A property inspection hidden danger intelligent identification and closed-loop management method and system

Abstract

The application discloses a kind of property inspection hidden danger intelligent identification and closed-loop management method and system, specifically related to property inspection and hidden danger management technical field, method includes: construct three-dimensional digital twin bottom plate and component knowledge graph;With the intelligent terminal of integrated visible light, near infrared and structured light sensor, through space-time joint calibration and global synchronous trigger, visible light image, moisture content thermal map and three-dimensional deformation vector field are synchronously collected;Data is uploaded to edge node, and the deformation field is calculated with the rotation speed and divergence, combined with moisture content distribution to infer physical cause;The diagnostic conclusion is mapped to digital twin, and the multi-physical field simulation deduction coupled with Richards equation is run to confirm the effective maintenance scheme;Generate parameter constraint self-adapting closed-loop work order;After maintenance, rigid reinspection is carried out through multimodal terminal.The application realizes the physical level accurate identification, cause and effect inference and closed-loop eradication of hidden danger, and significantly improves the intellectualization and long-term effectiveness of property inspection.

Description

Technical Field

[0001] This invention relates to the field of property inspection and hazard management technology, and more specifically, to a method and system for intelligent identification and closed-loop management of property inspection hazards. Background Technology

[0002] Traditional property inspections primarily rely on manual visual inspection and point-to-point measurements using simple handheld instruments (such as hygrometers and crack gauges), which has significant limitations. First, manual inspections are constrained by experience differences and subjective judgment, making it difficult to detect early-stage micro-leakage, sub-millimeter deformation, and other hidden hazards. Furthermore, inspection efficiency is low, records are inconsistent, and hazard information is difficult to trace and share. Second, existing inspection equipment is mostly single-sensor (such as visible light cameras or infrared thermal imagers), lacking the ability to simultaneously acquire multimodal data and spatially register it. This makes it impossible to accurately quantify the physical correlation between moisture content distribution and structural deformation, and to infer the true physical cause of hazards (such as expansion due to leakage or shearing due to settlement) from the perspective of the co-evolution of mechanical and moisture fields. Finally, current property management systems generally remain in a linear "discovery-reporting-repair" process. Repair plans rely on experience and lack simulation verification based on digital twins and adaptive optimization of construction parameters. Post-repair, there is a lack of rigid physical indicator re-verification and long-term continuous monitoring, leading to repeated occurrences and repairs of the same hazards, creating a vicious cycle of "treating the symptoms but not the root cause."

[0003] To address the aforementioned issues, some studies have attempted to introduce drone inspections, image recognition, or building information models (BIM) for information management. However, key shortcomings remain: First, image recognition can only handle surface defects and cannot perceive internal moisture fields and three-dimensional deformation fields; second, BIM is primarily used in the design and construction phases, lacking dynamic mapping with real-time multimodal sensing data and physical simulation-driven capabilities; third, the hazard identification results lack explanations at the physical cause level, leading to insufficiently targeted maintenance solutions; and fourth, a complete closed-loop mechanism is lacking, encompassing diagnosis, simulation, work order generation, rigid verification, and knowledge self-evolution. Therefore, achieving accurate physical-level identification, causal inference, solution verification, and closed-loop management of hazards in property inspections has become a pressing technical problem to be solved in this field. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for intelligent identification and closed-loop management of potential hazards during property inspections.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for intelligent identification and closed-loop management of potential hazards during property inspections includes the following steps:

[0007] Construct a three-dimensional digital twin base for the property building and create a knowledge graph for each component;

[0008] Using an intelligent terminal that integrates a visible light camera, a near-infrared spectral sensor, and a structured light micro-deformation sensing module, multimodal synchronous acquisition is performed on the target component to obtain visible light images, moisture content thermograms, and three-dimensional deformation vector fields.

[0009] The collected data is uploaded to the edge computing node to calculate the curl and divergence of the deformation vector field, and the physical causes of the hidden dangers are inferred by combining the moisture content thermogram.

[0010] The diagnostic conclusions are mapped to a digital twin, and effective repair solutions are confirmed through multiphysics simulation.

[0011] An adaptive closed-loop work order is generated based on the simulation results, wherein the parameter constraints of the operation steps are dynamically set by the collision detection of the digital twin.

[0012] After maintenance, the multimodal terminal is used for re-verification, and the results are rigidly compared with the physical verification standards in the work order. Once the standards are met, the work order is closed.

[0013] Deploy wireless sensors for long-term monitoring and feed back the successfully closed-loop mapping relationship to the knowledge graph to achieve self-evolution of diagnostic capabilities.

[0014] Specifically, multi-sensor spatiotemporal joint calibration is performed before the multimodal synchronous acquisition:

[0015] Using a three-modal common calibration target to capture images from multiple directions, the intrinsic parameter matrix of the visible light camera and the intrinsic parameter matrix of the near-infrared imager are calculated separately, and the structural parameters of the structured light module are optimized.

[0016] Using the structured light receiver coordinate system as the global reference, solve for the rigid body transformation matrix from the visible light camera and near-infrared imaging coordinate systems to the global coordinate system;

[0017] Configure a global synchronization pulse generator to lock the trigger signals of the three sensors to the same clock edge.

[0018] Specifically, the physical causes of the inferred potential hazards include:

[0019] Connectivity analysis of the divergence field is performed to extract regions with central radial features. If the spatial intersection ratio of this region with the high value region of the water content thermogram exceeds the preset value and the mean curl is less than 20% of the mean divergence, it is determined that the expansion is caused by internal water leakage.

[0020] If the boundary of the high water content area is sharp and extends along a specific line that matches the water supply pipeline path, it is determined to be a pipeline rupture and leakage;

[0021] If the high water content area is distributed in a diffuse, patchy manner and is adjacent to the outer wall, it is determined to be rainwater intrusion.

[0022] Specifically, the process of matching the water supply pipeline path is as follows:

[0023] The pipeline centerline in the digital twin model is retrieved and discretized into a three-dimensional path point sequence. The boundary skeleton points of the high water content area are extracted from the water content heat map as a two-dimensional path point sequence. An improved discrete Fréchet distance is constructed as the matching cost function and a tangent direction angle penalty factor is introduced. When the matching cost is less than a preset threshold and the water content of the area exceeds 1.5 times the global average water content, it is marked as a leakage source.

[0024] Specifically, the core of the multiphysics simulation is:

[0025] The extended Richards equations were run, coupling the pipeline leakage source term and the dominant flow channel of the wall cracks. The unsaturated hydraulic conductivity function introduced a wall damage correction term characterized by the normalized value of the divergence field. Simulations were performed with source term parameters set for different repair schemes to confirm the only effective scheme.

[0026] Specifically, the adaptive generation logic for the parameter constraints of the operation steps is as follows:

[0027] Extract the boundary of the high moisture content area and transform it to the engineering coordinate system to generate a three-dimensional bounding box as the calibration range;

[0028] The maximum allowable percentage of the area exceeding the calibrated range is dynamically set after collision detection between the hidden engineering model in the digital twin and the bounding box. If the extended boundary touches the reinforcing bars or pipelines, the percentage is automatically reduced until a safe breach boundary is generated.

[0029] Specifically, the rigidity verification method is as follows:

[0030] After the repair is completed, the multimodal terminal is used again to scan the repair area synchronously. The obtained post-repair moisture content thermal map and three-dimensional deformation vector field are compared with the preset physical verification standards in the work order item by item. The work order is only allowed to be closed when all physical indicators fall within the standard threshold range; otherwise, the non-compliant items are indicated and rework items are generated.

[0031] Specifically, the long-term monitoring and knowledge self-evolution are implemented as follows:

[0032] After a work order is closed, the wireless sensor continues to transmit data back to the digital twin. If an upward trend in moisture content or an abnormal rate of microdeformation is detected, an early warning will be automatically triggered and historical work orders will be traced back. At the same time, the mapping relationship between physical characteristics, causes, and effective intervention measures for each successful closed loop will be integrated into the knowledge graph, and the weight of the physical cause logic chain will be updated.

[0033] A property inspection hazard identification and closed-loop management system includes:

[0034] The multimodal synchronous acquisition and digital twin base plate construction module is used to build a full-element building information model and component knowledge graph, and integrates visible light, near infrared and structured light sensors. After spatiotemporal joint calibration and global synchronous triggering, it acquires visible light images, moisture content thermal maps and three-dimensional deformation vector fields of components.

[0035] The physical feature clustering and intelligent inference module for causes runs on edge computing nodes. It is used to calculate the curl and divergence of the deformation vector field. Combined with the gradient distribution and spatial overlap of the moisture content thermogram, it automatically identifies the physical causes of hidden dangers and outputs diagnostic conclusions.

[0036] The twin simulation and adaptive work order generation module is used to map diagnostic conclusions to a digital twin, run multiphysics simulation to verify the maintenance plan, and automatically generate a closed-loop work order containing a diagnostic report, operation prescription, and physical verification standards. The operation parameter constraints are adaptively set by the twin model collision detection.

[0037] The rigid verification and long-cycle knowledge self-evolution module is used to guide operations through augmented reality glasses during maintenance. After maintenance, a multimodal terminal is used for verification and rigid comparison with physical verification standards. Once the standards are met, the work order is closed. At the same time, the status of components is continuously monitored through wireless sensors, and the mapping relationship of successful closed loop is fed back to the knowledge graph to realize the self-evolution of diagnostic capabilities.

[0038] The technical effects and advantages of this invention are as follows:

[0039] This invention constructs a comprehensive digital twin base and component knowledge graph, and utilizes spatiotemporal joint calibration and global synchronous triggering of visible light, near-infrared, and structured light three-modal sensors to achieve high-precision synchronous acquisition of macroscopic defects, moisture content distribution, and sub-millimeter-level three-dimensional deformation fields of property components. Based on this, by calculating the curl and divergence of the deformation vector field, and combining the gradient characteristics and spatial overlap analysis of the moisture content thermogram, the causes of potential hazards (such as leakage-induced expansion, pipe rupture, and rainwater intrusion) can be accurately inferred from the physical characteristics level, overcoming the shortcomings of traditional inspections that rely solely on appearance, leading to unclear causes and blind repairs. Simultaneously, mapping the diagnostic conclusions to the digital twin and running the Richards equation with coupled damage correction terms for multiphysics simulation can pre-verify the effectiveness of different repair schemes, avoiding ineffective repairs.

[0040] This invention further automates and intelligentizes the entire process from diagnosis to closed-loop management. Based on effective solutions confirmed by simulation, the system automatically generates adaptive work orders containing diagnostic reports, operational prescriptions, and physical verification standards. Operational parameter constraints (such as the scope of breach) are dynamically set by twin model collision detection, improving construction safety and adaptability. During maintenance, AR glasses guide and automatically record key parameters. After maintenance, a multimodal terminal performs rigid verification; the work order is only closed when all physical indicators meet the standards, eliminating residual hazards. Furthermore, on-site deployed wireless sensors continuously monitor for long periods and feed back the mapping relationship between physical characteristics, causes, and intervention measures of successfully closed-loop systems to a knowledge graph, enabling self-evolution of diagnostic capabilities. Through this rigorous closed-loop system—from physical perception, causal inference, simulation verification to rigid verification and continuous monitoring—the recurring problem of hidden dangers is completely eliminated, significantly improving the accuracy, scientific rigor, and long-term effectiveness of property inspections. Attached Figure Description

[0041] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] The technical solutions of 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.

[0043] like Figure 1 As shown, the steps of a method for intelligent identification and closed-loop management of potential hazards during property inspections are as follows:

[0044] Step 1: Construct a knowledge graph of the 3D digital twin base plate and components:

[0045] A high-precision 3D laser scan of the building is performed to construct a comprehensive building information model, or digital twin, encompassing structural columns, shear walls, decorative layers, and water supply and drainage pipelines. On this model, each inspectable building component is assigned a unique identifier, and a component knowledge graph is established. The knowledge graph defines the material properties, physical characteristics, common hazard categories, and corresponding physical cause logic chains for each component. For example, for a "non-load-bearing partition wall," the physical cause logic chain includes possible paths such as thermal expansion and contraction, keel corrosion, wall dampness, and excessive load.

[0046] Step 2: Perform multimodal synchronous inspection and data collection:

[0047] Inspection personnel use a smart terminal that integrates a visible light camera, a near-infrared spectral sensor, and a structured light three-dimensional micro-deformation sensing module to scan the target component.

[0048] Perform the multi-sensor spatiotemporal joint calibration process:

[0049] The operator places a specially designed tri-modal common calibration target at the terminal's working distance. The target's surface is simultaneously printed with a checkerboard pattern, a standard diffuse reflection region with known reflectivity, and embossed geometric steps. The terminal sequentially photographs the calibration target from at least nine different angles, simultaneously triggering the acquisition of three data streams. The Zhang Zhengyou calibration method is then used to calculate the intrinsic parameter matrix of the visible light camera. Intrinsic parameter matrix of near-infrared spectral imager For the structured light module, the structural parameters between the projector and receiver are optimized by using the known mapping between the step height and deformation phase. Then, using the structured light receiver coordinate system as the global reference coordinate system, the rigid body transformation matrix from the visible light camera coordinate system to the global coordinate system is solved. And the rigid body transformation matrix from the near-infrared imaging coordinate system to the global coordinate system. ,in Let be the rigid body transformation matrix from the visible light camera coordinate system to the structured light global reference coordinate system. It is a 3×3 rotation matrix that describes the pose of the camera coordinate system relative to the global coordinate system. It is a 3×1 translation vector describing the position of the camera coordinate system origin relative to the global coordinate system; This represents the rigid body transformation matrix from the infrared imaging coordinate system to the structured light global reference coordinate system. It is a 3×3 rotation matrix that describes the attitude of the near-infrared coordinate system relative to the global coordinate system. It is a 3×1 translation vector that describes the position of the origin of the near-infrared coordinate system relative to the global coordinate system.

[0050] The transformation relationship satisfies: for any point in space, its three-dimensional non-homogeneous coordinates in the structured light global coordinate system are... Homogeneous coordinates of the corresponding pixel in the visible light image There is a mapping relationship between them. ,in Let be the three-dimensional non-homogeneous coordinates of a point in the structured light global coordinate system. Pixels obtained through stereo matching The depth value (scalar) at that location. Visible light camera intrinsic parameter matrix The inverse matrix, These are the homogeneous coordinates of a pixel in a visible light image;

[0051] Similarly, near-infrared image pixels Mapping to global coordinates is ,in Pixels acquired via the rangefinding or confocal system of a near-infrared sensor The depth value (scalar) at that location. Intrinsic parameter matrix of near-infrared imager The inverse matrix, These are the homogeneous coordinates of the pixels in the near-infrared image.

[0052] Meanwhile, at the terminal main control hardware level, a global synchronization pulse generator is configured to lock the frame trigger signal of the visible light camera, the line scan synchronization signal of the near-infrared spectral sensor, and the stripe switching signal of the structured light projector to the same clock edge, ensuring that any set of three acquired data channels are stamped with the same timestamp, with the maximum time deviation. Strictly constrained to within 1 microsecond, eliminating spatial misalignment caused by rolling shutter or asynchronous triggering. After completing the above calibration and synchronization, the terminal will... The data is stored as global parameters for this inspection session. All subsequent synchronously collected data from the three channels are registered in real time to the same global coordinate system through the corresponding transformation matrix, providing a physically accurate data foundation for the pixel-by-pixel spatial overlap calculation of the moisture content thermogram and the deformation divergence field in step 3.

[0053] The terminal simultaneously acquires three data streams: visible light images, used to record macroscopic visual defects; near-infrared spectral data, which analyzes the absorption characteristics of moisture in specific wavelength bands to quantify the moisture content distribution on the surface and shallow layers of the component and generate a moisture content thermogram; and micro-deformation sensing data, which obtains a high-density three-dimensional deformation vector field of the scanned area by projecting structured light onto the component surface and solving the phase distortion, accurately calculating the misalignment of the crack wings, the flatness deviation of the wall surface, and the height field distribution of the bulge, with deformation measurement accuracy reaching sub-millimeter level.

[0054] Step 3: Causal inference based on physical feature clustering:

[0055] The intelligent terminal uploads the collected moisture content thermogram and three-dimensional deformation vector field to the edge computing node in real time. The edge node runs a physical origin inference algorithm, performing the following operations: First, it calculates the curl and divergence of the three-dimensional deformation vector field. If the curl is much greater than the divergence, it indicates that the deformation is mainly shear displacement, which may correspond to structural settlement; if the divergence is significant and radially distributed from the center, and coincides with the high-value area of ​​the moisture content thermogram, it is determined to be material expansion caused by internal water leakage.

[0056] The specific execution process of the physical cause inference algorithm is as follows:

[0057] The raw 3D point cloud data acquired by the structured light module is preprocessed. The moving least squares method is used to smooth the discrete point cloud and reconstruct a continuous surface, obtaining a differentiable deformation function. A regular mesh is constructed on this continuous surface. Based on the finite difference method, the spatial derivative of the deformation vector is calculated for each mesh node, thereby generating discrete curl field scalar maps and divergence field scalar maps. Connectivity analysis is performed on the divergence field to extract a first region with positive divergence values ​​and absolute values ​​exceeding a preset threshold. The geometric center and boundary of this first region are calculated. Simultaneously, the... If the roundness value of a region is greater than 0.85 and the divergence value within the region monotonically decreases from the center to the boundary, then the region is determined to conform to the central radial pattern. The divergence anomaly region with the central radial pattern is spatially cross-union ratio (CUNR) calculated with the water content thermogram in the same coordinate system. If the CUNR exceeds a preset 50% and the mean curl value of the divergence region is less than 20% of the mean divergence value, then the quantification condition that the curl is much smaller than the divergence is met. The system finally outputs the diagnostic conclusion of internal water source leakage causing swelling and the corresponding CUNR value as the confidence level.

[0058] Secondly, the gradient distribution characteristics of the moisture content heatmap are extracted. If the boundary of the high moisture content area is sharp and extends along a specific linear shape, matching the water supply pipeline path, it is determined to be a pipeline rupture and leakage.

[0059] The specific process for determining the matching water supply pipeline path is as follows:

[0060] First, the edge computing node retrieves the digital twin model bound to the component, extracts the centerlines of all water supply pipelines passing through the current abnormal water content area, and discretizes the centerline of each pipeline into an ordered sequence of three-dimensional path points. ,in For the first An ordered three-dimensional path point sequence obtained by discretizing the centerline of a water supply pipeline. This represents the total number of points in the sequence. For the pipeline in the first The three-dimensional coordinates of each sampling point.

[0061] Simultaneously, adaptive threshold segmentation and Canny edge detection are performed on the moisture content heatmap to extract the skeleton line segments of the high moisture content region boundary. The center coordinates of the skeleton pixels are used to construct another set of ordered two-dimensional path point sequences. ;in This is an ordered two-dimensional path point sequence composed of the center points of the skeleton pixels of the high water content region boundary in the water content heatmap. The length of the skeleton point sequence. For the first in the sequence Two-dimensional coordinates (pixel coordinate system) of a point.

[0062] To quantify the similarity of their geometric shapes, an improved discrete Fréchet distance is constructed as the matching cost function: Let the pipeline path point sequence be... With skeleton sequence The matching cost between them is ;in, Pipeline path point sequence With skeleton point sequence The matching cost between them For matching pairs The total number of pairs included. To satisfy the set of monotonic path matching pairs with corresponding first and last points, express The Point and The Point pairing; ||·|2 represents the Euclidean distance; For pipelines in The tangent direction at that point is in relation to the skeleton. The angle between the tangent directions at a point is used to penalize inconsistencies in direction. As the angle penalty factor, take , This represents the physical scale corresponding to the image pixels.

[0063] If and only if the matching cost of a certain pipeline The system marks a pipeline as successfully matched only when the water content of the matched section is less than the preset pipeline matching threshold and exceeds 1.5 times the global average water content. The system then outputs the physical cause of the pipeline rupture and leakage, along with the matched pipeline number. If multiple pipelines are successfully matched, the pipeline with the lowest matching cost is selected as the source of the leakage.

[0064] If the high water content area is diffusely distributed in patches and adjacent to the exterior wall, it is determined to be rainwater intrusion. The algorithm finally outputs the hazard diagnosis conclusion for the component, including the category of physical cause and the confidence level. For example, the diagnosis conclusion is "Blistering of the wall putty layer caused by micro-leakage of water supply pipe, confidence level 95%".

[0065] Step 4: Simulation and deduction of maintenance solutions in a digital twin environment:

[0066] The diagnostic conclusions and quantified physical characteristic parameters from step 3 are mapped onto the digital twin of the component, generating a hidden danger twin model with physical attributes. The system matches candidate repair solutions for the physical cause from a pre-set intervention measure library. For example, for "micro-leakage in the pipeline," solution A is to simply remove and redo the wall putty, while solution B is to repair the pipeline, dry the wall, and then redo the wall. In the twin environment, multi-physics simulations are performed on solutions A and B. Based on parameters such as material porosity and moisture diffusion coefficient, the simulation engine simulates the movement of moisture inside the wall after solution A is implemented within the next 72 hours, predicting that the high internal moisture content will continue to rise and cause new bulges. For solution B, the simulation first simulates that after closing the upstream valve and repairing the pipeline, the wall moisture content decreases exponentially over time. Once it drops to a safe threshold, the putty layer is applied, and the stress distribution is uniform and without abnormalities. The simulation results are presented as a visual dynamic cloud map, directly rejecting solution A and confirming solution B as the only effective solution.

[0067] The core of multiphysics simulation is to run an extended Richards equation, which specifically couples the concentrated moisture source term formed by the pipe rupture and the dominant flow channel formed by the wall crack. Its specific form is as follows: ,in, For divergence operators (3D). To account for the unsaturated hydraulic conductivity caused by wall damage, For the total head gradient, This refers to the volumetric moisture content of the wall material. For time, For pressure head, It is a gravitational potential. For pipeline leakage sources, the three-dimensional coordinates of the leakage point and the estimated leakage rate given in the diagnostic conclusion of step 3 are directly used for setting;

[0068] Key parameters To account for unsaturated hydraulic conductivity due to wall damage: In the formula, It is a natural exponential function. The saturated hydraulic conductivity of the material, and To fit the shape parameters, This represents the saturated moisture content. For damage correction terms, The normalized value of the divergence field calculated in step 3 represents the degree of expansion damage to the wall at that location. This is an empirical scaling factor;

[0069] The simulation of scheme A is to set up The term remains positive, and the evolution of the moisture content field is observed; the simulation of scheme B first... Set the value to zero, simulate the repair, and run the above equations again until the average moisture content of the wall drops below the preset threshold. Finally, load the mechanical constitutive model of the putty layer, calculate the shrinkage stress field caused by the evaporation of residual moisture in the repair layer, and confirm that no tensile stress concentration area appears.

[0070] Step 5: Generate and issue a diagnostic-prescription adaptive work order:

[0071] The system automatically generates an adaptive closed-loop work order based on the only successful solution B derived from the deduction. This work order structure consists of three parts:

[0072] The diagnostic report, in the form of images, text, and screenshots of twin models, clearly shows the three-dimensional coordinates of the leakage point, the water content range, and the bulge deformation data, and includes an explanation of the physical causes.

[0073] The operational prescription breaks down Plan B into ordered, parameter-constrained operational steps. For example, Step 1: Close the water supply valve on the 5th floor of the west unit of Building 3 and confirm that the pressure gauge is zero; Step 2: Break through the wall 1.2 meters above the ground in the horizontal coordinate range A3-B4 to expose the damaged pipe. The area to be broken must not exceed 20% of the calibrated range; Step 3: Replace the pipe with a 30 cm long, DN20 PPR pipe section, heat-fused it, and after cooling, conduct a 0.8 MPa water pressure test, holding the pressure for 15 minutes without leakage.

[0074] In the prescription operation, the adaptive generation logic for specific parameter constraints is as follows:

[0075] First, the boundary of the high moisture content area output in step 3 is extracted, and through the mapping relationship, it is transformed from the sensor coordinate system to the engineering coordinate system of the building component, and a three-dimensional bounding box is automatically generated as the calibration range.

[0076] The constraint that the demolition area must not exceed 20% of the specified range is not preset, but a safety margin dynamically set after collision detection between the internal steel reinforcement and pipeline layout model in the digital twin and the three-dimensional bounding box. That is, the system extends outward by 20% at equal intervals along the boundary of the bounding box and intersects with the hidden engineering model embedded in the component. If the extended boundary does not touch any steel reinforcement or adjacent pipelines, the 20% is approved as an operability margin and written into the work order. If the extended boundary touches an obstacle, the system automatically reduces the percentage until a safe demolition boundary that does not damage the internal structure is generated, and the finally approved percentage value is dynamically written into the corresponding step of the operation prescription.

[0077] Physical verification standards define the quantitative physical indicators that the component must pass again after repair, such as "using a thermal imager to check, the temperature field distribution of the repaired section of the pipe and the surrounding wall surface is uniform, with no low temperature anomalies; after 72 hours, the wall surface flatness error measured by structured light is less than 1 mm".

[0078] Step 6: Maintenance Execution and Rigid Verification of Physical Indicators:

[0079] Upon arrival at the site, maintenance personnel use AR glasses to load a twin model of the work order and operational instructions. The model is precisely overlaid on the actual component, guiding the dismantling scope and operational steps. Each critical operation, such as heat fusion temperature and holding time, is automatically recorded and transmitted by sensors. If any parameter exceeds limits, the work order system immediately issues a corrective warning. After maintenance is completed, personnel use the multimodal terminal from step 2 for a re-inspection scan. The system automatically performs a rigid comparison between the post-maintenance moisture content thermogram and deformation vector field and the preset physical verification standards in the work order. Only when all physical indicators fall within the standard threshold range does the system allow the work order to enter the pending closure state; otherwise, it automatically rejects the work order and specifically identifies the non-compliant physical indicators, generating a rework item.

[0080] Step 7: Long-term digital twin monitoring and knowledge self-evolution:

[0081] After the work order is closed, the wireless humidity and deformation sensors deployed on-site will continuously transmit data back to the digital twin of the component. During long-term monitoring, if an increasing trend in moisture content or an abnormal micro-deformation rate is detected again, the system will automatically trigger an early warning and trace back historical work orders related to this component, entering a new round of diagnosis-simulation-treatment micro-cycle. Simultaneously, for each successfully closed-loop work order, the mapping relationship between "physical characteristics - causes - effective intervention measures" will be extracted and integrated into the component's knowledge graph, updating the weights of the physical cause logic chain and achieving self-evolution of the system's diagnostic capabilities. Through this rigorous closed loop from physical perception, causal inference, simulation verification to rigid re-verification and continuous monitoring, the defect of recurring hidden dangers is completely eliminated.

[0082] The following is a module of a property inspection hazard identification and closed-loop management system:

[0083] The multimodal synchronous acquisition and digital twin baseboard construction module constructs a full-element building information model (digital twin baseboard) through 3D laser scanning, and assigns a unique identity and knowledge graph to each building component. Simultaneously, it integrates a visible light camera, a near-infrared spectral sensor, and a structured light micro-deformation sensing module. Through spatiotemporal joint calibration and global synchronous triggering, it synchronously acquires visible light images, moisture content thermograms, and high-precision 3D deformation vector fields of the components.

[0084] The physical feature clustering and intelligent inference module for causes runs on edge computing nodes. It calculates the curl and divergence of the deformation vector field, combines the gradient distribution and spatial overlap of the moisture content thermogram, and uses connected domain analysis, roundness discrimination and improved discrete Fréchet distance matching to automatically identify physical causes such as leakage, expansion and shear, and outputs diagnostic conclusions and confidence levels.

[0085] The twin simulation and adaptive work order generation module maps diagnostic conclusions to a digital twin, calls the intervention measure library to generate candidate solutions, and runs a multiphysics simulation coupled with Richards equations to dynamically verify the maintenance effect. Subsequently, it automatically generates a diagnostic-prescription type closed-loop work order, in which the parameter constraints of the operation steps (such as the scope of failure) are adaptively set by the twin model collision detection, and includes quantified physical verification standards.

[0086] The rigid verification and long-cycle knowledge self-evolution module guides operations via AR glasses and transmits key parameters in real time during maintenance. After maintenance, a multimodal terminal rescans the system and rigidly compares it with the physical verification standards in the work order. The work order can only be closed after the standards are met. Simultaneously, deployed wireless sensors continuously monitor the component status. The mapping relationship between physical characteristics, causes, and intervention measures for each successful closed loop is fed back into the knowledge graph, enabling the self-evolution of diagnostic capabilities.

[0087] The above formulas are all dimensionless calculations. Dimensionless calculations can be performed using various methods such as standardization, which will not be elaborated here. The formulas are derived from software simulations based on a large amount of collected data, and the preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0088] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, ATA hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state ATA hard disk.

[0089] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0090] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0091] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0092] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0093] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0094] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable ATA hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0095] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent identification and closed-loop management of potential hazards during property inspections, characterized in that, Includes the following steps: Construct a three-dimensional digital twin base for the property building and create a knowledge graph for each component; Using an intelligent terminal that integrates a visible light camera, a near-infrared spectral sensor, and a structured light micro-deformation sensing module, multimodal synchronous acquisition is performed on the target component to obtain visible light images, moisture content thermograms, and three-dimensional deformation vector fields. The collected data is uploaded to the edge computing node to calculate the curl and divergence of the deformation vector field, and the physical causes of the hidden dangers are inferred by combining the moisture content thermogram. The diagnostic conclusions are mapped to a digital twin, and effective repair solutions are confirmed through multiphysics simulation. An adaptive closed-loop work order is generated based on the simulation results, wherein the parameter constraints of the operation steps are dynamically set by the collision detection of the digital twin. After maintenance, the multimodal terminal is used for re-verification, and the results are rigidly compared with the physical verification standards in the work order. Once the standards are met, the work order is closed. Deploy wireless sensors for long-term monitoring and feed back the successfully closed-loop mapping relationship to the knowledge graph to achieve self-evolution of diagnostic capabilities.

2. The method for intelligent identification and closed-loop management of potential hazards during property inspections according to claim 1, characterized in that, Multi-sensor spatiotemporal joint calibration is performed before the multimodal synchronous acquisition: Using a three-modal common calibration target to capture images from multiple directions, the intrinsic parameter matrix of the visible light camera and the intrinsic parameter matrix of the near-infrared imager are calculated separately, and the structural parameters of the structured light module are optimized. Using the structured light receiver coordinate system as the global reference, solve for the rigid body transformation matrix from the visible light camera and near-infrared imaging coordinate systems to the global coordinate system; Configure a global synchronization pulse generator to lock the trigger signals of the three sensors to the same clock edge.

3. The method for intelligent identification and closed-loop management of potential hazards in property inspections according to claim 1, characterized in that, The physical causes of the inferred potential hazards include: Connectivity analysis of the divergence field is performed to extract regions with central radial features. If the spatial intersection ratio of this region with the high value region of the water content thermogram exceeds the preset value and the mean curl is less than 20% of the mean divergence, it is determined that the expansion is caused by internal water leakage. If the boundary of the high water content area is sharp and extends along a specific line that matches the water supply pipeline path, it is determined to be a pipeline rupture and leakage; If the high water content area is distributed in a diffuse, patchy manner and is adjacent to the outer wall, it is determined to be rainwater intrusion.

4. The method for intelligent identification and closed-loop management of potential hazards in property inspection according to claim 3, characterized in that, The specific process of matching the water supply pipeline path is as follows: The pipeline centerline in the digital twin model is retrieved and discretized into a three-dimensional path point sequence. The boundary skeleton points of the high water content area are extracted from the water content heat map as a two-dimensional path point sequence. An improved discrete Fréchet distance is constructed as the matching cost function and a tangent direction angle penalty factor is introduced. When the matching cost is less than a preset threshold and the water content of the area exceeds 1.5 times the global average water content, it is marked as a leakage source.

5. The method for intelligent identification and closed-loop management of potential hazards during property inspections according to claim 1, characterized in that, The core of the multiphysics simulation is: The extended Richards equations were run, coupling the pipeline leakage source term and the dominant flow channel of the wall cracks. The unsaturated hydraulic conductivity function introduced a wall damage correction term characterized by the normalized value of the divergence field. Simulations were performed with source term parameters set for different repair schemes to confirm the only effective scheme.

6. The method for intelligent identification and closed-loop management of potential hazards in property inspections according to claim 1, characterized in that, The adaptive generation logic for the parameter constraints of the operation steps is as follows: Extract the boundary of the high moisture content area and transform it to the engineering coordinate system to generate a three-dimensional bounding box as the calibration range; The maximum allowable percentage of the area exceeding the calibrated range is dynamically set after collision detection between the hidden engineering model in the digital twin and the bounding box. If the extended boundary touches the reinforcing bars or pipelines, the percentage is automatically reduced until a safe breach boundary is generated.

7. The method for intelligent identification and closed-loop management of potential hazards during property inspections according to claim 1, characterized in that, The specific method for the rigidity verification is as follows: After the repair is completed, the multimodal terminal is used again to scan the repair area synchronously. The obtained post-repair moisture content thermal map and three-dimensional deformation vector field are compared with the preset physical verification standards in the work order item by item. The work order is only allowed to be closed when all physical indicators fall within the standard threshold range; otherwise, the non-compliant items are indicated and rework items are generated.

8. The method for intelligent identification and closed-loop management of potential hazards during property inspections according to claim 1, characterized in that, The specific implementation of the long-term monitoring and knowledge self-evolution is as follows: After a work order is closed, the wireless sensor continues to transmit data back to the digital twin. If an upward trend in moisture content or an abnormal rate of microdeformation is detected, an early warning will be automatically triggered and historical work orders will be traced back. At the same time, the mapping relationship between physical characteristics, causes, and effective intervention measures for each successful closed loop will be integrated into the knowledge graph, and the weight of the physical cause logic chain will be updated.

9. A system applied to the intelligent identification and closed-loop management method for potential hazards in property inspections as described in any one of claims 1-8, characterized in that, include: The multimodal synchronous acquisition and digital twin base plate construction module is used to build a full-element building information model and component knowledge graph, and integrates visible light, near infrared and structured light sensors. After spatiotemporal joint calibration and global synchronous triggering, it acquires visible light images, moisture content thermal maps and three-dimensional deformation vector fields of components. The physical feature clustering and intelligent inference module for causes runs on edge computing nodes. It is used to calculate the curl and divergence of the deformation vector field. Combined with the gradient distribution and spatial overlap of the moisture content thermogram, it automatically identifies the physical causes of hidden dangers and outputs diagnostic conclusions. The twin simulation and adaptive work order generation module is used to map diagnostic conclusions to a digital twin, run multiphysics simulation to verify the maintenance plan, and automatically generate a closed-loop work order containing a diagnostic report, operation prescription, and physical verification standards. The operation parameter constraints are adaptively set by the twin model collision detection. The rigid verification and long-cycle knowledge self-evolution module is used to guide operations through augmented reality glasses during maintenance. After maintenance, a multimodal terminal is used for verification and rigid comparison with physical verification standards. Once the standards are met, the work order is closed. At the same time, the status of components is continuously monitored through wireless sensors, and the mapping relationship of successful closed loop is fed back to the knowledge graph to realize the self-evolution of diagnostic capabilities.

Images