A building intelligent early warning method based on 5G Internet of Things integrated video surveillance

Abstract

The present invention relates to the field of building alarm technology, and in particular to a building intelligent early warning method based on 5G Internet of Things integrated video monitoring, wherein first load data of a building scaffold is obtained through an Internet of Things pressure sensor and standardized to obtain second load data of the building scaffold, the second load data of the building scaffold is loaded into a load analysis model to obtain global load assessment data, risk determination is further performed to obtain risk partitions of the building scaffold, and risk nodes in the risk partitions are prioritized to obtain a list of nodes to be optimized, risk partition videos are obtained and the position coordinates of personnel in the area are located through an image target detection algorithm, risk nodes in the list of nodes to be optimized are matched with the nearest personnel through a risk node path matching model and a load adjustment instruction is sent, and when the personnel do not move to the designated risk nodes, an alarm is issued and the matching personnel are adjusted, thereby realizing effective identification and management of uneven loads on building scaffolds.

Description

Technical Field

[0001] The present invention relates to the field of building alarm technology, and in particular to a building intelligent early warning method based on 5G Internet of Things integrated video monitoring. Background Art

[0002] During the construction process, temporary tools, materials and other items are often placed on the scaffolding. Because on-site construction workers are used to placing them randomly or for easy access, the load distribution is often uneven. This uneven load will cause excessive force on the local part of the scaffolding, thereby weakening the stability of the overall structure, and easily cause serious safety hazards such as overturning and collapse, threatening the safety of personnel and equipment on the construction site. The traditional scaffolding management method relies on the workers' experience and regular inspections. This method is not only difficult to detect the problem of uneven load in time, but also cannot provide effective measures to analyze and optimize the distribution of loads. It is difficult to achieve real-time and accurate load management, and it is impossible to effectively avoid accidents caused by load imbalance. Therefore, how to monitor and effectively manage the scaffolding in the construction process in real time has become an urgent problem to be solved. Summary of the invention

[0003] (1) Technical issues to be resolved

[0004] The present invention provides a building intelligent early warning method based on 5G Internet of Things integrated video monitoring, which solves the problem that it is difficult to timely discover and take effective management measures when uneven loads on building scaffolding occur during construction, thereby causing load imbalance accidents.

[0005] (2) Technical solution

[0006] To achieve the above object, the present invention provides a building intelligent early warning method based on 5G Internet of Things integrated video monitoring, the method comprising:

[0007] The first load data of the building scaffolding is obtained in real time by installing an Internet of Things pressure sensor at a preset key node of the building scaffolding, the first load data of the building scaffolding is standardized to obtain the second load data of the building scaffolding, and the second load data of the building scaffolding is transmitted to the central database through the 5G network for storage.

[0008] The second load data of the building scaffolding is retrieved from the central database at a preset time interval, the second load data of the building scaffolding at the same moment is loaded into the load analysis model to obtain the global load assessment data, risk determination is performed based on the global load assessment data to obtain the risk partition of the building scaffolding, and the risk nodes in the risk partition of the building scaffolding are prioritized to obtain a list of nodes to be optimized.

[0009] The risk zone video data collected by the risk zone camera of the building scaffolding is obtained. According to the risk zone video data, personnel are identified and the coordinates of the personnel in the area are located through the image target detection algorithm. Through the risk node path matching model, each risk node in the node list to be optimized is matched with the nearest person and a load adjustment instruction is sent to the person.

[0010] Monitor the changes in risk node load data. When the risk node load data does not decrease for more than the preset time, an alarm will be issued to the person matching the risk node. The person's displacement will be monitored by video to determine whether he moves toward the risk node. When the person's position coordinates move toward the risk node and coincide with the risk node coordinates, the alarm will be stopped. When the person's position coordinates do not move toward the designated risk node, the risk node will be matched with the second closest person and a load adjustment instruction will be sent to the person.

[0011] Further, the method of retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to obtain global load assessment data, performing risk determination according to the global load assessment data to obtain risk partitions of the building scaffolding, and prioritizing the risk nodes in the risk partitions of the building scaffolding to obtain a list of nodes to be optimized includes:

[0012] The second load data of the building scaffolding is retrieved from the central database at a preset time interval, and the second load data of the building scaffolding at the same time is loaded into the load analysis model to perform single-node load evaluation and regional load evaluation to obtain the node overload and regional load balance.

[0013] The regional risk coefficient is obtained based on the node overload and regional load balance in the region. for:

[0014] .

[0015] in It is Regional load balancing, is the upper limit of node load, It is In the region The node overload of nodes, is the overload adjustment factor, It is The total number of nodes in the region, It is The regional risk factor of a region.

[0016] The risk zoning of building scaffolding is obtained by comparing the regional risk coefficient with the preset risk determination threshold, including:

[0017] When the regional risk coefficient is greater than or equal to the preset risk determination threshold, the corresponding area is marked as a high-risk area.

[0018] When the regional risk coefficient is less than the preset risk determination threshold, the corresponding area is marked as a low-risk area.

[0019] The risk nodes in the risk partition of the building scaffolding marked as high-risk areas are sorted from high to low according to the size of the node overload to obtain a list of nodes to be optimized.

[0020] Furthermore, the method of loading the second load data of the building scaffold at the same time into the load analysis model to perform single node load evaluation and regional load evaluation to obtain the node overload and regional load balance includes:

[0021] According to the second load data of the building scaffolding at the same time, the node division is carried out through spatial adjacency analysis to obtain the scaffolding area.

[0022] According to the single node load value and the node load upper limit value in the scaffolding area, the node load is evaluated to obtain the node overload value. for:

[0023] .

[0024] in It is In the region The node load value of each node.

[0025] The regional load balance is obtained by evaluating the regional load according to the load values ​​of all nodes in the scaffolding area. for:

[0026] .

[0027] in It is In the region The node load value of each node, It is The total number of nodes in the region.

[0028] Furthermore, the method for obtaining the scaffolding area by performing node division through spatial adjacency analysis according to the second load data of the building scaffolding at the same time includes:

[0029] The scaffolding node coordinates are extracted from the second load data of the building scaffolding, and the spatial distances between the scaffolding nodes are calculated through the Euclidean distance according to the scaffolding node coordinates.

[0030] When the spatial distance between two scaffolding nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes. An adjacency matrix is ​​constructed based on the adjacent nodes and node connection edges.

[0031] The scaffolding region is obtained by traversing the adjacency matrix using the connected component algorithm.

[0032] Furthermore, when the spatial distance between two scaffolding nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes. The method for constructing an adjacency matrix according to the adjacent nodes and the node connection edges includes:

[0033] Create a size based on the total number of scaffolding nodes The zero matrix of .

[0034] When the spatial distance between the nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes.

[0035] The adjacency matrix is ​​obtained by updating the corresponding elements in the zero matrix according to the node connection edges, specifically:

[0036] .

[0037] in is the zero matrix Line The numeric value of the column elements.

[0038] Furthermore, the method of obtaining the scaffolding area by traversing the adjacency matrix through the connected component algorithm includes:

[0039] Mark all scaffold nodes as unvisited nodes and create a region set.

[0040] Select an unvisited node, mark the node as visited and add it to the first region of the region set, start from the node and traverse the adjacency matrix through depth-first search to obtain connected nodes, mark the connected nodes as visited and add them to the region set to obtain the first region.

[0041] The next unvisited node is selected from the remaining unvisited nodes, the node is marked as visited and added to the second region of the region set, and the adjacency matrix is ​​traversed from the node through depth-first search to obtain the connected nodes, the connected nodes are marked as visited and added to the region set to obtain the second region.

[0042] Repeat the above traversal process until all unvisited nodes are traversed and divided into one area of ​​the area set to obtain an updated area set as the scaffolding area.

[0043] Furthermore, the method of obtaining risk zone video data collected by a risk zone camera of a building scaffold, identifying personnel and locating the position coordinates of personnel in the area through an image target detection algorithm according to the risk zone video data, matching each risk node in the node list to be optimized with the nearest personnel through a risk node path matching model and sending a load adjustment instruction to the personnel includes:

[0044] The risk zone video data collected by the risk zone camera of the building scaffolding is obtained, and the YOLO algorithm is used to identify the people in the video and locate the coordinates of the people in the area.

[0045] Get the coordinates of the first risk node in the list of nodes to be optimized, calculate the path distance between each person and the first risk node using the Euclidean distance method, and select the person with the shortest path distance to the first risk node as the first target person.

[0046] Obtain the coordinates of the second risk node in the list of nodes to be optimized, calculate the path distance between each person except the first target person and the second risk node by the Euclidean distance method, and select the person with the shortest path distance to the second risk node as the second target person.

[0047] Obtain each risk node in turn according to the list of nodes to be optimized and repeat the above path distance calculation and personnel selection until all risk nodes are matched to the target personnel and then stop risk node path matching.

[0048] Load adjustment instructions are generated based on regional load balance and node overload and sent to target personnel through 5G Internet of Things.

[0049] Furthermore, the method of obtaining video data collected by a camera in a risk zone of a building scaffold, identifying people in the video and locating the coordinates of the people in the area by using a YOLO algorithm includes:

[0050] The video captured by the building scaffolding risk partition camera is obtained and the single-frame image is extracted to obtain the scaffolding risk partition image. The scaffolding risk partition image is used to detect the target of the personnel through the YOLO algorithm to obtain the personnel image coordinates.

[0051] A homography matrix from image coordinates to spatial coordinates is established by the least squares method according to the spatial coordinates of the spatial preset reference point and the coordinates of the spatial preset reference point in the image.

[0052] The person image coordinates are converted into person space coordinates as the person position coordinates according to the person image coordinates and the homography matrix from the image coordinates to the space coordinates.

[0053] Furthermore, the method of determining whether the person moves toward a risk node by monitoring the person's displacement through video includes:

[0054] The distance between a person and the risk node they match is calculated based on the person's location coordinates and the risk node coordinates.

[0055] When the distance shows a decreasing trend within a preset time, it is judged that the person has moved towards the risk node, otherwise it is judged that the person has not moved towards the risk node.

[0056] Further, generating a load adjustment instruction according to the regional load balance degree and the node overload includes:

[0057] The load of the overloaded node is transferred at a level not lower than the overload amount, so that the node load value is reduced to within the node load upper limit.

[0058] Nodes with smaller load values ​​are selected first, and load transfer is performed according to the adjustment rule that the overload of each node after load transfer is 0 and close.

[0059] (3) Beneficial effects

[0060] Compared with the prior art, the present invention has the following beneficial effects: scaffolding area division is achieved by node division through spatial adjacency analysis, risk zoning is achieved by single-node load evaluation and regional load evaluation of each area through a load analysis model, personnel detection is performed on risk zone videos through an image target detection algorithm to achieve personnel positioning, and risk nodes are further matched with the nearest personnel through a risk node path matching model to achieve efficient load allocation, thereby achieving effective identification and management of uneven loads on building scaffolding as a whole. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0062] Figure 1 This is a flowchart of a building intelligent early warning method based on 5G Internet of Things integrated video monitoring according to Embodiment 1 of the present invention; DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] Example 1: Figure 1 As shown, this embodiment provides a building intelligent early warning method based on 5G Internet of Things integrated video surveillance, the method comprising:

[0065] S1. Obtain the first load data of the building scaffolding in real time through the Internet of Things pressure sensor installed at the preset key node of the building scaffolding, standardize the first load data of the building scaffolding to obtain the second load data of the building scaffolding, and transmit the second load data of the building scaffolding to the central database through the 5G network for storage; the standardization processing is to clean the first load data of the building scaffolding to remove noise through the commonly used data preprocessing method, and combine the processed data with the corresponding sensor position data to form structured data including load value, coordinate value and timestamp as the second load data of the building scaffolding.

[0066] S2. Retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to obtain global load assessment data, performing risk determination based on the global load assessment data to obtain risk partitions of the building scaffolding, and prioritizing the risk nodes in the risk partitions of the building scaffolding to obtain a list of nodes to be optimized; the retrieving the second load data of the building scaffolding from the central database, for example, uses the database query language SQL to extract the second load data of the building scaffolding from the central database every five minutes, and the retrieved data includes load value, coordinate value and timestamp information.

[0067] S3. Obtain the risk zone video data collected by the risk zone camera of the building scaffolding, identify the personnel and locate the coordinates of the personnel in the area through the image target detection algorithm based on the risk zone video data, and match each risk node in the node list to be optimized with the nearest personnel through the risk node path matching model and send load adjustment instructions to them.

[0068] S4. Monitor the changes in the load data of the risk node. When the load data of the risk node does not decrease for more than the preset time, an alarm is issued to the personnel matched with the risk node. The displacement of the personnel is monitored by video to determine whether it moves to the risk node. When the personnel position coordinates move to the risk node and coincide with the risk node coordinates, the alarm is stopped. When the personnel position coordinates do not move to the designated risk node, the risk node is matched with the second closest person and a load adjustment instruction is sent to the person. By real-time monitoring of the personnel response, if the nearest person does not respond within the time limit, the person is switched to ensure that each risk node can get a timely response from the personnel and adjust the load.

[0069] It should be noted that the method of retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to obtain global load assessment data, performing risk determination according to the global load assessment data to obtain risk partitions of the building scaffolding, and prioritizing the risk nodes in the risk partitions of the building scaffolding to obtain a list of nodes to be optimized includes:

[0070] S21. Retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to perform single-node load evaluation and regional load evaluation to obtain the node overload and regional load balance.

[0071] S22. Obtain the regional risk coefficient based on the node overload in the region and the regional load balance for:

[0072] .

[0073] in It is Regional load balancing, is the upper limit of node load, It is In the region The node overload of nodes, is the overload adjustment factor, It is The total number of nodes in the region, It is The regional risk factor of the region; It is the maximum load value that can be achieved in normal use according to the load-bearing standard of scaffolding components. Indicates The overload coefficient of each node. The value of the overload coefficient is positive or 0. The larger the value, the more serious the overload. The overload adjustment factor is used to control the impact of node overload on the risk factor. The larger the node overload, the greater the impact on the risk factor. It can be set based on historical experience. For example, when , it means that the regional load balance and node overload have equal importance in affecting the risk factor.

[0074] S23. Compare the regional risk coefficient with the preset risk determination threshold to obtain the risk zoning of the building scaffolding, including:

[0075] When the regional risk coefficient is greater than or equal to the preset risk determination threshold, the corresponding area is marked as a high-risk area.

[0076] When the regional risk coefficient is less than the preset risk determination threshold, the corresponding area is marked as a low-risk area. The preset risk determination threshold is a value set based on historical experience to measure the size of regional risk.

[0077] S24. Sort the risk nodes in the risk partition of the building scaffolding marked as a high-risk area from high to low according to the size of the node overload to obtain a list of nodes to be optimized.

[0078] For example: Assume that there are two areas in the scaffolding area, area 1 contains 2 nodes, and the node overloads are =0, =1, load balancing , Region 2 contains 3 nodes, and the node overloads are =0, =3, =5, load balancing , when the overload adjustment factor is set to , the upper limit of node load Set to When the regional risk factor of region 1 is , the regional risk factor of region 2 When the risk judgment threshold is set to 2, since the regional risk coefficient of area 2 is greater than 2, area 2 is marked as a high-risk area, and the regional risk coefficient of area 1 is less than 2, area 1 is marked as a low-risk area. The risk nodes in area 2 are further sorted from high to low according to the size of the node overload, and the list of nodes to be optimized is: node 5, node 4, node 3. It should be noted that although node 3 is not overloaded, it may affect the load balancing degree, so it also needs to be considered in the list of nodes to be optimized.

[0079] It should be noted that the method of loading the second load data of the building scaffolding at the same time into the load analysis model to perform single node load evaluation and regional load evaluation to obtain the node overload and regional load balance includes:

[0080] S211, performing node division through spatial adjacency analysis according to the second load data of the building scaffold at the same time to obtain a scaffold area.

[0081] S212: Perform a single node load evaluation on the node according to the single node load value and the node load upper limit value in the scaffolding area to obtain the node overload value. for:

[0082] .

[0083] in It is In the region The node load value of each node.

[0084] S213: Perform regional load evaluation based on the load values ​​of all nodes in the scaffolding area to obtain the regional load balance for:

[0085] .

[0086] in It is In the region The node load value of each node, It is The total number of nodes in the region; represents the regional load average, Indicates the degree of dispersion of the load in the area. The smaller the dispersion, the greater the regional load balance and the more balanced the load.

[0087] For example: There are two areas in the scaffolding area. Area 1 contains 2 nodes, and the load values ​​of the nodes are , , Region 2 contains 3 nodes, and the load values ​​of the nodes are , , , when the node load upper limit When set to 7, a single node load evaluation is performed on region 1 to obtain =0, =1, load balancing degree of area 1 = , a single node load evaluation is performed on region 2 to obtain =0, =3, =5, load balancing degree of area 2 .

[0088] It should be noted that the method of performing node division to obtain the scaffolding area by spatial adjacency analysis according to the second load data of the building scaffolding at the same time includes:

[0089] S2111. Extract scaffolding node coordinates from the second load data of the building scaffolding, and calculate the spatial distance between the scaffolding nodes through the Euclidean distance according to the scaffolding node coordinates.

[0090] S2112. When the spatial distance between two scaffolding nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes, and an adjacency matrix is ​​constructed based on the adjacent nodes and the node connection edges; the adjacent distance threshold is set according to the spatial distribution structure of the scaffolding nodes, and is used to determine whether the nodes are adjacent nodes, ensuring that the spatial distance between adjacent nodes is within a reasonable range. For example, the average spacing distance between scaffolding nodes can be set as the adjacent distance threshold.

[0091] S2113. Obtain the scaffolding region by traversing the adjacency matrix through the connected component algorithm.

[0092] It should be noted that when the spatial distance between two scaffolding nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes. The method for constructing an adjacency matrix according to the adjacent nodes and the node connection edges includes:

[0093] S21121. Create a node with the size of The zero matrix of .

[0094] S21122. When the spatial distance between the nodes is less than a preset adjacency distance threshold, two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes.

[0095] S21123. Update the corresponding elements in the zero matrix according to the node connection edges to obtain an adjacency matrix, specifically:

[0096] .

[0097] in is the zero matrix Line The numeric value of the column elements.

[0098] For example: If the total number of scaffolding nodes is , then the zero matrix is , assuming that the node and nodes The spatial distance between them is less than the preset adjacency distance threshold, that is, ,but , then the updated adjacency matrix is .

[0099] It should be noted that the method of traversing the adjacency matrix through the connected component algorithm to obtain the scaffolding area includes:

[0100] S21131. Mark all scaffold nodes as unvisited nodes and create a region set.

[0101] S21132. Select an unvisited node, mark the node as visited and add it to the first region of the region set, traverse the adjacency matrix from the node through depth-first search to obtain connected nodes, mark the connected nodes as visited and add them to the region set to obtain the first region.

[0102] S21133. Select the next unvisited node from the remaining unvisited nodes, mark the node as visited and add it to the second region of the region set, traverse the adjacency matrix from the node through depth-first search to obtain connected nodes, mark the connected nodes as visited and add them to the region set to obtain the second region.

[0103] S21134. Repeat the above traversal process until all unvisited nodes are traversed and divided into one area of ​​the area set to obtain an updated area set as the scaffolding area.

[0104] For example: If the adjacency matrix is , mark all scaffolding nodes as unvisited nodes and create a region set, select nodes , marking the node as visited and adding it to the first region of the region set, from node Start by traversing the adjacency matrix through a depth-first search to get nodes Connected nodes , get the first region of the region set , select a node from the remaining unvisited nodes , the node Mark as visited and add to the second region of the region set, from node Start by traversing the adjacency matrix through a depth-first search to get nodes Connected nodes , the node Add to the zone set The second zone continues to access the node With Node Connect the nodes Add to the second region of the region set to get the second region of the region set , so that the regional set , that is, the scaffolding area is composed of nodes and nodes The first region and node formed ,node and nodes The second region is formed.

[0105] It should be noted that the method of obtaining risk zone video data collected by the risk zone camera of the building scaffolding, identifying personnel and locating the position coordinates of personnel in the area through the image target detection algorithm according to the risk zone video data, matching each risk node in the node list to be optimized with the nearest personnel through the risk node path matching model and sending load adjustment instructions to them includes:

[0106] S31. Obtain risk zone video data collected by the risk zone camera of the building scaffolding, identify the personnel in the video through the YOLO algorithm and locate the position coordinates of the personnel in the area.

[0107] S32. Obtain the coordinates of the first risk node in the list of nodes to be optimized, calculate the path distance between each person and the first risk node by using the Euclidean distance method, and select the person with the shortest path distance to the first risk node as the first target person.

[0108] S33. Obtain the coordinates of the second risk node in the list of nodes to be optimized, calculate the path distance between each person except the first target person and the second risk node by the Euclidean distance method, and select the person with the shortest path distance to the second risk node as the second target person.

[0109] S34. Obtain each risk node in turn according to the list of nodes to be optimized and repeat the above path distance calculation and personnel selection until all risk nodes are matched to the target personnel, and then stop the risk node path matching.

[0110] S35. Generate a load adjustment instruction based on the regional load balance and node overload and send it to each target person through the 5G Internet of Things.

[0111] It should be noted that the method of obtaining the video data collected by the camera of the building scaffolding risk zone, identifying the personnel in the video and locating the position coordinates of the personnel in the area by using the YOLO algorithm includes:

[0112] S311, obtaining the video captured by the building scaffolding risk partition camera and extracting the single-frame image to obtain the scaffolding risk partition image, and performing target detection on the scaffolding risk partition image through the YOLO algorithm to obtain the image coordinates of the person.

[0113] S312. Establishing a homography matrix from image coordinates to spatial coordinates by the least squares method according to the spatial coordinates of the spatial preset reference points and the coordinates of the spatial preset reference points in the image; establishing a homography matrix from image coordinates to spatial coordinates by the least squares method according to the spatial coordinates of the spatial preset reference points and the coordinates of the spatial preset reference points in the image is a common method for converting image coordinates into physical spatial coordinates in the image processing industry. The spatial preset reference points can be selected and marked according to fixed visual markers, fixed equipment, etc. that can be captured by the camera. For example, the Internet of Things pressure sensors at the preset key nodes of the building scaffolding can be used as spatial preset reference points.

[0114] S313 . Convert the person image coordinates into the person space coordinates as the person position coordinates according to the person image coordinates and the homography matrix from the image coordinates to the space coordinates.

[0115] It should be noted that the method of determining whether the person moves toward the risk node by monitoring the person's displacement through video includes:

[0116] S41. Calculate the distance between the person and the risk node that matches the person according to the person's location coordinates and the risk node coordinates.

[0117] S42. When the distance shows a decreasing trend within a preset time, it is determined that the person has moved toward the risk node; otherwise, it is determined that the person has not moved toward the risk node.

[0118] It should be noted that the generating of the load adjustment instruction according to the regional load balance degree and the node overload includes:

[0119] S351. Transfer the load of the overloaded node at a level not lower than the overload amount, so that the node load value is reduced to within the node load upper limit value.

[0120] S352. Nodes with smaller load values ​​are selected preferentially, and load transfer is performed according to the adjustment rule that the overload of each node after load transfer is 0 and close.

[0121] For example, suppose there are three nodes in the list of nodes to be optimized: node A has a load value of 9, node B has a load value of 6, and node C has a load value of 4. The load limit of the node is 7. , the load of overloaded node A needs to be transferred at a rate not less than 2. Since the node with the smallest load value is node C, node C is selected first for load transfer. According to the adjustment rule that the overload of each node is 0 and similar after the load transfer, node A can transfer 2.5 loads to node C. After the transfer, the load value of node A is 6.5, and the load value of node C is 6.5. The overload of the nodes is 0 and similar.

[0122] Finally, it should be noted that: Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A building intelligent early warning method based on 5G Internet of Things integrated video monitoring, characterized in that: The method comprises: The first load data of the building scaffolding is obtained in real time by installing an IoT pressure sensor at a preset key node of the building scaffolding, the first load data of the building scaffolding is standardized to obtain the second load data of the building scaffolding, and the second load data of the building scaffolding is transmitted to a central database through a 5G network for storage; Retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to obtain global load assessment data, performing risk determination based on the global load assessment data to obtain risk partitions of the building scaffolding, and prioritizing the risk nodes in the risk partitions of the building scaffolding to obtain a list of nodes to be optimized; Obtain risk zone video data collected by the risk zone camera of the building scaffolding, identify personnel and locate the coordinates of personnel in the area through the image target detection algorithm based on the risk zone video data, and match each risk node in the node list to be optimized with the nearest personnel through the risk node path matching model and send load adjustment instructions to them; Monitor the changes in the load data of the risk node. When the load data of the risk node does not decrease for more than the preset time, an alarm will be issued to the person matching the risk node. The displacement of the person will be monitored by video to determine whether he moves toward the risk node. When the person's position coordinates move toward the risk node and coincide with the risk node coordinates, the alarm will be stopped. When the person's position coordinates do not move toward the designated risk node, the risk node will be matched with the second closest person and a load adjustment instruction will be sent to the person. The method of retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to obtain global load assessment data, performing risk determination according to the global load assessment data to obtain risk partitions of the building scaffolding, and prioritizing the risk nodes in the risk partitions of the building scaffolding to obtain a list of nodes to be optimized includes: Retrieving the second load data of the building scaffolding from the central database at a preset time interval, loading the second load data of the building scaffolding at the same time into the load analysis model to perform single node load evaluation and regional load evaluation to obtain the node overload and regional load balance; The regional risk coefficient is obtained based on the node overload and regional load balance in the region. for: ； in It is Regional load balancing, is the upper limit of node load, It is In the region The node overload of nodes, is the overload adjustment factor, It is The total number of nodes in the region, It is The regional risk factor of each region; The risk zoning of building scaffolding is obtained by comparing the regional risk coefficient with the preset risk determination threshold, including: When the regional risk coefficient is greater than or equal to the preset risk determination threshold, the corresponding area is marked as a high-risk area; When the regional risk coefficient is less than the preset risk determination threshold, the corresponding area is marked as a low-risk area; The risk nodes in the risk partition of the building scaffolding marked as high-risk areas are sorted from high to low according to the size of the node overload to obtain a list of nodes to be optimized.

2. According to claim 1, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method of loading the second load data of the building scaffold at the same time into the load analysis model to perform single node load evaluation and regional load evaluation to obtain the node overload and regional load balance includes: According to the second load data of the building scaffold at the same time, node division is performed through spatial adjacency analysis to obtain the scaffold area; According to the single node load value and the node load upper limit value in the scaffolding area, the node load is evaluated to obtain the node overload value. for: ； in It is In the region The node load value of each node; The regional load balance is obtained by evaluating the regional load according to the load values ​​of all nodes in the scaffolding area. for: ； in It is In the region The node load value of each node, It is The total number of nodes in the region.

3. According to claim 2, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method for obtaining a scaffolding area by performing node division through spatial adjacency analysis according to the second load data of the building scaffolding at the same time comprises: Extracting scaffolding node coordinates from the second load data of the building scaffolding, and calculating the spatial distance between the scaffolding nodes by Euclidean distance according to the scaffolding node coordinates; When the spatial distance between two scaffolding nodes is less than a preset adjacency distance threshold, the two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes, and an adjacency matrix is ​​constructed according to the adjacent nodes and node connection edges; The scaffolding region is obtained by traversing the adjacency matrix using the connected component algorithm.

4. According to claim 3, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method of marking two scaffolding nodes as adjacent nodes and generating node connection edges between the nodes when the spatial distance between the two scaffolding nodes is less than a preset adjacency distance threshold, and constructing an adjacency matrix according to the adjacent nodes and the node connection edges includes: Create a size based on the total number of scaffolding nodes The zero matrix of ; When the spatial distance between the nodes is less than a preset adjacency distance threshold, two scaffolding nodes are marked as adjacent nodes and node connection edges are generated between the nodes; The adjacency matrix is ​​obtained by updating the corresponding elements in the zero matrix according to the node connection edges, specifically: ； in is the zero matrix Line The numeric value of the column elements.

5. According to claim 3, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method of traversing the adjacency matrix through the connected component algorithm to obtain the scaffolding area includes: Mark all scaffold nodes as unvisited nodes and create region sets; Select an unvisited node, mark the node as visited and add it to the first region of the region set, start from the node and traverse the adjacency matrix through depth-first search to obtain connected nodes, mark the connected nodes as visited and add them to the region set to obtain the first region; Select the next unvisited node from the remaining unvisited nodes, mark the node as visited and add it to the second region of the region set, start from the node and traverse the adjacency matrix through depth-first search to obtain connected nodes, mark the connected nodes as visited and add them to the region set to obtain the second region; Repeat the above traversal process until all unvisited nodes are traversed and divided into one area of ​​the area set to obtain an updated area set as the scaffolding area.

6. According to claim 2, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method of obtaining risk zone video data collected by a risk zone camera of a building scaffold, identifying personnel and locating the position coordinates of personnel in the area through an image target detection algorithm according to the risk zone video data, matching each risk node in the node list to be optimized with the nearest personnel through a risk node path matching model and sending a load adjustment instruction to the personnel includes: Obtain risk zone video data collected by the risk zone camera of the building scaffolding, identify the people in the video and locate the coordinates of the people in the area through the YOLO algorithm; Obtain the coordinates of the first risk node in the list of nodes to be optimized, calculate the path distance between each person and the first risk node using the Euclidean distance method, and select the person with the shortest path distance to the first risk node as the first target person; Obtain the coordinates of the second risk node in the list of nodes to be optimized, calculate the path distance between each person except the first target person and the second risk node by the Euclidean distance method, and select the person with the shortest path distance to the second risk node as the second target person; Obtain each risk node in turn according to the list of nodes to be optimized and repeat the above path distance calculation and personnel selection until all risk nodes are matched to the target personnel and then stop risk node path matching; Load adjustment instructions are generated based on regional load balance and node overload and sent to target personnel through 5G Internet of Things.

7. According to claim 6, a building intelligent early warning method based on 5G Internet of Things integrated video monitoring is characterized in that: The method of obtaining video data collected by a camera for risk partitioning of building scaffolding, identifying personnel in the video and locating the position coordinates of personnel in the area by using a YOLO algorithm includes: The video captured by the building scaffolding risk partition camera is obtained and a single frame image is extracted to obtain the scaffolding risk partition image. The scaffolding risk partition image is subjected to target detection by the YOLO algorithm to obtain the image coordinates of the person. A homography matrix from image coordinates to spatial coordinates is established by least square method according to the spatial coordinates of the spatial preset reference point and the coordinates of the spatial preset reference point in the image; The person image coordinates are converted into person space coordinates as the person position coordinates according to the person image coordinates and the homography matrix from the image coordinates to the space coordinates.

8. The intelligent building early warning method based on 5G Internet of Things integrated video monitoring according to claim 7 is characterized in that: The method of determining whether the person moves toward a risk node by monitoring the person's displacement through video includes: Calculate the distance between the person and the risk node that matches him based on the person's location coordinates and the risk node coordinates; When the distance shows a decreasing trend within a preset time, it is judged that the person has moved towards the risk node, otherwise it is judged that the person has not moved towards the risk node.

9. The intelligent building early warning method based on 5G Internet of Things integrated video monitoring according to claim 6 is characterized in that: Generating a load adjustment instruction according to the regional load balance degree and the node overload amount includes: Transfer the load of the overloaded node to a value not less than the overload amount, so that the node load value is reduced to within the node load upper limit; Nodes with smaller load values ​​are selected first, and load transfer is performed according to the adjustment rule that the overload of each node after load transfer is 0 and close.

Images