Building construction site safety analysis system and method based on artificial intelligence

Abstract

The invention discloses a building construction site safety analysis system and method based on artificial intelligence. The method comprises the steps that scaffold images and scaffold basic data of a construction site are collected; the deviation angle of the scaffold material on the construction site is collected; inputting the scaffold image into a pre-trained recognition model, and outputting scaffold component labeling information; the scaffold component labeling information, the scaffold basic data and the deviation angle are input into a pre-trained stability decision model, and a stability result is output and comprises stability and instability; if the stability result is unstable, optimizing the scaffold structure through an optimization heuristic algorithm based on the set target condition and constraint condition, outputting the optimal scaffold structure layout, and transforming the existing scaffold according to the optimal scaffold structure layout; according to the whole system, through scientific modeling and optimization analysis, the collapse risk of the scaffold is greatly reduced, and the overall safety level of a construction site is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building safety analysis, and in particular to an artificial intelligence-based construction site safety analysis system and method. Background Art

[0002] At the construction site, scaffolding is the main construction auxiliary tool, and its safety directly affects the life safety of construction workers and the smooth progress of construction projects. However, the current scaffolding design, construction, management and safety analysis methods are generally insufficient and fail to fully meet the requirements of modern construction for efficiency and safety.

[0003] Traditional scaffolding design mostly relies on the experience of construction workers or simple static mechanical calculations, and it is difficult to fully consider the dynamic impact of various factors in complex construction environments (such as wind loads, construction loads, etc.) on the stability of the scaffolding. At the same time, the construction of scaffolding usually requires a high degree of manual operation, and errors are prone to occur during the construction process, resulting in structural instability. Especially in the construction of high-rise buildings and complex structures, these problems are particularly prominent and can easily lead to safety accidents.

[0004] In terms of scaffolding inspection and management, existing technologies mainly rely on manual visual inspection or simple inspection equipment. The inspection results are often incomplete and inefficient, making it difficult to detect potential problems in the scaffolding structure in a timely manner. In addition, scaffolding optimization design technology is relatively backward. Existing technologies lack scientific optimization algorithms and intelligent auxiliary tools, and cannot effectively reduce construction material costs or improve construction efficiency while ensuring safety.

[0005] Therefore, it is an urgent problem for technical personnel in this field to propose a construction site safety analysis system and method based on artificial intelligence to solve the difficulties existing in the prior art. Summary of the invention

[0006] The purpose of the present invention is to provide a construction site safety analysis system and method based on artificial intelligence to address the deficiencies in the background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a construction site safety analysis method based on artificial intelligence, comprising:

[0008] Collect scaffolding images and scaffolding basic data at the construction site;

[0009] Collect the offset angle of scaffolding materials at the construction site;

[0010] Input the scaffolding image into the pre-trained recognition model and output the scaffolding component annotation information;

[0011] Input the scaffolding component annotation information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable;

[0012] If the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, and the optimal scaffolding structure layout is output. The existing scaffolding is modified according to the optimal scaffolding structure layout.

[0013] Furthermore, the basic data of the scaffolding includes the material of each component, the cross-sectional area of ​​each component, the design strength of each component, the diameter of each component, the length of each component, the height of the scaffolding and the number of scaffolding layers; the components include cross bars, vertical bars, diagonal braces and fixed anchor rods; the scaffolding component annotation information includes the scaffolding component annotation drawing and the number of each component.

[0014] Furthermore, the objective conditions are to maximize the stability score and minimize the usage of components.

[0015] Furthermore, the constraints include component material selection restrictions and position range restrictions; component materials are restricted by existing scaffolding materials available at the construction site; position range restrictions include ground coordinate range restrictions, fixed anchor point coordinate range restrictions and space coordinate range restrictions.

[0016] Furthermore, the method for outputting the optimal scaffolding structure layout includes:

[0017] Step 101, using the existing number of components to randomly generate R groups of executable initial scaffolding structure layouts, which can be obtained by screening by staff, R is an integer greater than 1, and a scaffolding structure layout includes multiple particles, each particle represents a component and a corresponding component material;

[0018] Step 102: Calculate the stability score of each scaffolding structure layout according to the sum of the number of components exceeding the position range limit in each scaffolding structure layout, the total usage of each component in each scaffolding structure layout, the structural deformation of each scaffolding structure layout, and the sum of the number of selected materials of each component in each scaffolding structure layout exceeding the component material selection limit.

[0019] Step 103: update the speed and position of the particles, that is, update the coordinates of each component and the material of each component to generate a new scaffolding structure layout;

[0020] The speed update methods include:

[0021] V i (t+1)=σ×V i (t)+c1×a1×(P i -X i (t))+c2×a2×(GXi (t));

[0022] Where V i (t+1) is the velocity of the ith particle at the t+1th iteration; σ is the inertia weight, which controls the dependence of the velocity on the previous velocity, and its value is between 0 and 1; c1 and c2 are acceleration constants, and their values ​​are 2; P i is the historical best position of the ith particle in the scaffolding structure layout, that is, if the current stability score S i If it is higher than the particle's previous best score, update P i ;X i (t) is the position of the i-th particle at the t-th iteration; G is the global optimal position, that is, among all particles, the one with the highest stability score is the global optimal position; a1 and a2 are random numbers in the range [0,1];

[0023] The position update formula is as follows:

[0024] X i (t+1)=X i (t)+V i (t+1);

[0025] X i (t+1) is the new position of the i-th particle at the t+1th iteration.

[0026] Step 104, repeating steps 102-103 until a specified number of iterations is reached, and then selecting the scaffolding structure layout with the highest stability score. The scaffolding structure layout with the highest stability score is the optimal scaffolding structure layout, which meets the set target conditions.

[0027] Furthermore, the stability score of the scaffolding structure layout is:

[0028] S r =w1×z r -w2×B r -w3×D r -w4×C r ;

[0029] In the formula, S r is the stability score of the r-th scaffolding structure layout, r∈R; z r is the sum of the number of components that exceed the position range limit in the rth scaffolding structure layout; B r is the total usage of each component of the rth scaffolding structure layout; D r is the structural deformation of the rth scaffolding structure layout; C rThe sum of the number of materials selected for each component of r types of scaffolding structure layouts that exceeds the component material selection limit; w1, w2, w3, w4 are weight coefficients, which are values ​​greater than 0.

[0030] Furthermore, the method for obtaining the structural deformation includes:

[0031] Step 201, establishing a structural model of a scaffold according to the scaffold structural layout; defining the geometric structure and material properties of the scaffold; defining the geometric structure of the scaffold includes defining nodes, units and node degrees of freedom of each layer of the scaffold;

[0032] The nodes are the locations of the connection points of the scaffolding, and the node locations are expressed by coordinates; the units are the rods connecting the nodes, and the rods include crossbars, vertical rods, diagonal braces and fixed anchor rods; determine the nodes connecting the horizontal crossbars, vertical rods, diagonal braces, fixed anchor rods of each layer of the scaffolding and each unit;

[0033] Step 202, calculating a single unit stiffness matrix; assembling all unit stiffness matrices into a global stiffness matrix according to node connection relationships;

[0034] Step 203, establishing an equilibrium equation, wherein the equilibrium equation is that the product of the global stiffness matrix and the node displacement vector is equal to the external load vector;

[0035] Step 204, calculate each node displacement vector according to the equilibrium equation, calculate the displacement of each node according to the node displacement vector, and take the maximum displacement as the structural deformation, which represents the maximum displacement under the action of the load F of the corresponding scaffolding structure layout.

[0036] Furthermore, the element stiffness matrix is:

[0037]

[0038] K e is the stiffness matrix of the e-th unit; E e The elastic modulus of the e-th unit, in Pascals (Pa); A e is the cross-sectional area of ​​the e-th unit; L e is the length of the e-th unit; the numbers in the matrix are fixed values.

[0039] Furthermore, the method of assembling all unit stiffness matrices into a global stiffness matrix according to the node connection relationship includes:

[0040] Step 2021, taking the product of the number of nodes and the degree of freedom of the nodes as the size of the global stiffness matrix;

[0041] Step 2022, initializing the global stiffness matrix, converting the unit stiffness matrix into the global coordinate system, and obtaining the unit global stiffness matrix;

[0042] Step 2023, adding the unit global stiffness matrix to the global stiffness matrix;

[0043] Step 2024, repeat steps 2022-2023, add each unit stiffness matrix to the global stiffness matrix K, and obtain the global stiffness matrix K.

[0044] Furthermore, the size of the global stiffness matrix is:

[0045] K = (1 × N) × (1 × N);

[0046] Where K is the global stiffness matrix and N is the number of nodes; for example, if there are 4 nodes, the size of the global stiffness matrix is ​​a 4×4 matrix.

[0047] Furthermore, the element global stiffness matrix is ​​obtained.

[0048]

[0049] is the global stiffness matrix of the e-th unit, that is, the stiffness matrix of the e-th unit in the global coordinate system; T e is the coordinate transformation matrix; is the transposed matrix of the coordinate transformation matrix;

[0050] Conversion steps:

[0051] Step 1: Substitute the element stiffness matrix K e Left multiplication Transform it to the displacement of the global coordinate system;

[0052] Step 2: Multiply right by T e , mapping the transformed matrix to the force in the global coordinate system.

[0053] Furthermore, the training method of the recognition model includes:

[0054] Collecting a group A of identification data in advance, where A is an integer greater than 1, and the identification data includes a scaffolding image and scaffolding component annotation information, and the scaffolding component annotation information includes a scaffolding component annotation picture and the number of each component;

[0055] The recognition model uses the scaffolding component labeling information corresponding to each group of scaffolding images as output, and the actual scaffolding component labeling information corresponding to each group of scaffolding images as a prediction target; minimizing the sum of prediction errors of all scaffolding component labeling information is used as a training target; the recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped; the recognition model is a YOLO model or a Mask R-CNN model.

[0056] Furthermore, the training method of the stability decision model includes:

[0057] Collect q groups of decision data in advance, where q is an integer greater than 1, and the decision data includes decision feature data and stability results; the decision feature data includes scaffolding component marking information, scaffolding foundation data and offset angle;

[0058] Convert the decision data and the time required for the corresponding task into a corresponding set of feature vectors;

[0059] Each group of feature vectors is used as the input of a stability decision model, and the stability decision model uses the stability result corresponding to each group of decision feature data as the output, and the actual stability result corresponding to each group of decision feature data as the prediction target; minimizing the sum of prediction errors of all decision feature data is used as the training target; the stability decision model is trained until the sum of prediction errors reaches convergence and the training is stopped; the stability decision model is a multilayer perceptron or a long short-term memory network.

[0060] An artificial intelligence-based construction site safety analysis system is used to implement an artificial intelligence-based construction site safety analysis method, including:

[0061] The first acquisition module is used to collect scaffolding images and scaffolding basic data at the construction site;

[0062] The second acquisition module is used to collect the offset angle of the scaffolding materials at the construction site;

[0063] The first analysis module is used to input the scaffold image into the pre-trained recognition model and label the scaffold component information;

[0064] The second analysis module is used to input the scaffolding component annotation information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable;

[0065] In the third analysis module, if the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, the optimal scaffolding structure layout is output, and the existing scaffolding is modified according to the optimal scaffolding structure layout.

[0066] The technical effects of the artificial intelligence-based construction site safety analysis system and method provided by the present invention are as follows:

[0067] Through the first and second acquisition modules, the system can comprehensively collect scaffolding data (including material, size, angle, etc.) at the construction site, laying the foundation for subsequent analysis, thereby improving the accuracy of scaffolding safety assessment; with the help of pre-trained YOLO model or Mask R-CNN model, scaffolding components can be automatically identified and labeled, which can reduce manual identification errors and improve work efficiency; through the pre-trained stability decision model, combined with the scaffolding basic data and offset angle, the stability assessment results are output, which can timely discover potential safety hazards of the scaffolding; with the help of optimization algorithm, the system can generate the optimal scaffolding structure layout on the premise of meeting the target conditions and constraints, so as to achieve efficient utilization of the scaffolding and improve safety.

[0068] The system uses optimization algorithms to minimize component usage and maximize stability scores, avoiding waste of resources while ensuring construction safety. It generates transformation plans based on the optimized scaffolding layout, which can directly guide on-site operations and reduce the disconnect between design and implementation. The system can dynamically optimize the scaffolding layout by updating particle positions and velocities in real time, reduce reliance on manual labor, adapt to changes in the construction site, and enhance adaptability. The entire system uses scientific modeling and optimization analysis to significantly reduce the risk of scaffolding collapse and improve the overall safety level of the construction site. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] 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 will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0070] Figure 1 A schematic diagram of a construction site safety analysis system based on artificial intelligence according to the present invention;

[0071] Figure 2 The present invention is a flow chart of a construction site safety analysis method based on artificial intelligence. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.

[0073] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be an intermediate element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be an intermediate element at the same time.

[0074] Example 1

[0075] like Figure 1 As shown, this embodiment provides an artificial intelligence-based construction site safety analysis system and method, including, via wired and / or wireless connection.

[0076] The first acquisition module is used to collect scaffolding images and basic data of scaffolding at the construction site. The basic data of scaffolding includes the material of each component, the cross-sectional area of ​​each component, the design strength of each component, the diameter of each component, the length of each component, the height of scaffolding and the number of scaffolding layers; the components include cross bars, vertical bars, diagonal braces and fixed anchor rods; the material, cross-sectional area, design strength, diameter and length can reflect the stability coefficient and anti-overturning moment of each component. The larger the anti-overturning moment and the stability coefficient, the more stable the scaffolding.

[0077] The second acquisition module is used to acquire the offset angle of the scaffolding at the construction site. The offset angle can be directly obtained through a tilt sensor installed on the scaffolding.

[0078] The first analysis module is used to input the scaffolding image into the pre-trained recognition model and output the scaffolding component labeling information. The scaffolding structure labeling diagram is to label the scaffolding components with different colors.

[0079] The training methods for the recognition model include:

[0080] Collect A group of identification data in advance, where A is an integer greater than 1, and the identification data includes scaffolding images and scaffolding component annotation information, and the scaffolding component annotation information includes scaffolding component annotation pictures (the scaffolding structure annotation picture is different color annotations for different scaffolding components) and the number of each component; the scaffolding component annotation picture is an image of different angles and scenes;

[0081] The recognition model uses the scaffolding component labeling information corresponding to each group of scaffolding images as output, and the actual scaffolding component labeling information corresponding to each group of scaffolding images as a prediction target; minimizing the sum of prediction errors of all scaffolding component labeling information is used as a training target; the recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped; the recognition model is a YOLO model or a Mask R-CNN model.

[0082] The second analysis module is used to input the scaffolding component labeling information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable.

[0083] The training method of the stability decision model includes:

[0084] Collect q groups of decision data in advance, where q is an integer greater than 1, and the decision data includes decision feature data and stability results; the decision feature data includes scaffolding component marking information, scaffolding foundation data and offset angle.

[0085] Convert the decision data and the time required for the corresponding task into a corresponding set of feature vectors;

[0086] Each group of feature vectors is used as the input of a stability decision model, and the stability decision model uses the stability result corresponding to each group of decision feature data as the output, and the actual stability result corresponding to each group of decision feature data as the prediction target; minimizing the sum of prediction errors of all decision feature data is used as the training target; the stability decision model is trained until the sum of prediction errors reaches convergence and the training is stopped; the stability decision model is a multilayer perceptron or a long short-term memory network.

[0087] In the third analysis module, if the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, the optimal scaffolding structure layout is output, and the existing scaffolding is modified according to the optimal scaffolding structure layout.

[0088] The objective condition is to maximize the stability score and minimize the usage of components; the constraints include component material selection restrictions and location range restrictions; component materials are restricted by the existing material restrictions of scaffolding on the construction site, such as steel, composite materials, etc.; location range restrictions include ground coordinate range restrictions, fixed anchor point coordinate range restrictions and space coordinate range restrictions.

[0089] To further explain, the ground coordinate range limitation refers to the coordinate range limitation of the connection point between the component and the ground, the fixed anchor point coordinate range limitation refers to the coordinate range limitation of the building's load-bearing wall; the spatial position limitation means that the entire scaffolding must not exceed the spatial coordinate range allowed by the construction site.

[0090] The optimization heuristic algorithm of this embodiment uses a particle swarm algorithm to optimize the scaffolding structure.

[0091] Methods for outputting optimal scaffolding structure layout include:

[0092] Step 101, using the existing number of components to randomly generate R groups of executable initial scaffolding structure layouts, which can be obtained by screening by staff, R is an integer greater than 1, and a scaffolding structure layout includes multiple particles, each particle represents a component and the corresponding component material.

[0093] Step 102: Calculate the stability score of each scaffolding structure layout.

[0094] S r =w1×z r -w2×B r -w3×D r -w4×C r ;

[0095] In the formula, S r is the stability score of the r-th scaffolding structure layout, r∈R; z r is the sum of the number of components that exceed the position range limit in the rth scaffolding structure layout; B r is the total usage of each component of the rth scaffolding structure layout; D r is the structural deformation of the rth scaffolding structure layout; C r The sum of the number of materials selected for each component of r types of scaffolding structure layouts that exceeds the component material selection limit; w1, w2, w3, w4 are weight coefficients, which are values ​​greater than 0.

[0096] The methods for obtaining structural deformation include:

[0097] Step 201, establish a structural model of the scaffolding according to the scaffolding structure layout; define the geometric structure and material properties of the scaffolding; define the geometric structure of the scaffolding including defining the nodes, units and node degrees of freedom of each layer of the scaffolding; each node has 1 degree of freedom, which is vertical displacement.

[0098] The nodes are the locations of the connection points of the scaffolding, and the node locations are expressed by coordinates; the units are the rods connecting the nodes, including crossbars, vertical poles, diagonal braces and fixed anchor rods; determine the horizontal crossbars, vertical poles (connecting different layers), diagonal braces (providing additional stability for the scaffolding), fixed anchor rods and nodes connecting each unit of each layer of the scaffolding.

[0099] Step 202, calculating a single unit stiffness matrix; assembling all unit stiffness matrices into a global stiffness matrix according to node connection relationships;

[0100]

[0101] K e is the stiffness matrix of the e-th unit; E e The elastic modulus of the e-th unit, in Pascals (Pa); A eis the cross-sectional area of ​​the e-th unit; L e is the length of the e-th unit; the numbers in the matrix are fixed values.

[0102] Methods for assembling all element stiffness matrices into a global stiffness matrix according to the node connection relationship include:

[0103] Step 2021: The product of the number of nodes and the node degrees of freedom is taken as the size of the global stiffness matrix K.

[0104] K = (1 × N) × (1 × N);

[0105] Where N is the number of nodes; for example, if there are 4 nodes, the size of the global stiffness matrix is ​​a 4×4 matrix.

[0106] Step 2022: Initialize the global stiffness matrix, transform the unit stiffness matrix into the global coordinate system, and obtain the unit global stiffness matrix

[0107]

[0108] is the global stiffness matrix of the e-th unit, that is, the stiffness matrix of the e-th unit in the global coordinate system; T e is the coordinate transformation matrix; is the transposed matrix of the coordinate transformation matrix.

[0109] Conversion steps:

[0110] Step 1: Substitute the element stiffness matrix K e Left multiplication Convert it to displacement in the global coordinate system.

[0111] Step 2: Multiply right by T e , mapping the transformed matrix to the force in the global coordinate system.

[0112] Step 2023: transform the element stiffness matrix K e Added to the global stiffness matrix.

[0113] Step 2024, repeat steps 2022-2023, and Add to the global stiffness matrix K to obtain the global stiffness matrix K.

[0114] It should be noted that all units According to the degree of freedom index of the node, the global stiffness matrix is ​​assembled.

[0115]

[0116] m, n are the row and column indices of the global stiffness matrix; are the row and column indices in the element stiffness matrix.

[0117] Step 203, establish the equilibrium equation: Kδ=F; δ is the node displacement vector, including the displacement of all nodes in all directions, in meters; F is the external load vector, including the load of all nodes; through K, the displacement δ of the node under the external load F can be solved. ; The external load is the load applied by the weight of the workers on the scaffold, the load applied by the weight of the equipment, and the load applied by the crosswind. The weight of the workers and the weight of the equipment are determined according to the specific construction site, and the crosswind direction is the direction blowing to the side of the scaffold.

[0118] Step 204, calculate each node displacement vector according to the equilibrium equation, calculate the displacement of each node according to the node displacement vector, and take the maximum displacement as the structural deformation, which represents the maximum displacement under the action of the load F of the corresponding scaffolding structure layout.

[0119] An example of obtaining the displacement of each node is as follows:

[0120] For example, the scaffolding structure layout consists of 4 nodes and 3 units; unit 1 connects node 1 and node 2; unit 2 connects node 2 and node 3; unit 3 connects node 3 and node 4; each node has only one degree of freedom, namely vertical displacement, and the unit attributes are all steel, with a cross-sectional area A of 0.01 square meters, a length L of 1 meter, and an elastic modulus E = 200×10 9 Pascal. A downward force E = 1000 Newtons is applied to node 4.

[0121] The stiffness coefficient of each element is:

[0122] Unit 1:

[0123]

[0124] Unit 2:

[0125] k2=k1=×10 9 N / m;

[0126] Unit 3:

[0127] 3=k1=×10 9 N / m.

[0128] The global stiffness matrix is ​​a 4×4 matrix, node 1, degree of freedom 1 (δ1); node 2, degree of freedom 2 (δ2); node 3, degree of freedom 3 (δ3); node 4, degree of freedom 4 (δ4); for each element, update the global stiffness matrix as follows, for the element connecting node y and node g:

[0129] k yy =k yy+K e , k yg =k yg +K e , K gy =K gy +K e , k gg =k gg +K e .

[0130] Initial global stiffness matrix (all zeros):

[0131]

[0132] Add cell 1 (connecting nodes 1 and 2):

[0133]

[0134] Add cell 2 (connecting nodes 2 and 3):

[0135]

[0136] Add cell 3 (connecting nodes 3 and 4):

[0137]

[0138] The complete global stiffness matrix K:

[0139]

[0140] The complete known force F = -1000 Newton (the downward force is negative); there is no external force on other nodes, F1 = F2 = F3 = 0; force vector F:

[0141]

[0142] Node 1 and node 2 are fixed and do not move (i.e., δ1=0, δ2=0), the global stiffness matrix and force vector are modified, the rows and columns corresponding to the known displacements are removed, and the simplified global stiffness matrix K′ is obtained; the unknown node displacement vectors are δ3 and v4.

[0143]

[0144] Simplified force vector F′:

[0145]

[0146] The equilibrium equation is expressed in matrix form:

[0147]

[0148] Substituting the values:

[0149]

[0150] Solve for the node displacement vector:

[0151] (4×10 9 )×δ3-(2×10 9 )×δ4=0; Equation (1)

[0152] -(2×10 9 )×δ3+(2×10 9 )×δ4=-1000; Equation (2)

[0153] Divide both sides of the equation by 10 9 :

[0154] 4δ3-2δ4=0; Equation (1.1)

[0155]

[0156] Solve δ4 from equation (1.1) and substitute δ4 into equation (2.1) to obtain:

[0157] Node 3: δ3 = -5 × 10 -7 Meters (downward displacement).

[0158] Node 4: δ4 = -1 × 10 -6 Meters (downward displacement), the displacement of node 4 is greater than the displacement of node 3, because node 4 bears more load and therefore bears more deformation. The displacement of node 4 is taken as the structural deformation.

[0159] Step 103: update the speed and position of the particles, that is, update the coordinates of each component and the material of each component to generate a new scaffolding structure layout; the speed update method includes:

[0160] V i (t+1)=σ×V i (t)+c1×a1×(P i -X i (t))+c2×a2×(GX i (t));

[0161] Where V i (t+1) is the velocity of the ith particle at the t+1th iteration; σ is the inertia weight, which controls the dependence of the velocity on the previous velocity, and its value is between 0 and 1; c1 and c2 are acceleration constants, and their values ​​are 2; P i is the historical best position of the ith particle in the scaffolding structure layout, that is, if the current stability score S iIf it is higher than the particle's previous best score, update P i ;X i (t) is the position of the i-th particle at the t-th iteration; G is the global optimal position, that is, among all particles, the one with the highest stability score is the global optimal position; a1 and a2 are random numbers in the range [0,1].

[0162] The position update formula is as follows:

[0163] X i (t+1)=X i (t)+V i (t+1);

[0164] X i (t+1) is the new position of the i-th particle at the t+1th iteration.

[0165] Step 104, repeating steps 102-103 until the specified number of iterations is reached, selecting the scaffolding structure layout with the highest stability score (i.e., maximizing the stability score), the scaffolding structure layout with the highest stability score is the optimal scaffolding structure layout, and meets the set target conditions.

[0166] Example 2

[0167] See also Figure 2 As shown, this embodiment provides a construction site safety analysis method based on artificial intelligence, including:

[0168] Collect scaffolding images and scaffolding basic data at the construction site;

[0169] Collect the offset angle of scaffolding materials at the construction site;

[0170] Input the scaffolding image into the pre-trained recognition model and output the scaffolding component annotation information;

[0171] Input the scaffolding component annotation information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable;

[0172] If the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, and the optimal scaffolding structure layout is output. The existing scaffolding is modified according to the optimal scaffolding structure layout.

[0173] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A construction site safety analysis method based on artificial intelligence, characterized in that: include: Collect scaffolding images and scaffolding basic data at the construction site; Collect the offset angle of scaffolding materials at the construction site; Input the scaffolding image into the pre-trained recognition model and output the scaffolding component annotation information; Input the scaffolding component annotation information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable; If the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, and the optimal scaffolding structure layout is output. The existing scaffolding is modified according to the optimal scaffolding structure layout.

2. The method for safety analysis of construction sites based on artificial intelligence according to claim 1 is characterized in that: The basic data of scaffolding include the material of each component, the cross-sectional area of ​​each component, the design strength of each component, the diameter of each component, the length of each component, the height of scaffolding and the number of scaffolding layers; the components include cross bars, vertical bars, diagonal braces and fixed anchor rods; the scaffolding component annotation information includes the scaffolding component annotation drawing and the number of each component.

3. The method for safety analysis of construction sites based on artificial intelligence according to claim 2 is characterized in that: The objective condition is to maximize the stability score and minimize the usage of components; the constraints include component material selection restrictions and location range restrictions; component materials are restricted by the existing materials of scaffolding available at the construction site; location range restrictions include ground coordinate range restrictions, fixed anchor point coordinate range restrictions and space coordinate range restrictions.

4. The method for safety analysis of construction sites based on artificial intelligence according to claim 3 is characterized in that: Methods for outputting optimal scaffolding structure layout include: Step 101, using the existing number of components to randomly generate R groups of executable initial scaffolding structure layouts, R is an integer greater than 1, and a scaffolding structure layout includes multiple particles, each particle represents a component and a corresponding component material; Step 102, calculating the stability score of each scaffolding structure layout according to the sum of the number of components exceeding the position range limit in each scaffolding structure layout, the total usage of each component in each scaffolding structure layout, the structural deformation of each scaffolding structure layout, and the sum of the number of selected materials of each component in each scaffolding structure layout exceeding the component material selection limit; Step 103: update the speed and position of the particles, that is, update the coordinates of each component and the material of each component to generate a new scaffolding structure layout; Step 104, repeating steps 102-103 until a specified number of iterations is reached, and then selecting the scaffolding structure layout with the highest stability score. The scaffolding structure layout with the highest stability score is the optimal scaffolding structure layout, which meets the set target conditions.

5. The method for safety analysis of construction sites based on artificial intelligence according to claim 4 is characterized in that: The methods for obtaining structural deformation include: Step 201, establishing a structural model of a scaffold according to the scaffold structural layout; defining the geometric structure and material properties of the scaffold; defining the geometric structure of the scaffold includes defining nodes, units and node degrees of freedom of each layer of the scaffold; The nodes are the locations of the connection points of the scaffolding, and the node locations are expressed by coordinates; the units are the rods connecting the nodes, and the rods include crossbars, vertical rods, diagonal braces and fixed anchor rods; determine the nodes connecting the horizontal crossbars, vertical rods, diagonal braces, fixed anchor rods of each layer of the scaffolding and each unit; Step 202, calculating a single unit stiffness matrix; assembling all unit stiffness matrices into a global stiffness matrix according to node connection relationships; Step 203, establishing an equilibrium equation, wherein the equilibrium equation is that the product of the global stiffness matrix and the node displacement vector is equal to the external load vector; Step 204, calculate each node displacement vector according to the equilibrium equation, calculate the displacement of each node according to the node displacement vector, and take the maximum displacement as the structural deformation, which represents the maximum displacement under the load of the corresponding scaffolding structure layout.

6. The method for safety analysis of construction sites based on artificial intelligence according to claim 5 is characterized in that: Methods for assembling all element stiffness matrices into a global stiffness matrix according to the node connection relationship include: Step 2021, taking the product of the number of nodes and the degree of freedom of the nodes as the size of the global stiffness matrix; Step 2022, initializing the global stiffness matrix, converting the unit stiffness matrix into the global coordinate system, and obtaining the unit global stiffness matrix; Step 2023, adding the element stiffness matrix to the global stiffness matrix; Step 2024, repeat steps 2022-2023, add each unit stiffness matrix to the global stiffness matrix, and obtain the global stiffness matrix.

7. The method for safety analysis of construction sites based on artificial intelligence according to claim 6 is characterized in that: The conversion steps of the element stiffness matrix to the global coordinate system are: Step 1: Transpose the element stiffness matrix with the coordinate transformation matrix on the left to transform it into the displacement of the global coordinate system; Step 2: Right-click the coordinate transformation matrix and map the transformed matrix to the force of the global coordinate system.

8. The method for safety analysis of construction sites based on artificial intelligence according to claim 1 is characterized in that: The training methods for the recognition model include: Collecting a group A of identification data in advance, where A is an integer greater than 1, and the identification data includes a scaffolding image and scaffolding component annotation information, and the scaffolding component annotation information includes a scaffolding component annotation picture and the number of each component; The recognition model uses the scaffolding component labeling information corresponding to each group of scaffolding images as output, and the actual scaffolding component labeling information corresponding to each group of scaffolding images as a prediction target; minimizing the sum of prediction errors of all scaffolding component labeling information is used as a training target; the recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped; the recognition model is a YOLO model or a Mask R-CNN model.

9. The method for safety analysis of construction sites based on artificial intelligence according to claim 1, characterized in that: The training method of the stability decision model includes: Collect q groups of decision data in advance, where q is an integer greater than 1, and the decision data includes decision feature data and stability results; the decision feature data includes scaffolding component marking information, scaffolding foundation data and offset angle; Convert the decision data and the time required for the corresponding task into a corresponding set of feature vectors; Each group of feature vectors is used as the input of a stability decision model, and the stability decision model uses the stability result corresponding to each group of decision feature data as the output, and the actual stability result corresponding to each group of decision feature data as the prediction target; minimizing the sum of prediction errors of all decision feature data is used as the training target; the stability decision model is trained until the sum of prediction errors reaches convergence and the training is stopped; the stability decision model is a multilayer perceptron or a long short-term memory network.

10. A construction site safety analysis method based on artificial intelligence, used to implement the construction site safety analysis method based on artificial intelligence according to any one of claims 1 to 9, characterized in that: include: The first acquisition module is used to collect scaffolding images and scaffolding basic data at the construction site; The second acquisition module is used to collect the offset angle of the scaffolding materials at the construction site; The first analysis module is used to input the scaffold image into the pre-trained recognition model and label the scaffold component information; The second analysis module is used to input the scaffolding component annotation information, scaffolding basic data and offset angle into the pre-trained stability decision model, and output the stability result, which includes stable and unstable; In the third analysis module, if the stability result is unstable, the scaffolding structure is optimized through the optimization heuristic algorithm based on the set target conditions and constraints, the optimal scaffolding structure layout is output, and the existing scaffolding is modified according to the optimal scaffolding structure layout.

Images