A digital base-based operation and maintenance method and system for existing building glass curtain walls

Abstract

The present invention relates to a digital base-based operation and maintenance method and system for existing building glass curtain walls. The method comprises the following steps: obtaining curtain wall vibration response data, visible light image data, and three-dimensional point cloud data; comparing current modal parameters with historical modal parameters based on the curtain wall vibration response data to determine the dynamic characteristics of the curtain wall panels; identifying the type of apparent damage to the curtain wall and the corresponding degree of damage based on the visible light image data; constructing a real-life three-dimensional point cloud model of the curtain wall based on the three-dimensional point cloud data and mapping it into a numerical model; calculating local stiffness reduction coefficients based on the dynamic characteristics of the curtain wall panels and the type of apparent damage to the curtain wall and the corresponding degree of damage, and correcting the local stiffness matrix of the corresponding units or nodes in the numerical model; and performing finite element analysis on the corrected numerical model to evaluate and deduce the performance of the glass curtain wall. Compared with existing technologies, the present invention has the advantages of being able to accurately and dynamically perceive and predict changes in glass curtain wall performance.

Description

Technical Field

[0001] The present invention relates to the technical field of building curtain wall operation and maintenance, and in particular to a method and system for operating and maintaining an existing building glass curtain wall based on a digital base. Background Art

[0002] With the continuous advancement of modern construction technology, glass curtain walls, as a crucial component of building facades, are crucial for ensuring urban safety and resilience. Existing high-rise building curtain walls face the threat of severe dynamic loads such as typhoons and earthquakes during their service life. Traditional curtain wall inspection methods rely primarily on manual inspections, which are inefficient, costly, and heavily reliant on subjective judgment by inspectors. Furthermore, existing building glass curtain wall assessments focus on analyzing single modal data, making it difficult to reveal the performance status of glass curtain walls from multiple dimensions. CN113887091A discloses a prefabricated building simulation test system and method, which includes: a model assembly unit, used to assemble prefabricated components into a prefabricated building model using equivalent connectors of prefabricated components; a testing unit, used to carry the prefabricated building model and apply load to the prefabricated building model; a data acquisition unit, used to scan the prefabricated building model before testing to generate a three-dimensional point cloud building model, and also used to collect sensor data, displacement data and strain data during the test process; a calculation and analysis unit, used to perform finite element analysis on the three-dimensional point cloud building model to generate a finite element model and a measured model that can reflect the actual state of the model, and also used to process the finite element model and the measured model to obtain a digital twin model, and use the digital twin model to determine the load state of the prefabricated building model. However, this method is only applicable to prefabricated buildings, with a limited scope of application, and is not suitable for building glass curtain walls. Moreover, the collected data and analysis dimensions are single, focusing only on data on structural performance, and lack the ability to collect and analyze data on other important dimensions such as the building's appearance and dynamic response. In addition, this method determines the load state of the prefabricated building model through finite element analysis and digital twin models, which only involves the evaluation of the overall structural performance of the model under load, and cannot comprehensively evaluate and deduce the performance of existing building glass curtain walls under different working conditions.

[0003] Therefore, there is currently a lack of a method that can integrate multimodal data collection and combine curtain wall panel dynamic characteristics evaluation and curtain wall apparent damage evaluation to achieve comprehensive performance evaluation and deduction of glass curtain walls under different working conditions, thereby realizing comprehensive operation and maintenance of existing building glass curtain walls. Summary of the Invention

[0004] The purpose of the present invention is to provide an operation and maintenance method and system for existing building glass curtain walls based on a digital base. Multimodal data collection is performed through periodic and efficient inspections. On this basis, a digital base for glass curtain walls is constructed through curtain wall panel dynamic characteristics evaluation, curtain wall apparent damage evaluation, numerical simulation deduction and visual interaction to achieve full life cycle monitoring and evaluation of building glass curtain walls.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for operating and maintaining an existing building glass curtain wall based on a digital base comprises the following steps:

[0007] S1, obtaining curtain wall vibration response data, curtain wall visible light image data and curtain wall three-dimensional point cloud data;

[0008] S2, based on the curtain wall vibration response data, comparing current modal parameters with historical modal parameters to determine the dynamic characteristics of the curtain wall panel;

[0009] S3, based on the visible light image data of the curtain wall, identifying the type of apparent damage to the curtain wall and the corresponding degree of damage through an artificial intelligence visual large model;

[0010] S4, based on the curtain wall three-dimensional point cloud data, constructing a real-scene three-dimensional point cloud model of the curtain wall as a digital base, and mapping the three-dimensional point cloud model into a numerical model in combination with measured material performance parameters of the curtain wall components, calculating a local stiffness reduction factor based on the dynamic characteristics of the curtain wall panels and the apparent damage type and corresponding damage degree of the curtain wall, and correcting the local stiffness matrix of the corresponding unit or node in the numerical model;

[0011] S5, perform finite element analysis on the modified numerical model to evaluate and deduce the performance of the glass curtain wall under static load, wind load, and earthquake.

[0012] For curtain wall glass panels of typical specifications, the curtain wall vibration response data is collected using an acceleration sensor; for curtain wall glass panels of similar specifications in large quantities, the curtain wall vibration response data is collected using a laser vibrometer.

[0013] The S2 specifically includes the following steps:

[0014] S21, using the random subspace method to perform modal identification on the curtain wall vibration response data collected by the acceleration sensor to obtain the first n modal frequencies of the curtain wall glass panels of typical specifications;

[0015] S22, performing spectrum analysis on the curtain wall vibration response data collected by the laser vibrometer, using the modal identification result in step S21 to eliminate interference modes, and obtaining the first m modal frequencies of a large batch of curtain wall glass panels with similar specifications, where m < n;

[0016] S23, comparing the first m modal frequencies of a large batch of curtain wall glass panels with similar specifications with the historical modal frequencies, and judging whether the curtain wall glass panels are abnormal based on the average reduction rate of the first m modal frequencies, and classifying the degree of abnormality.

[0017] The types of apparent damage to the curtain wall include glass panel falling off, glass panel self-explosion, supporting component deformation, glass panel cracks and structural adhesive falling off.

[0018] The degree of damage is specifically as follows: for glass panel falling off, glass panel self-explosion and supporting member deformation, the corresponding apparent damage degree of the curtain wall is severe damage; for glass panel cracks and structural adhesive falling off, evaluation indicators are calculated for damage classification, wherein, for glass panel cracks, the evaluation indicator is calculated based on the glass panel peeling area and / or the glass panel crack length, and for structural adhesive falling off, the evaluation indicator is calculated based on the structural adhesive falling length.

[0019] The mapping of the three-dimensional point cloud model into a numerical model is specifically as follows:

[0020] The curtain wall panels and supporting components in the 3D point cloud model are segmented through the point cloud segmentation algorithm to obtain the geometric dimensions and spatial coordinates of the components. The component geometric parameters are corrected by locally refining the 3D point cloud model. Combined with the measured material performance parameters of the curtain wall components, the overall 3D point cloud model of the curtain wall is mapped into a numerical model, in which the supporting components are line units and the curtain wall panels are plate units.

[0021] Based on the dynamic characteristics of the curtain wall panel and the apparent damage type and corresponding damage degree of the curtain wall, the local stiffness reduction coefficient is calculated, and the local stiffness matrix of the corresponding unit or node in the numerical model is modified. Specifically, it is:

[0022] According to the change of the dynamic characteristics of the curtain wall glass panel, the first local stiffness reduction coefficient is calculated according to the average value of the reduction rate of the first m modal frequencies in the dynamic characteristics of the curtain wall panel. :

[0023] ,

[0024] Where, is the empirical coefficient, is the mean value of the frequency reduction rate of the first m-order modes;

[0025] According to the different types of curtain wall apparent damage, the second local stiffness reduction factor is determined according to the degree of damage. :

[0026] ,

[0027] Where, is the apparent damage type of the curtain wall. is an evaluation index of apparent damage. Corresponding to the curtain wall apparent damage type i The empirical coefficient of Corresponding injury category i Evaluation metrics j The value of, among which, for glass panel falling off, glass panel self-explosion and support member deformation, Take it as 1.

[0028] Based on the first local stiffness reduction factor and the second local stiffness reduction factor Modify the local stiffness matrix in a numerical model:

[0029] ,

[0030] Where, is the original local stiffness matrix, is the modified local stiffness matrix.

[0031] An existing building glass curtain wall operation and maintenance system based on a digital base, for implementing the method described above, comprising:

[0032] Multimodal data acquisition module: includes an accelerometer, a laser vibrometer, an unmanned aerial vehicle platform, and a 3D laser scanner. The unmanned aerial vehicle platform is equipped with an optical camera and an RTK laser radar. The accelerometer and laser vibrometer are used to collect curtain wall vibration response data, the optical camera is used to collect curtain wall visible light image data, and the RTK laser radar and 3D laser scanner are used to collect curtain wall 3D point cloud data.

[0033] Dynamic characteristics evaluation module: connected to the multimodal data acquisition module, used to compare current modal parameters with historical modal parameters based on the curtain wall vibration response data to determine the dynamic characteristics of the curtain wall panel;

[0034] Apparent damage identification module: connected to the multimodal data acquisition module, used to identify the type of apparent damage to the curtain wall and the corresponding damage degree based on the visible light image data of the curtain wall through the artificial intelligence visual large model;

[0035] Numerical simulation and deduction module: connected to the multimodal data acquisition module, the dynamic characteristics evaluation module and the apparent damage identification module, and used to construct a real-scene three-dimensional point cloud model of the curtain wall as a digital base based on the curtain wall three-dimensional point cloud data, and combine the measured material performance parameters of the curtain wall components to map the three-dimensional point cloud model into a numerical model, calculate the local stiffness reduction coefficient based on the dynamic characteristics of the curtain wall panel and the curtain wall apparent damage type and corresponding damage degree, and correct the local stiffness matrix of the corresponding unit or node in the numerical model; perform finite element analysis on the corrected numerical model to realize the performance evaluation and deduction of the glass curtain wall under static load, wind load and earthquake.

[0036] The system also includes a visualization interaction module: used to support roaming the glass curtain wall digital model at any angle, and visualize the performance evaluation and deduction results of the building glass curtain wall.

[0037] The user interacts with the system through the visualization interaction module. When the user is roaming in a three-dimensional real scene, the user can freely adjust the viewing angle, distance and observation range through interaction to view the damaged part, and use different color marks to distinguish and display glass panels in different states.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) The present invention focuses on the operation and maintenance management of existing building glass curtain walls. In view of the usage characteristics and practical problems faced by existing building glass curtain walls, it provides a special operation and maintenance method, which can accurately evaluate and deduce the performance status of existing building glass curtain walls, and provide a more targeted decision-making basis for operation and maintenance work.

[0040] (2) The present invention comprehensively collects curtain wall vibration response data, curtain wall visible light image data and curtain wall three-dimensional point cloud data, covering multiple dimensions such as the curtain wall's dynamic response, appearance status and overall geometric shape. The data collection is more comprehensive and can more comprehensively reflect the actual status of the curtain wall.

[0041] (3) The present invention accurately determines the dynamic characteristics of the curtain wall panel by comparing its current modal parameters with its historical modal parameters; uses an artificial intelligence visual large model to identify the apparent damage of the curtain wall, and can quickly and accurately identify the type and extent of the damage; and combines the measured material performance parameters to convert the real-scene three-dimensional point cloud model constructed by the three-dimensional point cloud data into a numerical model, making the data processing more in-depth and accurate, and providing a higher quality data foundation for subsequent numerical simulation and performance evaluation.

[0042] (4) The present invention calculates the local stiffness reduction coefficient and modifies the model, so that the numerical model can more realistically reflect the actual performance status of the curtain wall, improves the accuracy of performance evaluation and deduction, helps to discover potential problems in advance, formulate more scientific and reasonable operation and maintenance plans, and reduce the safety risks of the curtain wall during use. After modifying the local stiffness matrix of the numerical model, finite element analysis is performed to achieve performance evaluation and deduction of the glass curtain wall under various working conditions such as static load, wind load, and earthquake. It not only focuses on the performance of the curtain wall in its current state, but also simulates and deduces complex working conditions that may arise in the future, providing more comprehensive and in-depth support for operation and maintenance decisions.

[0043] (5) The present invention realizes digital management of glass curtain walls based on the construction and modification of digital bases and numerical models, enabling operation and maintenance personnel to intuitively understand the performance status and damage of curtain walls through virtual models, facilitating remote monitoring and data analysis, and improving the intelligence level and decision-making efficiency of operation and maintenance management. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flow chart of the method of the present invention;

[0045] Figure 2 This is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0046] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0047] Example 1

[0048] This embodiment provides an operation and maintenance method for an existing building glass curtain wall based on a digital base. Figure 1 As shown, the following steps are included:

[0049] S1, obtain curtain wall vibration response data, curtain wall visible light image data and curtain wall three-dimensional point cloud data.

[0050] In this embodiment, a periodic and rapidly implemented curtain wall inspection mechanism is used to continuously collect data on the glass curtain wall and obtain dynamic data on the performance of the glass curtain wall in the time dimension.

[0051] Before implementing multimodal data collection, a sensor network was deployed inside the building. Typical glass curtain wall panel specifications were identified and densely populated with accelerometers. A drone airport was deployed outside the building, and drones equipped with ultra-high-definition optical cameras and RTK lidar equipment were deployed inside the airport. Laser scanners were deployed in selected areas of the curtain wall to collect 3D point cloud data and construct a detailed local model.

[0052] Specifically, 16 accelerometers are evenly distributed across panels of typical specifications, collecting vibration response data for typical-sized glass panels. Single-point laser vibrometers are evenly distributed across a large number of panels of similar specifications, collecting vibration response data for similar-sized glass panels. A drone platform equipped with an optical camera regularly collects high-precision visible light image data of the curtain wall along a pre-set, customized cruise route for surface damage assessment. An RTK laser radar is also equipped on the drone platform, regularly collecting 3D point cloud data of the entire glass curtain wall along a pre-set route for reconstruction of the overall point cloud model. A laser scanner collects 3D point cloud data of the curtain wall region for detailed local model construction, obtaining high-precision geometric parameters of local curtain wall components.

[0053] During the normal service of the glass curtain wall, multimodal data collection is implemented every 3-6 months, and the frequency of multimodal data collection is increased after extreme weather conditions, including typhoons and earthquakes.

[0054] S2, based on the curtain wall vibration response data, compares the current modal parameters with the historical modal parameters to determine the dynamic characteristics of the curtain wall panel.

[0055] S2 specifically includes the following steps:

[0056] S21, using the random subspace method to perform modal identification on the curtain wall vibration response data collected by the acceleration sensor to obtain the first n modal frequencies of the curtain wall glass panels of typical specifications;

[0057] S22, performing spectrum analysis on the curtain wall vibration response data collected by the laser vibrometer, using the modal identification results in step S21 to eliminate interference modes, and obtaining the top three modal frequencies of a large batch of curtain wall glass panels with similar specifications;

[0058] S23 compares the first three modal frequencies of a large batch of curtain wall glass panels of similar specifications with historical modal frequencies. Based on the average of the reduction rates of the first three modal frequencies, the curtain wall glass panels are judged to be abnormal. The degree of abnormality is then classified as no abnormality, slight abnormality, moderate abnormality, and severe abnormality. In one embodiment, a reduction rate of the first three modal frequencies of 0-5% indicates no abnormality, 5%-15% indicates slight abnormality, 15%-30% indicates moderate abnormality, and 30% or more indicates severe abnormality.

[0059] S3, based on the visible light image data of the curtain wall, uses the artificial intelligence visual big model to identify the apparent damage type and corresponding damage degree of the curtain wall.

[0060] In this example, a pre-trained large-scale visual model based on the Transformer architecture is used, and fine-tuned using fine-tuning technology on a glass curtain wall damage dataset. The fine-tuned model identifies and classifies curtain wall damage, and grades the damage based on evaluation indicators to obtain the degree of damage. The details are as follows:

[0061] High-definition curtain wall image data is pre-processed and damage annotated, including abnormal image removal and image resizing. Image annotation tools are then used to perform pixel-level masking of different curtain wall damage types, accurately marking the location and shape of damaged areas. Curtain wall damage includes glass panel detachment, glass panel explosion, support member deformation, glass panel cracks, and structural adhesive loss.

[0062] The pre-trained model based on the Transformer architecture can use large visual models such as ViT and DinoV2, and adapt to the curtain wall surface damage dataset through LoRA fine-tuning technology, and fine-tune the Transformer through LoRA technology. and Parameter matrix, LoRA module fine-tuning process is as follows:

[0063] Fixed pre-trained model weight matrix ; Random Gaussian distribution initializes the dimensionality reduction matrix , zero matrix initializes the dimension-raising matrix ,in ;

[0064] The parameter update matrix is calculated as:

[0065] ,

[0066] Calculate the forward propagation output :

[0067] ,

[0068] LoRA fine-tuning technology updates matrix parameters much smaller than those in pre-trained models, thereby reducing the number of parameter updates during large-scale visual model training, improving model fine-tuning efficiency and reducing model fine-tuning costs. During large-scale visual model inference, curtain wall image data is input into the large model to obtain curtain wall damage types and corresponding damage masks.

[0069] The types of curtain wall surface damage classified by the pre-trained visual large model include glass panel falling off, glass panel self-explosion, supporting component deformation, glass panel cracks and structural adhesive falling off.

[0070] For glass panel falling off, glass panel self-explosion and supporting member deformation, the corresponding apparent damage degree of the curtain wall is severe damage; for glass panel cracks and structural adhesive falling off, the damage is graded by calculating the evaluation index, which is divided into slight damage, moderate damage and severe damage. Among them, for glass panel cracks, the evaluation index is calculated based on the glass panel peeling area and / or glass panel crack length, and for structural adhesive falling off, the evaluation index is calculated based on the structural adhesive falling length. Specifically:

[0071] Evaluation indicator 1: The ratio of the potential peeling area of the glass panel and / or the closed area of the glass panel crack to the glass panel area;

[0072] Evaluation indicator 2: Ratio of glass panel crack length to glass panel circumference;

[0073] Evaluation indicator 3: The ratio of the length of structural adhesive peeling to the total length of structural adhesive on the glass panel.

[0074] In one embodiment, an evaluation index of 0-10% is mild injury, 10%-30% is moderate injury, and above 30% is severe injury.

[0075] S4, based on the curtain wall 3D point cloud data, constructs a real-scene 3D point cloud model of the curtain wall as a digital base, and combines the measured material performance parameters of the curtain wall components to map the 3D point cloud model into a numerical model. Based on the dynamic characteristics of the curtain wall panels and the apparent damage type and corresponding damage degree of the curtain wall, the local stiffness reduction coefficient is calculated, and the local stiffness matrix of the corresponding unit or node in the numerical model is corrected.

[0076] S41, establishes a 3D point cloud model based on the curtain wall 3D point cloud data, which serves as the platform basic model and carrier for the operation and maintenance of existing building glass curtain walls.

[0077] Filtering and denoising are performed on RTK lidar data to remove noise and outliers from the point cloud data, improving data clarity and accuracy. Subsequently, a simplification operation based on a voxel downsampling algorithm converts the data into a voxel grid and adjusts the resolution to achieve simplification. This reduces the complexity and storage requirements of the point cloud data, improving the efficiency of data processing and transmission while maintaining the data's geometric characteristics. Finally, a stitching operation based on an iterative closest point algorithm iteratively optimizes the alignment of the point cloud data to minimize overlap between different point clouds. Point cloud data acquired from multiple locations or angles is then stitched together into a complete model, eliminating overlap and gaps between different point clouds.

[0078] This embodiment utilizes control point optimization to improve model construction accuracy. Control points are placed using both artificial and natural control points, both on the structure surface and on the surrounding ground. A total station is used to measure 3D coordinates to ensure accurate reconstruction of the 3D point cloud model. During 3D model reconstruction, computer vision algorithms are employed to analyze the model's geometric features and texture information, remove suspended objects, and render the model, optimizing both 3D model quality and visual quality.

[0079] S42, mapping the three-dimensional point cloud model into a numerical model.

[0080] The curtain wall panels and supporting components in the 3D point cloud model are segmented through a random sampling consistency algorithm to obtain the geometric dimensions and spatial coordinates of the components. The component geometric parameters are corrected by locally refining the 3D point cloud model. Combined with the measured material performance parameters of the curtain wall components, the overall 3D point cloud model of the curtain wall is mapped into a numerical model, in which the supporting components are line units and the curtain wall panels are plate units.

[0081] S43, mapping the analysis results of steps S2 and S3 to the numerical model in a stiffness reduction form, and correcting the numerical model.

[0082] Specifically, the local stiffness reduction factor is calculated based on the dynamic characteristics of the curtain wall panel and the apparent damage type and corresponding damage degree of the curtain wall, and the local stiffness matrix of the corresponding unit or node in the numerical model is corrected.

[0083] According to the change of the dynamic characteristics of the curtain wall glass panel, the first local stiffness reduction coefficient is calculated according to the average value of the reduction rate of the first m modal frequencies in the dynamic characteristics of the curtain wall panel. :

[0084] ,

[0085] Where, is the empirical coefficient, is the mean value of the frequency reduction rate of the first m-order modes;

[0086] According to the different types of curtain wall apparent damage, the second local stiffness reduction factor is determined according to the degree of damage. :

[0087] ,

[0088] Where, is the apparent damage type of the curtain wall. is an evaluation index of apparent damage. Corresponding to the curtain wall apparent damage type i The empirical coefficient of Corresponding injury category i Evaluation metrics jThe value of, among which, for glass panel falling off, glass panel self-explosion and support member deformation, Take it as 1.

[0089] Based on the first local stiffness reduction factor and the second local stiffness reduction factor Modify the local stiffness matrix in a numerical model:

[0090] ,

[0091] Where, is the original local stiffness matrix, is the modified local stiffness matrix.

[0092] S5, perform finite element analysis on the modified numerical model to evaluate and deduce the performance of the glass curtain wall under static load, wind load, and earthquake.

[0093] Example 2

[0094] This embodiment provides an existing building glass curtain wall operation and maintenance system based on a digital base, which is used to implement the method of embodiment 1, such as Figure 2 As shown, the system includes:

[0095] (1) Multimodal data acquisition module: including accelerometer, laser vibrometer, UAV platform, and 3D laser scanner. The UAV platform is equipped with an optical camera and RTK laser radar. The accelerometer and laser vibrometer are used to collect curtain wall vibration response data, the optical camera is used to collect curtain wall visible light image data, and the RTK laser radar and 3D laser scanner are used to collect curtain wall 3D point cloud data.

[0096] (2) Dynamic characteristics evaluation module: connected to the multimodal data acquisition module, used to compare the current modal parameters with the historical modal parameters based on the curtain wall vibration response data to determine the dynamic characteristics of the curtain wall panel;

[0097] (3) Surface damage identification module: connected to the multimodal data acquisition module, used to identify the surface damage type and corresponding damage degree of the curtain wall based on the curtain wall visible light image data through the artificial intelligence visual large model;

[0098] (4) Numerical simulation module: connects the multimodal data acquisition module, the dynamic characteristics evaluation module and the apparent damage identification module, and is used to construct a real-scene three-dimensional point cloud model of the curtain wall as a digital base based on the curtain wall three-dimensional point cloud data, and map the three-dimensional point cloud model into a numerical model in combination with the measured material performance parameters of the curtain wall components. Based on the dynamic characteristics of the curtain wall panel and the curtain wall apparent damage type and the corresponding damage degree, the local stiffness reduction coefficient is calculated, and the local stiffness matrix of the corresponding unit or node in the numerical model is corrected; the finite element analysis is performed on the corrected numerical model to realize the performance evaluation and deduction of the glass curtain wall under static, wind load and earthquake action;

[0099] (5) Visualization interaction module: used to support roaming the digital model of glass curtain wall at any angle, and visualize the performance evaluation and deduction results of architectural glass curtain wall.

[0100] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the multimodal data acquisition module, dynamic characteristics evaluation module, apparent damage identification module, and numerical simulation deduction module can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0101] For the visualization interaction module, a roamable glass curtain wall digital model (including fixed-track roaming and free roaming) is built based on the UnrealEngine engine. Users interact with the system through the visualization interaction module. When roaming in a 3D real scene, they can freely adjust the viewing angle, distance, and observation range through interaction to view the damaged parts. Including:

[0102] The damage characteristics of the building facade, such as cracks, peeling, and deformation, are marked and prompted. Different color marks are used to distinguish and display glass panels in different states. Green indicates normal, orange indicates abnormal, and red indicates serious abnormality.

[0103] Display historical monitoring data, structural vibration response, and historical changes in building facade defects, compare damage changes during multiple inspections, and predict the development of damage indicators.

[0104] The UE platform has high-fidelity real-time rendering capabilities and can support complex lighting effects and material performance, providing a more realistic visual experience for the visual interaction of glass curtain walls.

[0105] Under the UE engine, the historical changes in the vibration response and apparent damage of the historical curtain wall are displayed, and the damage changes during multiple inspections are compared through the built-in fine-tuned visual large model, and the development of damage indicators is predicted through the LSTM algorithm.

[0106] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A method for operating and maintaining an existing building glass curtain wall based on a digital base, characterized in that: The following steps are involved: S1, obtaining curtain wall vibration response data, curtain wall visible light image data and curtain wall three-dimensional point cloud data; S2, based on the curtain wall vibration response data, comparing current modal parameters with historical modal parameters to determine the dynamic characteristics of the curtain wall panel; The S2 specifically includes the following steps: S21, using the random subspace method to perform modal identification on the curtain wall vibration response data collected by the acceleration sensor to obtain the first n modal frequencies of the curtain wall glass panels of typical specifications; S22, performing spectrum analysis on the curtain wall vibration response data collected by the laser vibrometer, using the modal identification result in step S21 to eliminate interference modes, and obtaining the first m modal frequencies of a large batch of curtain wall glass panels with similar specifications, where m < n; S23, comparing the first m modal frequencies of a large batch of curtain wall glass panels with similar specifications with the historical modal frequencies, and judging whether the curtain wall glass panels are abnormal based on the average of the reduction rates of the first m modal frequencies, and classifying the degree of abnormality; S3, based on the visible light image data of the curtain wall, identifying the type of apparent damage to the curtain wall and the corresponding degree of damage through an artificial intelligence visual large model; S4, based on the curtain wall three-dimensional point cloud data, constructing a real-scene three-dimensional point cloud model of the curtain wall as a digital base, and mapping the three-dimensional point cloud model into a numerical model in combination with measured material performance parameters of the curtain wall components, calculating a local stiffness reduction factor based on the dynamic characteristics of the curtain wall panels and the apparent damage type and corresponding damage degree of the curtain wall, and correcting the local stiffness matrix of the corresponding unit or node in the numerical model; The local stiffness reduction factor is calculated based on the dynamic characteristics of the curtain wall panel and the apparent damage type and corresponding damage degree of the curtain wall, and the local stiffness matrix of the corresponding unit or node in the numerical model is modified, specifically: According to the change of the dynamic characteristics of the curtain wall glass panel, the first local stiffness reduction coefficient is calculated according to the average value of the reduction rate of the first m modal frequencies in the dynamic characteristics of the curtain wall panel. : ， Where, is the empirical coefficient, is the mean value of the frequency reduction rate of the first m-order modes; According to the different types of curtain wall apparent damage, the second local stiffness reduction factor is determined according to the degree of damage. : ， Where, is the apparent damage type of the curtain wall. is an evaluation index of apparent damage. Corresponding to the curtain wall apparent damage type i The empirical coefficient of Corresponding injury category i Evaluation metrics j The value of, among which, for glass panel falling off, glass panel self-explosion and support member deformation, Take it as 1; Based on the first local stiffness reduction factor and the second local stiffness reduction factor Modify the local stiffness matrix in a numerical model: ， Where, is the original local stiffness matrix, is the modified local stiffness matrix; S5, perform finite element analysis on the modified numerical model to evaluate and deduce the performance of the glass curtain wall under static load, wind load, and earthquake.

2. The method for operating and maintaining an existing building glass curtain wall based on a digital base according to claim 1, characterized in that: For curtain wall glass panels of typical specifications, the curtain wall vibration response data is collected using an acceleration sensor; for curtain wall glass panels of similar specifications in large quantities, the curtain wall vibration response data is collected using a laser vibrometer.

3. The method for operating and maintaining an existing building glass curtain wall based on a digital base according to claim 1, characterized in that: The types of apparent damage to the curtain wall include glass panel falling off, glass panel self-explosion, supporting component deformation, glass panel cracks and structural adhesive falling off.

4. The method for operating and maintaining an existing building glass curtain wall based on a digital base according to claim 3 is characterized in that: The degree of damage is specifically as follows: for glass panel falling off, glass panel self-explosion and supporting member deformation, the corresponding apparent damage degree of the curtain wall is severe damage; for glass panel cracks and structural adhesive falling off, evaluation indicators are calculated for damage classification, wherein, for glass panel cracks, the evaluation indicator is calculated based on the glass panel peeling area and / or the glass panel crack length, and for structural adhesive falling off, the evaluation indicator is calculated based on the structural adhesive falling length.

5. The method for operating and maintaining an existing building glass curtain wall based on a digital base according to claim 1, characterized in that: The mapping of the three-dimensional point cloud model into a numerical model is specifically as follows: The curtain wall panels and supporting components in the 3D point cloud model are segmented through the point cloud segmentation algorithm to obtain the geometric dimensions and spatial coordinates of the components. The component geometric parameters are corrected by locally refining the 3D point cloud model. Combined with the measured material performance parameters of the curtain wall components, the overall 3D point cloud model of the curtain wall is mapped into a numerical model, in which the supporting components are line units and the curtain wall panels are plate units.

6. An existing building glass curtain wall operation and maintenance system based on a digital base, characterized in that: For implementing the method according to any one of claims 1 to 5, the system comprises: Multimodal data acquisition module: includes an accelerometer, a laser vibrometer, an unmanned aerial vehicle platform, and a 3D laser scanner. The unmanned aerial vehicle platform is equipped with an optical camera and an RTK laser radar. The accelerometer and laser vibrometer are used to collect curtain wall vibration response data, the optical camera is used to collect curtain wall visible light image data, and the RTK laser radar and 3D laser scanner are used to collect curtain wall 3D point cloud data. Dynamic characteristics evaluation module: connected to the multimodal data acquisition module, used to compare current modal parameters with historical modal parameters based on the curtain wall vibration response data to determine the dynamic characteristics of the curtain wall panel; Apparent damage identification module: connected to the multimodal data acquisition module, used to identify the type of apparent damage to the curtain wall and the corresponding damage degree based on the visible light image data of the curtain wall through the artificial intelligence visual large model; Numerical simulation and deduction module: connected to the multimodal data acquisition module, the dynamic characteristics evaluation module and the apparent damage identification module, and used to construct a real-scene three-dimensional point cloud model of the curtain wall as a digital base based on the curtain wall three-dimensional point cloud data, and combine the measured material performance parameters of the curtain wall components to map the three-dimensional point cloud model into a numerical model, calculate the local stiffness reduction coefficient based on the dynamic characteristics of the curtain wall panel and the curtain wall apparent damage type and corresponding damage degree, and correct the local stiffness matrix of the corresponding unit or node in the numerical model; perform finite element analysis on the corrected numerical model to realize the performance evaluation and deduction of the glass curtain wall under static load, wind load and earthquake.

7. The digital base-based operation and maintenance system for existing building glass curtain walls according to claim 6 is characterized in that: The system also includes a visualization interaction module: used to support roaming the glass curtain wall digital model at any angle, and visualize the performance evaluation and deduction results of the building glass curtain wall.

8. The digital base-based operation and maintenance system for existing building glass curtain walls according to claim 7 is characterized in that: The user interacts with the system through the visualization interaction module. When the user is roaming in a three-dimensional real scene, the user can freely adjust the viewing angle, distance and observation range through interaction to view the damaged part, and use different color marks to distinguish and display glass panels in different states.

Images