A real-time safety monitoring method for large-span composite phase-change glass photovoltaic curtain wall

Abstract

The present invention discloses a real-time safety monitoring method for a large-span composite phase-change glass photovoltaic curtain wall, which relates to the field of photovoltaic curtain wall operation and maintenance monitoring technology, and realizes all-round and dynamic safety supervision of the photovoltaic curtain wall. The present invention constructs a twin three-dimensional scene model through the real-time status data of the photovoltaic curtain wall, divides the glass model and the photovoltaic component model in the twin three-dimensional scene model, generates a deformation monitoring network of the glass model and the photovoltaic component model through deformation checkpoints, updates the twin three-dimensional scene model through the real-time status data of the photovoltaic curtain wall, and determines whether there is an abnormality in the deformation detection point through the real-time status data. The deformation monitoring network generates a primary stress analysis network and a secondary stress analysis network based on the abnormal deformation detection point, and then traverses the primary stress abnormal part and the secondary stress abnormal part in the twin three-dimensional scene model through the primary stress analysis network and the secondary stress analysis network.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic curtain wall operation and maintenance monitoring, and in particular to a real-time safety monitoring method for a large-span composite phase-change glass photovoltaic curtain wall. Background Art

[0002] With the growing global demand for clean energy, photovoltaic power generation has gained widespread adoption as a sustainable energy solution. As a key component of photovoltaic power generation systems, photovoltaic curtain walls not only generate electricity but also serve as a decorative and protective facade for buildings. However, over the long term, photovoltaic curtain walls are susceptible to various factors, such as temperature fluctuations, wind, and earthquakes. These factors can cause deformation of the glass and photovoltaic modules, impacting their power generation efficiency and safety.

[0003] Currently, monitoring of photovoltaic curtain walls primarily relies on traditional sensor technology. These sensors can only capture basic parameters such as temperature and light intensity, but are unable to comprehensively and intuitively reflect the real-time status of the curtain wall and potential safety hazards. Furthermore, traditional monitoring methods lack effective means to monitor and analyze curtain wall deformation, making it difficult to promptly detect and address abnormalities. Therefore, accurately and in real time monitor the status of photovoltaic curtain walls and promptly identify and address potential safety hazards, which has become a pressing issue in the photovoltaic power generation field. To address this issue, a real-time safety monitoring method for large-span composite phase-change glass photovoltaic curtain walls is proposed. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a real-time safety monitoring method for a large-span composite phase-change glass photovoltaic curtain wall.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for real-time safety monitoring of a large-span composite phase-change glass photovoltaic curtain wall comprises the following steps:

[0007] Step 1: Set up a sensor network in the photovoltaic scene and collect real-time status data of each photovoltaic curtain wall in the photovoltaic scene through the sensor network;

[0008] Step 2: Build a scene framework model of the photovoltaic scene based on the real-time status data of each photovoltaic curtain wall, mark several data nodes in the scene framework model, and then map all real-time status data onto the scene framework model based on the location distribution of the data nodes to obtain a twin 3D scene model;

[0009] Step 3: Segment the glass model and the PV module model within the twin 3D scene model. Set several deformation detection points on the glass model and the PV module model, and then generate a deformation monitoring network for the glass model and the PV module model based on the deformation detection points.

[0010] Step 4: Update the twin 3D scene model using the real-time status data of the photovoltaic curtain wall, and use the real-time status data to determine whether there are any abnormalities in the deformation detection points. The deformation monitoring network generates a primary stress analysis network and a secondary stress analysis network based on the abnormal deformation detection points. Then, the primary stress analysis network and the secondary stress analysis network are used to traverse the primary stress abnormal part and the secondary stress abnormal part in the twin 3D scene model.

[0011] Furthermore, the division process of the glass model and the photovoltaic module model includes:

[0012] Marking multiple fuzzy boundary areas in the twin 3D scene model and dividing the twin 3D scene model into a number of scene voxels, each of which is a cube and has a pixel value corresponding to each face;

[0013] Randomly select no less than 100 boundary recognition voxels in the fuzzy boundary area of ​​the twin 3D scene model, and select no less than 10 scene voxels around each data node as inward recognition voxels;

[0014] Each inward identification voxel grows in the vertical direction of its fuzzy boundary area, and the boundary identification voxel grows in the vertical direction of the adjacent fuzzy boundary area. Then, each boundary identification voxel and inward identification voxel matches the scene voxels at adjacent positions along the growth direction, bidirectionally or unidirectionally, and obtains the pixel value difference of each surface between the two;

[0015] A pixel difference threshold is set for each fuzzy boundary area. If the pixel value difference of more than four faces is less than or equal to the pixel difference threshold, the scene voxel matched by the boundary recognition voxel or the inward recognition voxel is recorded as the boundary recognition voxel or the inward recognition voxel. Otherwise, the matching operation of the boundary recognition voxel or the inward recognition voxel is stopped.

[0016] When there is a scene voxel that is recorded as both a boundary recognition voxel and an inward recognition voxel, a voxel path is traversed between the corresponding data node and the fuzzy boundary area along the matching path;

[0017] The voxel paths between the fuzzy boundary areas are connected to each other, and the glass model and photovoltaic module model are divided within the curtain wall area model based on the connection results.

[0018] Furthermore, the process of establishing a deformation monitoring network includes:

[0019] Set multiple temperature detection intervals, each interval being 5 degrees Celsius, and obtain several sets of historical status data of the photovoltaic scene in each temperature detection interval, the historical status data including historical temperature change curves and historical stress change curves;

[0020] Establish a two-dimensional coordinate system, map several sets of historical status data corresponding to the same photovoltaic curtain wall in the same temperature detection range onto the two-dimensional coordinate system, and then establish multi-segment linear regression equations in different temperature detection ranges based on the historical temperature change curve and the historical stress change curve;

[0021] Several dynamic deformation detection points are randomly set on the glass model and the photovoltaic module model. The positions of the dynamic deformation detection points change synchronously with the acquisition frequency, but the total number remains unchanged.

[0022] The dynamic deformation detection points on the glass model and the photovoltaic module model are connected nearby to obtain a deformation monitoring network.

[0023] Furthermore, the process of constructing the primary stress analysis network and the secondary stress analysis network through the deformation monitoring network includes:

[0024] According to the temperature detection range of the real-time temperature value and the position of the deformation detection point, the corresponding multi-segment linear regression equation is retrieved, and the real-time temperature value is input into the multi-segment linear regression equation to obtain the estimated stress value;

[0025] If the estimated stress value is greater than or equal to the real-time stress value, the position of the corresponding deformation detection point is judged to be normal;

[0026] If the estimated stress value is less than the real-time stress value, the location of the corresponding deformation detection point is judged to be abnormal, and the corresponding deformation detection point is recorded as an abnormal deformation detection point. Then, the deformation monitoring network constructs a primary stress analysis network and a secondary stress analysis network with the abnormal deformation detection point as the starting point.

[0027] Furthermore, the process of traversing the main stress anomaly part and the secondary stress anomaly part on the twin 3D scene model includes:

[0028] The primary stress analysis network uses the stress direction of the abnormal deformation detection point as the main force analysis direction. The primary stress analysis network then extends from the main force analysis direction, traversing deformation detection points whose stress directions have an angle between (-45° and 45°) and the main force analysis direction, or are also abnormal deformation detection points, and incorporating the traversal results into the primary stress analysis network until the edge positions of the glass model and the photovoltaic module model are reached;

[0029] When the first-level stress analysis network traversal is completed, the deformation detection points in the first-level stress analysis network are sequentially connected according to their spatial positions, and the connection results are recorded as the main stress abnormal part;

[0030] The secondary stress analysis network first checks whether there is an abnormality in the deformation detection points at the adjacent spatial positions of the abnormal deformation detection point. If it is judged that there is no abnormality, no operation is performed;

[0031] If it is judged to exist, the stress direction of the abnormal deformation detection point at the adjacent spatial position is used as the main force analysis direction, and then the process of incorporating the first-level stress analysis network into the deformation detection point is repeated to traverse the secondary stress abnormal part on the twin three-dimensional scene model.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. This invention updates the twin 3D scene model using real-time status data and determines whether anomalies exist based on deformation detection points. Once an anomaly is detected, the deformation monitoring network rapidly generates primary and secondary stress analysis networks, accurately traversing the primary and secondary stress anomalies within the twin 3D scene model.

[0034] 2. The present invention effectively improves the safety and reliability of the photovoltaic curtain wall of composite phase change glass through all-round and dynamic safety supervision, reduces the risk of accidents, and provides strong guarantees for the normal operation and service life of the curtain wall. It is of great significance to improving the safety level of buildings and the long-term stable operation of photovoltaic curtain walls. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention.

[0036] Figure 1 This is a flow chart of a method for real-time safety monitoring of a large-span composite phase-change glass photovoltaic curtain wall according to the present invention. DETAILED DESCRIPTION

[0037] To make the objectives, technical solutions, and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementations obtained by those of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0038] like Figure 1As shown, a real-time safety monitoring method for a large-span composite phase-change glass photovoltaic curtain wall includes the following steps:

[0039] Step 1: Set up a sensor network in the photovoltaic scene and collect real-time status data of each photovoltaic curtain wall in the photovoltaic scene through the sensor network;

[0040] Step 2: Build a scene framework model of the photovoltaic scene based on the real-time status data of each photovoltaic curtain wall, mark several data nodes in the scene framework model, and then map all real-time status data onto the scene framework model based on the location distribution of the data nodes to obtain a twin 3D scene model;

[0041] Step 3: Segment the glass model and the PV module model within the twin 3D scene model. Set several deformation detection points on the glass model and the PV module model, and then generate a deformation monitoring network for the glass model and the PV module model based on the deformation detection points.

[0042] Step 4: Update the twin 3D scene model using the real-time status data of the photovoltaic curtain wall, and use the real-time status data to determine whether there are any abnormalities in the deformation detection points. The deformation monitoring network generates a primary stress analysis network and a secondary stress analysis network based on the abnormal deformation detection points. Then, the primary stress analysis network and the secondary stress analysis network are used to traverse the primary stress abnormal part and the secondary stress abnormal part in the twin 3D scene model.

[0043] Furthermore, the step 1 is achieved through the following process:

[0044] Step 101: Install n fiber Bragg grating sensor arrays in the photovoltaic scene, and install each fiber Bragg grating sensor array in a grid-like distribution on the photovoltaic curtain wall along the main force direction of the photovoltaic curtain wall. The grid distance between each fiber Bragg grating sensor array is dynamically adjusted according to the curtain wall span, for example, the span is 5m. Nine fiber Bragg grating sensor arrays are installed in a 5m area, where n is a natural number greater than 10;

[0045] It should be noted that the data acquisition ranges of the fiber Bragg grating sensor arrays at adjacent spatial locations partially overlap. The photovoltaic curtain wall is encapsulated by layers of composite phase change glass, adhesive film, and photovoltaic modules.

[0046] The fiber Bragg grating sensor array is composed of an infrared laser sensor, a pulse laser sensor, a stress sensor, a wireless communication device and a stress-sensitive sheet;

[0047] Data nodes are set according to the positions of wireless communication devices of the fiber grating sensor array, and adjacent data nodes are connected to each other according to the spatial position relationship between the data nodes to obtain a sensor network.

[0048] Step 102: setting the same acquisition frequency for each fiber Bragg grating sensor array, and setting a synchronous data verification period during the acquisition frequency, wherein the synchronous data verification period is used to verify various status data collected by the fiber Bragg grating sensor arrays at adjacent spatial positions;

[0049] Before the photovoltaic scene is put into use, the laser image data of each photovoltaic curtain wall is collected through the fiber grating sensor array;

[0050] When the photovoltaic scene is put into use, each fiber grating sensor array transmits the real-time status data collected by it through the wireless communication device at each collection frequency. The real-time status data includes real-time infrared image data, real-time laser image data and real-time regional stress change curve.

[0051] According to the connection status between the data nodes in the sensor network, the data is sent to the wireless communication device corresponding to the data node with the connection relationship, and synchronously enters the synchronization data verification period.

[0052] Step 103: During the synchronization data verification period, the real-time status data collected by the fiber Bragg grating sensor array corresponding to each data node is placed in the center, and the real-time status data of the remaining fiber Bragg grating sensor arrays are distributed around the data node according to the spatial position relationship in the sensor network.

[0053] Setting a number of regional identification points in each real-time status data, and retrieving the connections between the regional identification points, thereby dividing a number of matching identification areas in the real-time status data;

[0054] Match the matching identification area in the surrounding real-time status data with the matching identification area in the real-time status data at the center position. Since the data acquisition range of the fiber grating sensor arrays at adjacent spatial positions is partially the same, the corresponding real-time status data collected synchronously have the same part.

[0055] If it is determined that there are identical parts in the two matching identification areas, then based on the location of the identical parts in the matching identification areas in the surrounding real-time status data, the corresponding matching identification areas are retrieved and matched again with the matching identification areas in the real-time status data at the center position, and the identical parts are marked in the two real-time status data;

[0056] If it is determined that there is no identical part in the two matching recognition areas, no operation is performed;

[0057] Repeat the above matching operation until the real-time status data at the center position and the same part in the real-time status data of the surrounding areas appear in the matching process, and then integrate the same parts in the real-time status data of the surrounding areas to obtain the verified real-time status data segment;

[0058] The verification real-time status data segment is overlapped with the corresponding data segment in the real-time status data at the center position. The corresponding data segment is averaged according to the data segment overlap result. That is, for infrared image data and laser image data, the pixel values ​​of the corresponding pixels are directly averaged. For the regional stress change curve, the corresponding curve values ​​are averaged.

[0059] When using, combine the contents in steps 101 to 103:

[0060] By setting up a sensor network to collect real-time status data of the photovoltaic curtain wall and constructing a twin 3D scene model, the real-time status of the photovoltaic curtain wall can be comprehensively and intuitively reflected. Compared with traditional monitoring methods, this method can obtain more dimensional data, including not only basic parameters such as temperature and light intensity, but also integrate various real-time status data into the 3D model through scene laser image data and data node mapping, providing richer information for photovoltaic curtain wall status assessment.

[0061] Furthermore, the step 2 is achieved through the following process:

[0062] Step 201: Retrieve laser image data collected by each fiber Bragg grating sensor array before the photovoltaic scene is put into use, and splice the corresponding laser image data according to the position of each data node in the sensor network to obtain scene laser image data before the photovoltaic scene is put into use;

[0063] Then, a scene framework model is constructed based on the scene laser image data before the photovoltaic scene is put into use. According to the position distribution of each photovoltaic curtain wall in the photovoltaic scene, m curtain wall area models are divided in the scene framework model, where m is a positive integer greater than 0.

[0064] At the same time, according to the spatial position of the fiber grating sensor array in the photovoltaic scene, several data nodes are marked on each curtain wall area model;

[0065] After the photovoltaic scene is put into use, each curtain wall area model is dynamically updated based on the real-time laser image data verified by each data node after each synchronization data verification period.

[0066] Step 202: Using the location of each data node as the center and the distance to the adjacent data node as the radius, map the real-time infrared image data and real-time regional stress change curve collected by each data node onto the curtain wall area model where the data node is located, thereby obtaining a twin 3D scene model at the previous acquisition frequency.

[0067] The real-time infrared image data is mapped onto the curtain wall area model in the form of thermal patterns, and the real-time regional stress change curve is mapped onto the curtain wall area model in the form of stress point cloud.

[0068] When using, combine the contents in steps 201 to 202:

[0069] Based on the collected real-time status data, a laser image of the photovoltaic scene is constructed, and all real-time status data is mapped onto this image to generate a twin 3D scene model. This achieves the digitization and virtualization of the actual photovoltaic scene, allowing managers to intuitively observe the operation of the photovoltaic curtain wall in a virtual environment.

[0070] Furthermore, step three is achieved through the following process:

[0071] Step 301: Because the photovoltaic curtain wall is encapsulated layer by layer of composite phase-change glass, adhesive film, and photovoltaic modules, and the fiber Bragg grating sensor array is adsorbed on the photovoltaic curtain wall, when the fiber Bragg grating sensor array transmits a pulsed laser signal to the photovoltaic curtain wall via a pulsed laser sensor, the pulsed laser signal sequentially passes through the composite phase-change glass, adhesive film, and photovoltaic modules, generating pulsed reflection signals of varying intensities. This is reflected in the real-time laser image data, resulting in multiple distinct fuzzy boundaries therein, and thus multiple fuzzy boundary regions being simultaneously present in the corresponding generated twin 3D scene model.

[0072] Divide the twin 3D scene model into a number of scene voxels, each of which is a cube and each face of the scene voxel has a corresponding pixel value.

[0073] Randomly select no less than 100 boundary recognition voxels in the fuzzy boundary area of ​​the twin 3D scene model, and select no less than 10 scene voxels around each data node as inward recognition voxels;

[0074] Each inward identification voxel grows in the vertical direction of its fuzzy boundary area, and the boundary identification voxel grows in the vertical direction of the adjacent fuzzy boundary area. Then, each boundary identification voxel and inward identification voxel matches the scene voxels at adjacent positions along the growth direction, bidirectionally or unidirectionally, and obtains the pixel value difference of each surface between the two;

[0075] A pixel difference threshold is set for each fuzzy boundary area. If the pixel value difference of more than four faces is less than or equal to the pixel difference threshold, the scene voxel matched by the boundary recognition voxel or the inward recognition voxel is recorded as the boundary recognition voxel or the inward recognition voxel. Otherwise, the matching operation of the boundary recognition voxel or the inward recognition voxel is stopped.

[0076] When there is a scene voxel that is recorded as both a boundary recognition voxel and an inward recognition voxel, a voxel path is traversed between the corresponding data node and the fuzzy boundary area along the matching path;

[0077] The voxel paths between the fuzzy boundary areas are connected to each other, and the glass model and photovoltaic module model are divided within the curtain wall area model based on the connection results.

[0078] Step 302: Set multiple temperature detection intervals, with each temperature detection interval being 5 degrees Celsius apart, and obtain several sets of historical status data of the photovoltaic scene in each temperature detection interval, the historical status data including historical temperature change curves and historical stress change curves;

[0079] Establish a two-dimensional coordinate system, map several sets of historical status data corresponding to the same photovoltaic curtain wall in the same temperature detection range onto the two-dimensional coordinate system, and then establish multi-segment linear regression equations in different temperature detection ranges based on the historical temperature change curve and the historical stress change curve;

[0080] Several dynamic deformation detection points are randomly set on the glass model and the photovoltaic module model. The positions of the dynamic deformation detection points change synchronously with the acquisition frequency, but the total number remains unchanged.

[0081] The dynamic deformation detection points on the glass model and the photovoltaic module model are connected nearby to obtain a deformation monitoring network, and the multi-segment linear regression equations under each temperature detection area are bound to the dynamic deformation detection points.

[0082] When using, combine the contents in steps 301 to 302:

[0083] By segmenting the glass and PV module models within the twin 3D scene model and setting deformation detection points to create a deformation monitoring network, the system can accurately monitor the deformation of the PV curtain wall's glass and PV modules. This enables timely detection of even minor deformations and helps prevent safety incidents and power generation efficiency losses caused by these deformations.

[0084] Furthermore, the step 4 is achieved through the following process:

[0085] Step 401: Update the twin 3D scene model based on the real-time status data of the photovoltaic scene, and then obtain the real-time temperature and stress values ​​at each deformation detection point based on the thermal patterns and stress point clouds on the glass model and the photovoltaic module model;

[0086] Step 402: retrieve a corresponding multi-segment linear regression equation based on the temperature detection interval of the real-time temperature value and the position of the deformation detection point, and input the real-time temperature value into the multi-segment linear regression equation to obtain an estimated stress value;

[0087] If the estimated stress value is greater than or equal to the real-time stress value, the position of the corresponding deformation detection point is judged to be normal;

[0088] If the estimated stress value is less than the real-time stress value, the location of the corresponding deformation detection point is judged to be abnormal, and the corresponding deformation detection point is recorded as an abnormal deformation detection point. Then, the deformation monitoring network constructs a primary stress analysis network and a secondary stress analysis network with the abnormal deformation detection point as the starting point.

[0089] Step 403: The primary stress analysis network uses the stress direction of the abnormal deformation detection point as the primary stress analysis direction. The primary stress analysis network then extends from the primary stress analysis direction, traversing deformation detection points whose stress directions have angles between (-45° and 45°) and the primary stress analysis direction, or are abnormal deformation detection points, and incorporating the traversal results into the primary stress analysis network until the edges of the glass model and the photovoltaic module model are reached.

[0090] When the first-level stress analysis network traversal is completed, the deformation detection points in the first-level stress analysis network are sequentially connected according to their spatial positions, and the connection results are recorded as the main stress abnormal part;

[0091] The secondary stress analysis network first checks whether there is an abnormality in the deformation detection points at the adjacent spatial positions of the abnormal deformation detection point. If it is judged that there is no abnormality, no operation is performed;

[0092] If it is judged to exist, the stress direction of the abnormal deformation detection point at the adjacent spatial position is used as the main force analysis direction, and the process of incorporating the deformation detection point into the primary stress analysis network is repeated to traverse the secondary stress abnormal part on the twin 3D scene model;

[0093] It should be noted that, since there are multiple deformation detection points in the adjacent spatial positions of the abnormal deformation detection point, there may be multiple main force analysis directions of the secondary stress analysis network;

[0094] The parts where the main stress anomaly and the secondary stress anomaly are located are marked on the twin 3D scene model, and the marking results are synchronized with the maintenance personnel, so that the maintenance personnel can maintain the photovoltaic curtain wall at the corresponding position according to the marking results.

[0095] When using, combine the contents in steps 401 to 403:

[0096] Real-time status data is used to determine whether there are abnormalities in the deformation detection points, and a primary stress analysis network and a secondary stress analysis network are generated based on the abnormal conditions. While intelligently analyzing the stress distribution of the photovoltaic curtain wall, the primary stress abnormality and the secondary stress abnormality are accurately found, providing a scientific basis for the maintenance and repair of the photovoltaic curtain wall, and improving maintenance efficiency and effectiveness.

[0097] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A real-time safety monitoring method for a large-span composite phase-change glass photovoltaic curtain wall, characterized in that: The following steps are involved: Step 1: Set up a sensor network in the photovoltaic scene and collect real-time status data of each photovoltaic curtain wall in the photovoltaic scene through the sensor network; Step 2: Build a scene framework model of the photovoltaic scene based on the real-time status data of each photovoltaic curtain wall, mark several data nodes in the scene framework model, and then map all real-time status data onto the scene framework model based on the location distribution of the data nodes to obtain a twin 3D scene model; The process of establishing a twin 3D scene model includes: With the location of each data node as the center and the distance to the adjacent data node as the radius, the real-time infrared image data and real-time regional stress change curve collected by each data node are mapped onto the curtain wall area model where the data node is located, thereby obtaining a twin 3D scene model. Step 3: Segment the glass model and the photovoltaic module model within the twin 3D scene model, set several deformation detection points on the glass model and the photovoltaic module model, and then generate a deformation monitoring network for the glass model and the photovoltaic module model based on the deformation detection points; Step 4: Update the twin 3D scene model using the real-time status data of the photovoltaic curtain wall. Use the real-time status data to determine whether there are any anomalies at the deformation detection points. The deformation monitoring network generates a primary stress analysis network and a secondary stress analysis network based on the abnormal deformation detection points. Then, through the primary and secondary stress analysis networks, traverse the twin 3D scene model to identify the abnormal primary and secondary stress areas. The division process of the glass model and the photovoltaic module model includes: Mark multiple fuzzy boundary areas in the twin 3D scene model and divide the twin 3D scene model into several scene voxels. Randomly select no less than 100 boundary identification voxels in the fuzzy boundary areas in the twin 3D scene model, and select no less than 10 scene voxels around each data node as inward identification voxels. Each inward identification voxel grows in the vertical direction of its fuzzy boundary area, and the boundary identification voxel grows in the vertical direction of the adjacent fuzzy boundary area. Then, each boundary identification voxel and inward identification voxel matches the scene voxels at adjacent positions along the growth direction, bidirectionally or unidirectionally, and obtains the pixel value difference of each surface between the two; A pixel difference threshold is set for each fuzzy boundary area. If the pixel value difference of more than four faces is less than or equal to the pixel difference threshold, the scene voxel matched by the boundary recognition voxel or the inward recognition voxel is recorded as the boundary recognition voxel or the inward recognition voxel. Otherwise, the matching operation of the boundary recognition voxel or the inward recognition voxel is stopped. When a scene voxel is recorded as both a boundary identification voxel and an inward identification voxel, a voxel path is traversed between the corresponding data node and the fuzzy boundary area along the matching path. The voxel paths between the fuzzy boundary areas are connected to each other. Based on the connection results, the glass model and the photovoltaic module model are divided within the curtain wall area model. The process of establishing a deformation monitoring network includes: Set multiple temperature detection intervals and obtain several copies of historical status data of photovoltaic scenes in each temperature detection interval. Establish a two-dimensional coordinate system, map the historical status data of the same temperature detection interval and corresponding to the same photovoltaic curtain wall on the two-dimensional coordinate system, and establish multi-segment linear regression equations in different temperature detection intervals based on the historical temperature change curve and the historical stress change curve. A number of dynamic deformation detection points are randomly set on the glass model and the photovoltaic module model, and the dynamic deformation detection points on the glass model and the photovoltaic module model are connected nearby to obtain a deformation monitoring network; The process of constructing the primary stress analysis network and the secondary stress analysis network through the deformation monitoring network includes: According to the temperature detection range of the real-time temperature value and the position of the deformation detection point, the corresponding multi-segment linear regression equation is retrieved, and the real-time temperature value is input into the multi-segment linear regression equation to obtain the estimated stress value; If the estimated stress value is greater than or equal to the real-time stress value, the location of the corresponding deformation detection point is judged to be normal. If the estimated stress value is less than the real-time stress value, the location of the corresponding deformation detection point is judged to be abnormal, and the corresponding deformation detection point is recorded as an abnormal deformation detection point. Then, the deformation monitoring network constructs a primary stress analysis network and a secondary stress analysis network with the abnormal deformation detection point as the starting point; The process of traversing the main stress anomaly part and the secondary stress anomaly part on the twin 3D scene model includes: The primary stress analysis network uses the stress direction of the abnormal deformation detection point as the main force analysis direction. The primary stress analysis network then extends from the main force analysis direction, traversing deformation detection points whose stress directions have an angle between (-45° and 45°) and the main force analysis direction, or are also abnormal deformation detection points, and incorporating the traversal results into the primary stress analysis network until the edge positions of the glass model and the photovoltaic module model are reached; When the first-level stress analysis network traversal is completed, the deformation detection points in the first-level stress analysis network are sequentially connected according to their spatial positions, and the connection results are recorded as the main stress abnormal part; The secondary stress analysis network first analyzes whether there is an abnormality in the deformation detection points in the adjacent spatial positions of the abnormal deformation detection point. If it is determined that there is no abnormality, no operation is performed; If it is judged to exist, the stress direction of the abnormal deformation detection point at the adjacent spatial position is used as the main force analysis direction, and then the process of incorporating the first-level stress analysis network into the deformation detection point is repeated to traverse the secondary stress abnormal part on the twin three-dimensional scene model.

2. A method for real-time safety monitoring of a large-span composite phase-change glass photovoltaic curtain wall according to claim 1, characterized in that: The sensor network setup process includes: Install n fiber Bragg grating sensor arrays in a photovoltaic scene, where n is a natural number greater than 10; The fiber Bragg grating sensor array is composed of an infrared laser sensor, a pulse laser sensor, a stress sensor, a wireless communication device and a stress-sensitive sheet; Data nodes are set according to the positions of wireless communication devices of the fiber grating sensor array, and adjacent data nodes are connected to each other according to the spatial position relationship between the data nodes to obtain a sensor network.

3. The method for real-time safety monitoring of a large-span composite phase-change glass photovoltaic curtain wall according to claim 2, characterized in that: The process of collecting real-time status data of photovoltaic curtain walls includes: The same acquisition frequency is set for each fiber Bragg grating sensor array. Before the photovoltaic scene is put into use, the laser image data of each photovoltaic curtain wall is collected through the fiber Bragg grating sensor array. When the photovoltaic scene is put into use, each fiber grating sensor array collects various real-time status data through wireless communication devices at each collection frequency. The real-time status data includes real-time infrared image data, real-time laser image data and real-time regional stress change curve.

4. The method for real-time safety monitoring of a large-span composite phase-change glass photovoltaic curtain wall according to claim 1, characterized in that: The construction process of the scene framework model includes: Retrieve laser image data of the photovoltaic scene before it is put into use, and splice the corresponding laser image data according to the position of each data node in the sensor network to obtain the scene laser image data before the photovoltaic scene is put into use; Then, a scene framework model is constructed based on the scene laser image data before the photovoltaic scene is put into use. According to the position distribution of each photovoltaic curtain wall in the photovoltaic scene, m curtain wall area models are divided in the scene framework model, where m is a positive integer greater than 0. At the same time, according to the spatial position of the fiber Bragg grating sensor array in the photovoltaic scene, several data nodes are marked on each curtain wall area model. After the photovoltaic scene is put into use, each curtain wall area model is dynamically updated based on the real-time laser image data verified by each data node after each synchronization data verification period.

Images