present application also provide an AR technology-based building water supply and drainage system teaching demonstration device for implementing the above-mentioned AR technology-based building water supply and drainage system teaching demonstration method. The problem-solving implementation scheme provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more AR technology-based building water supply and drainage system teaching demonstration device embodiments provided below can be referred to the limitations of the AR technology-based building water supply and drainage system teaching demonstration method described above, which will not be repeated here.

[0106] In one exemplary embodiment, as Figure 4As shown, a building water supply and drainage system teaching demonstration device based on AR technology is provided, comprising:

[0107] The registration module 401 acquires the AR client camera image of the teaching site, the teaching site plan and the reference coordinates, and performs mixed space registration processing according to the AR client camera image, the teaching site plan and the reference coordinates to obtain a space transformation matrix; the space transformation matrix is used to establish a mapping relationship between the virtual model coordinate system and the real scene coordinate system;

[0108] The scenario view module 402 is used to convert the three-dimensional water supply and drainage model and the visualization resources in the AR resource package into an augmented view superimposed on the teaching site according to the space transformation matrix based on the pre-constructed AR resource package, and to obtain an event scenario by calling the scenario engine and the scenario trigger library to perform matching and linkage rule processing on the augmented view according to the teaching target; the event scenario includes a triggerable linkage control command;

[0109] The interaction module 403 is used to convert the operation event of the user on the event scenario into a scene control command to obtain an interaction command stream in response to acquiring the operation event;

[0110] The simulation solving module 404 is used to perform hydraulic state solving according to the interaction command stream and the linkage control command based on the pre-constructed reduced-order simulation model and the scenario pre-computation library to obtain a simulation state stream; the simulation state stream includes a node pressure vector, a pipe segment flow vector and device state information;

[0111] The evaluation module 405 is used to extract a key performance indicator after aligning the interaction command stream and the simulation state stream according to a time axis based on a pre-set scoring rule to obtain a learning evaluation report and provide an operation playback.

[0112] In one of the embodiments, the registration module 401 is further used to:

[0113] Perform visual synchronization positioning and map construction in sequence according to the AR client camera image to obtain a local coordinate system and a feature point cloud;

[0114] Perform drawing alignment processing on the teaching site plan based on the reference coordinates and the local coordinate system, and extract key features from the teaching site plan to obtain feature reference points;

[0115] Match the feature reference points with the site features in the AR client camera image, and estimate a global transformation matrix;

[0116] When the automatic matching accuracy is insufficient, set identifiable markers for the AR client camera image and the teaching site plan, and correct the global transformation matrix through marker recognition to obtain the space transformation matrix.

[0117] In one of the embodiments, an AR resource construction module is further included for:

[0118] obtaining an architectural design file and architectural water supply and drainage design parameters;

[0119] performing semantic extraction according to the architectural design file and the architectural water supply and drainage design parameters to obtain semantic information, and establishing a hydraulic topology graph and corresponding 3D geometric data based on the semantic information; the hydraulic topology graph is a structure of component nodes and component connection edges, and each component node saves a corresponding 3D geometric data node attribute; the semantic information includes drainage components and attributes thereof; the drainage components include pipes, nodes, valves, pumps, and sprinklers;

[0120] generating visual resources for AR rendering based on the hydraulic topology graph and the 3D geometric data and establishing a component attribute library to obtain an AR resource package.

[0121] In one of the embodiments, a simulation scenario construction module is further included for:

[0122] mapping the hydraulic topology graph into an engineering hydraulic equation set based on an engineering fluid mechanics conservation equation, and performing order reduction approximation on a flow of each pipe section according to the engineering hydraulic equation set to obtain a reduced-order simulation model; the reduced-order simulation model maintains a node flow conservation and an energy approximation relationship; the engineering fluid mechanics conservation equation includes flow conservation and energy conservation;

[0123] performing high-fidelity transient simulation to calculate time series samples under each scenario for interactive operations and fault scenarios required by a teaching goal, and storing the time series samples into a scenario pre-computation library according to a parameterized index.

[0124] In one of the embodiments, the simulation solving module 404 is further configured to:

[0125] when an operation event corresponding to the interactive command stream is a steady-state change, calling the reduced-order simulation model to perform steady-state engineering hydraulic equation solving to obtain a simulation state stream;

[0126] when an operation event corresponding to the interactive command stream is a transient event, searching a pre-computed time series sample matching a parameter of the operation event in the scenario pre-computation library, and reconstructing an approximate transient response through sample interpolation to obtain a simulation state stream.

[0127] In one of the embodiments, the scenario view module 402 is further configured to:

[0128] encoding a fire prevention rule and a sprinkler control logic into an event-driven finite state machine or a rule engine to obtain a scenario trigger library; the scenario trigger library takes the fire prevention rule, a detector-sprinkler linkage relationship, and a pump control rule as retrievable trigger items;

[0129] The scenario engine retrieves the linkage rules that match the teaching objectives from the scenario trigger library and generates event scenarios based on the matched linkage rules. The event scenarios include linkage control commands to be activated for the reduced-order simulation model and operation control. The teaching objectives correspond to fire source events or sensor signals.

[0130] In one embodiment, a scenario synchronization module is also included, for:

[0131] Collect the viewpoint and conversation information of each student's AR client, and associate the viewpoint and conversation information with the teacher's AR client;

[0132] With the teacher's AR client as the master node, the system broadcasts state differences to each student's AR client according to the master-slave synchronization architecture, and enables each student's AR client to receive and render the synchronization state stream published by the teacher's master node based on session association management permissions.

[0133] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps in the above method embodiments.

[0134] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0135] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0136] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A teaching demonstration method for building water supply and drainage systems based on AR technology, characterized in that, The method includes: The system acquires AR client camera images, a floor plan of the teaching site, and reference coordinates. It then performs hybrid spatial registration processing based on these elements to obtain a spatial transformation matrix. This spatial transformation matrix is ​​used to establish a mapping relationship between the virtual model coordinate system and the real-world scene coordinate system. Based on a pre-built AR resource package, the 3D water supply and drainage model and visualization resources in the AR resource package are transformed into an augmented view superimposed on the teaching site according to the spatial transformation matrix. Then, according to the teaching objectives, the scenario engine and scenario trigger library are called to perform matching and linkage rule processing on the augmented view to obtain an event scene. The event scene includes triggerable linkage control commands. In response to obtaining user operation events in the event scenario, the operation events are converted into scene control commands to obtain an interactive command stream; Based on the pre-built reduced-order simulation model and scenario pre-calculation library, the hydraulic state is solved according to the interactive command flow and the linkage control command to obtain the simulation state flow; the simulation state flow includes node pressure vector, pipe segment flow vector and equipment state information; Based on preset scoring rules, the interactive command stream and the simulation state stream are aligned along the time axis to extract key performance indicators, obtain a learning evaluation report, and provide operation playback.

2. The method according to claim 1, characterized in that, The step of performing hybrid spatial registration processing based on the AR client camera image, the teaching site plan, and reference coordinates to obtain a spatial transformation matrix includes: Based on the AR client camera images, visual synchronous positioning and map construction are performed sequentially to obtain the local coordinate system and feature point cloud; Based on the reference coordinates and the local coordinate system, the teaching site plan is aligned with the map, and key features are extracted from the teaching site plan to obtain feature reference points. The feature reference points are matched with the on-site features in the AR client camera image, and the global transformation matrix is ​​estimated. When the automatic matching accuracy is insufficient, identifiable markers are set for the AR client camera image and the teaching site plan, and the global transformation matrix is ​​corrected by identifying the markers to obtain the spatial transformation matrix.

3. The method according to claim 1, characterized in that, The AR resource package is obtained through the following method: Obtain architectural design documents and building water supply and drainage design parameters; Semantic information is obtained by extracting semantic information from the architectural design documents and the building water supply and drainage design parameters, and a hydraulic topology diagram and its corresponding 3D geometric data are established based on the semantic information. The hydraulic topology diagram is a structure of component nodes and component connection edges, and the corresponding 3D geometric data node attributes are stored in each component node. The semantic information includes drainage components and their attributes. The drainage components include pipes, nodes, valves, pumps, and sprinklers. Based on the hydraulic topology map and the 3D geometric data, a visualization resource for AR rendering is generated and a component attribute library is established to obtain the AR resource package.

4. The method according to claim 3, characterized in that, The reduced-order simulation model and scenario pre-calculation library were obtained through the following methods: Based on the engineering fluid dynamics conservation equations, the hydraulic topology diagram is mapped to a set of engineering hydraulic equations, and the flow rate of each pipe segment is approximated by a reduced order according to the set of engineering hydraulic equations to obtain a reduced-order simulation model; the reduced-order simulation model maintains the nodal flow rate conservation and energy approximation relationship; the engineering fluid dynamics conservation equations include flow rate conservation and energy conservation; For the interactive operations and fault scenarios required for the teaching objectives, high-fidelity transient simulation is used to calculate time series samples under each scenario, and the time series samples are stored in the scenario pre-calculation library according to the parameterized index.

5. The method according to claim 4, characterized in that, The simulation state flow, based on the pre-built reduced-order simulation model and scenario pre-calculation library, is obtained by solving the hydraulic state according to the interactive command flow and the linkage control command, including: When the operation event corresponding to the interactive command flow is a steady-state change, the reduced-order simulation model is invoked to solve the steady-state engineering hydraulic equations to obtain the simulation state flow. When the operation event corresponding to the interactive command flow is a transient event, a pre-calculated time series sample matching the parameters of the operation event is retrieved from the scenario pre-calculation library, and the approximate transient response is reconstructed through sample interpolation to obtain the simulation state flow.

6. The method according to claim 1, characterized in that, The process of matching and linking rules to the enhanced view based on the teaching objectives by calling the scenario engine and scenario trigger library to obtain event scenarios includes: Fire prevention rules and sprinkler control logic are encoded into an event-driven finite state machine or rule engine to obtain a scenario trigger library; the scenario trigger library uses fire prevention rules, detector-sprinkler linkage relationships, and pump control rules as searchable trigger items; The scenario engine retrieves the linkage rules that match the teaching objectives from the scenario trigger library and generates event scenarios based on the matching linkage rules; the event scenarios include linkage control commands to be activated for the reduced-order simulation model and operation control; the teaching objectives correspond to fire source events or sensor signals.

7. The method according to claim 1, characterized in that, The method further includes: Collect the perspective and conversation information of each student's AR client, and associate the perspective and conversation information with the teacher's AR client; Using the teacher's AR client as the master node, the system broadcasts state differences to each student's AR client according to the master-slave synchronization architecture, and enables each student's AR client to receive and render the synchronization state stream published by the teacher's master node based on session association management permissions.

8. A teaching demonstration device for building water supply and drainage systems based on AR technology, characterized in that, The device includes: The registration module acquires the AR client camera image, the teaching site plan, and reference coordinates of the teaching site, and performs hybrid spatial registration processing based on the AR client camera image, the teaching site plan, and the reference coordinates to obtain a spatial transformation matrix; the spatial transformation matrix is ​​used to establish a mapping relationship between the virtual model coordinate system and the real scene coordinate system. The scenario view module is used to transform the 3D water supply and drainage model and visualization resources in the pre-built AR resource package into an augmented view superimposed on the teaching site according to the spatial transformation matrix. Based on the teaching objectives, the scenario engine and scenario trigger library are called to perform matching and linkage rule processing on the augmented view to obtain event scenarios. The event scenarios include triggerable linkage control commands. The interaction module is used to respond to the user's operation events in the event scenario, convert the operation events into scene control commands, and obtain an interaction command stream; The simulation solution module is used to solve the hydraulic state based on the pre-built reduced-order simulation model and scenario pre-calculation library, according to the interactive command flow and the linkage control command, to obtain the simulation state flow; the simulation state flow includes node pressure vector, pipe segment flow vector and equipment state information; The evaluation module is used to extract key performance indicators by aligning the interactive command stream and the simulation state stream along the time axis based on preset scoring rules, thereby obtaining a learning evaluation report and providing operation playback.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images