A template scaffold safety risk assessment method based on strain energy sensitive analysis and physical constraint prediction

Abstract

The application discloses a template scaffold safety risk assessment method based on strain energy sensitive analysis and physical constraint prediction, relates to the technical field of building construction safety monitoring, acquires structure parameters, material parameters, connection parameters and construction working condition parameters and establishes a finite element stress simulation model; determines high-risk sensitive nodes according to strain energy changes of each rod under different construction working conditions, effectively covers the constraint optimization monitoring points and returns a candidate layout scheme to the finite element stress simulation model for coverage checking; collects real-time monitoring data, filters abnormal data and reconstructs signals to obtain structure response data; determines physical constraint conditions based on the finite element stress simulation model, predicts future structure states of the structure response data, determines a safety risk grade according to a prediction result and a preset safety control condition and outputs a safety risk assessment result.

Description

Technical Field

[0001] This invention relates to the field of construction safety monitoring technology, and in particular to a method for assessing the safety risks of formwork and scaffolding based on strain energy sensitivity analysis and physical constraint prediction. Background Technology

[0002] Formwork and scaffolding are widely used in the construction of concrete structures such as buildings and municipal infrastructure. During the construction phase, they need to withstand various loads, including the weight of the scaffold itself, the weight of the formwork and newly poured concrete, the loads of construction personnel and equipment, vibration loads, and wind loads. Furthermore, the structural load state and stress path exhibit significant phased and dynamic characteristics as the construction progresses. Therefore, continuous monitoring of the stress state, deformation state, and stability of formwork and scaffolding, as well as safety risk assessments, are crucial aspects of construction safety management.

[0003] Current safety monitoring of formwork scaffolding typically employs methods such as manual inspection, on-site sensor monitoring, and finite element or BIM-assisted analysis. Manual inspection relies primarily on on-site observation and experience to determine key inspection areas. Sensor monitoring usually involves deploying sensors for axial force, displacement, and tilt angle at pre-selected nodes, comparing measured parameters with preset thresholds to determine structural anomalies. Some technologies also utilize structural models for stress analysis and determine monitoring points based on locations of significant stress. Existing technologies can achieve real-time acquisition of local stress, displacement, and tilt status of formwork scaffolding and can combine monitoring data to generate risk alarms.

[0004] However, formwork scaffolding has numerous components and complex spatial connections. The sensitivity of each component to load changes varies under different construction conditions. Relying solely on experience, local stress extremes, or fixed locations for monitoring point placement can easily lead to omissions of critical risk areas or inadequate allocation of monitoring resources. Furthermore, existing alarm methods based on fixed thresholds primarily reflect the current structural state, failing to adequately consider the evolution of the structural state over a future period. Factors such as wind loads, mechanical vibrations, and vehicle movement at the construction site can also interfere with monitoring signals, affecting subsequent state assessments. In addition, some data-driven prediction methods rely heavily on learning input-output relationships from historical samples, and the consistency between the prediction results and the actual mechanical relationships of the formwork scaffolding still has room for improvement.

[0005] How to solve the above-mentioned technical problems is the problem faced by this invention. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for assessing the safety risks of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction. This method is beneficial for improving the targeting of risk-sensitive location monitoring and the mechanical consistency and risk identification capabilities of structural state prediction.

[0007] The technical solution adopted by this invention to solve its technical problem is as follows: This invention provides a method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction, comprising the following steps: S1. Obtain the structural parameters, material parameters, connection parameters, and construction condition parameters of the formwork scaffold, and establish a finite element stress simulation model based on the parameters. S2. Based on the finite element stress simulation model, determine the strain energy of each member under the benchmark construction condition and the target construction condition, calculate the strain energy change rate, and determine the high-risk sensitive nodes according to the strain energy change rate and the preset sensitive judgment conditions. S3. Based on the spatial location of the high-risk sensitive nodes and the physical quantities to be monitored, determine the candidate monitoring points, optimize the points with the effective coverage of the high-risk sensitive nodes as the layout constraint, return the obtained candidate layout scheme to the finite element stress simulation model to verify the coverage of the high-risk sensitive nodes, and determine the target monitoring point layout scheme based on the verification results. S4. Collect real-time monitoring data according to the target monitoring point layout scheme, and obtain structural response data after abnormal data filtering and signal reconstruction. S5. Based on the finite element stress simulation model, determine the physical constraints including mechanical equilibrium relationship, pole stability relationship and structural deformation relationship. Under the physical constraints, predict the structural state of the structural response data within a future preset time range to obtain the structural state prediction result. Determine the safety risk level based on the comparison result of the structural state prediction result and the preset safety control conditions, and output the safety risk assessment result.

[0008] Further, step S1 includes: S11. Establish a geometric structure model of the template scaffolding based on the structural parameters, perform finite element discretization on the geometric structure model, and discretize the rods and template panels in the template scaffolding into corresponding finite element elements to obtain a finite element discretization model composed of multiple finite element elements. S12. Determine the material properties of each finite element unit according to the material parameters, and assign the material properties to the corresponding finite element unit; determine the connection relationship and corresponding connection stiffness between the corresponding finite element units according to the connection parameters, and set the connection relationship and connection stiffness in the finite element discrete model; S13. Determine the boundary conditions of the finite element discrete model based on the structural parameters and connection parameters, and determine the load conditions and load application sequence corresponding to different construction conditions based on the construction condition parameters. S14. Configure the boundary conditions in the finite element discrete model after completing the material properties, connection relationships and connection stiffness settings, and configure the load conditions corresponding to different construction conditions according to the load application sequence to form the finite element stress simulation model.

[0009] Further, in step S2, the calculation of the strain energy change rate of each member includes: S21. Determine the reference load conditions corresponding to the reference construction condition based on the construction condition parameters, and determine the target load conditions corresponding to the target construction condition based on the reference load conditions and the incremental construction load generated during construction. S22. Apply the reference load conditions and target load conditions to the finite element stress simulation model to solve the stress, and determine the reference strain energy of each member under the reference construction condition and the target strain energy of the same member under the target construction condition. S23. Using the reference strain energy of each member as a comparison benchmark, determine the strain energy change rate of the corresponding member based on the change of the target strain energy of the same member relative to the reference strain energy.

[0010] Further, in step S2, after calculating the strain energy change rate of each member, the step of determining high-risk sensitive nodes includes: S24. According to the structural function of each member in the formwork scaffold, the members are divided into multiple member types, and the corresponding strain energy change rate thresholds are determined for different member types. Each strain energy change rate threshold is used as the preset sensitive judgment condition for the corresponding member type. S25. For each member, determine the corresponding strain energy change rate threshold according to the member type, compare the strain energy change rate of the member with the corresponding strain energy change rate threshold, and obtain the sensitivity judgment result corresponding to the member. S26. Based on the sensitivity determination result, determine the sensitive members that meet the corresponding preset sensitivity determination conditions, and determine the high-risk sensitive nodes based on the spatial position of the sensitive members in the finite element stress simulation model.

[0011] Further, in step S3, the step of determining candidate monitoring points includes: S31. Determine the spatial location of each of the high-risk sensitive nodes from the finite element stress simulation model to obtain the node location information of each of the high-risk sensitive nodes. S32. Determine the sensor type for monitoring the corresponding physical quantity to be monitored based on the physical quantity to be monitored corresponding to each of the high-risk sensitive nodes; S33. For each of the high-risk sensitive nodes, determine at least one candidate monitoring point based on the node location information corresponding to the high-risk sensitive node; associate the corresponding sensor type with the candidate monitoring point based on the physical quantity to be monitored corresponding to the high-risk sensitive node.

[0012] Furthermore, in step S3, the step of determining the target monitoring point deployment scheme includes: S34. Based on the spatial location of each high-risk sensitive node and the corresponding physical quantity to be monitored, as well as the sensor type associated with each candidate monitoring point, determine the coverage correspondence between each high-risk sensitive node and the candidate monitoring point that can collect its corresponding physical quantity to be monitored; take the condition that each high-risk sensitive node corresponds to at least one candidate monitoring point with the coverage correspondence in the candidate deployment scheme as an effective coverage condition, and use the effective coverage condition as the deployment constraint for the point optimization; S35. Based on the candidate monitoring points, a genetic algorithm is used to generate multiple candidate deployment schemes; according to the candidate monitoring points and their associated sensor types included in each candidate deployment scheme, the number of sensors and sensor deployment cost corresponding to the candidate deployment scheme are determined, and the number of sensors and sensor deployment cost are used as optimization targets, combined with Pareto front analysis to iteratively optimize the candidate deployment schemes. S36. In each iteration, each candidate deployment scheme is returned to the finite element stress simulation model. Based on the spatial location of the candidate monitoring points included in each candidate deployment scheme and the coverage correspondence, the coverage of the corresponding candidate deployment scheme to all the high-risk sensitive nodes is verified, and the corresponding coverage verification result is obtained. S37. Eliminate candidate deployment schemes that do not meet the effective coverage conditions based on the coverage verification results, and continue iterative optimization based on candidate deployment schemes that meet the effective coverage conditions until the iteration converges. S38. Select candidate deployment schemes that meet the effective coverage conditions from the candidate deployment schemes after iterative convergence, and determine the target monitoring point deployment scheme from the candidate deployment schemes that are at the Pareto front.

[0013] Furthermore, in step S4, the structural response data obtained after abnormal data filtering and signal reconstruction includes: S41. Perform wavelet transform on the original monitoring signal in the real-time monitoring data to decompose the original monitoring signal into multiple sub-band signals of different frequencies; S42. Perform intrinsic mode decomposition on the sub-band signal to obtain multiple intrinsic components corresponding to the original monitoring signal; S43. Input the intrinsic components into a pre-trained classifier for identification, determine the intrinsic components that characterize the real structural response as structural response components, and determine the intrinsic components that characterize environmental noise as noise components. S44. Remove the noise component from the intrinsic components and reconstruct the signal from the remaining structural response components to obtain the structural response data.

[0014] Further, in step S5, the physical constraints are determined based on the finite element stress simulation model, and the structural state is predicted within a future preset time range using the physical constraints on the structural response data, including: S51. Based on the mechanical equilibrium relationship, pole stability relationship and structural deformation relationship represented by the finite element force simulation model, establish mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations respectively, and use the mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations as the physical constraint conditions. S52. Obtain historical monitoring data of existing formwork and scaffolding projects, and generate simulation sample data corresponding to different construction conditions based on the finite element stress simulation model. Combine the historical monitoring data and the simulation sample data to form a training sample set for training the physical information neural network. S53. Construct the physical information neural network and train it based on the training sample set to obtain the network output corresponding to the training sample set; substitute the output quantities corresponding to each physical constraint equation in the network output into the corresponding physical constraint equation to obtain the equation residuals corresponding to each physical constraint equation, and determine the corresponding physical residual loss based on the equation residuals; determine the data fitting loss based on the deviation between the network output and the corresponding response data in the training sample set. S54. The data fitting loss and each of the physical residual losses are weighted and combined according to their respective preset weights to obtain a composite loss function, wherein the sum of the weights corresponding to each of the physical residual losses is greater than the weight corresponding to the data fitting loss. S55. Adjust the network parameters of the physical information neural network with the composite loss function as the optimization objective to obtain the trained physical information neural network; input the structural response data into the trained physical information neural network to obtain the mechanical characteristics constrained by the physical constraints.

[0015] Further, in step S5, predicting the structural state within a future preset time range based on the mechanical characteristics and the structural response data includes: S56. Organize the structural response data according to the acquisition time sequence of the structural response data, and divide the time-series organized structural response data using a sliding window method to obtain structural response time sequence segments. S57. Input the structural response time series segment into a long short-term memory network, and extract the temporal correlation information between different time steps through the long short-term memory network to obtain the first temporal feature; S58. Input the first temporal feature into a temporal convolutional network, and perform multi-scale temporal convolution processing on the first temporal feature through the temporal convolutional network to obtain dynamic temporal features. S59. Configure learnable fusion weight parameters corresponding to the mechanical features and the dynamic temporal features respectively, and perform weighted fusion of the mechanical features and the dynamic temporal features based on the learnable fusion weight parameters to obtain fused prediction features; S60. Based on the fusion prediction features, determine the predicted values ​​of the structural state parameters corresponding to the structural response data within the future preset time range, and construct the structural state prediction result from the predicted values ​​of the structural state parameters.

[0016] Further, in step S5, the step of outputting the security risk assessment result includes: S61. Determine one or more structural state parameters for safety risk assessment from the structural state prediction results, and extract the prediction results corresponding to each structural state parameter to obtain structural state prediction information corresponding to each structural state parameter. S62. Based on the preset safety control conditions, determine the risk classification judgment conditions corresponding to each of the structural state parameters, wherein the risk classification judgment conditions include parameter state judgment conditions corresponding to different risk levels. S63. Match the structural state prediction information corresponding to each of the structural state parameters with the corresponding risk classification judgment conditions to obtain the parameter risk judgment result corresponding to each of the structural state parameters, and determine the safety risk level based on each of the parameter risk judgment results. S64. Associate the safety risk level, the structural state prediction information corresponding to each of the structural state parameters, and the parameter risk judgment results to form and output the safety risk assessment result.

[0017] This invention uses a finite element stress simulation model to determine the strain energy of each member under both the baseline and target construction conditions. It then uses the strain energy change rate of the same member combined with sensitivity criteria corresponding to different member types to identify high-risk sensitive nodes. Compared to methods that determine key monitoring locations solely based on the absolute values ​​of stress and displacement at a given moment or empirical locations, this approach reflects the sensitivity of members to changes in construction conditions. This facilitates the identification of members and nodes that significantly impact structural safety during load changes, providing risk location data with structural mechanics basis for subsequent monitoring point determination.

[0018] This invention generates candidate monitoring points based on the spatial location of high-risk sensitive nodes and the physical quantities to be monitored, and uses effective coverage of these nodes as a constraint for point optimization. During the optimization process, candidate deployment schemes are returned to the finite element stress simulation model to verify their coverage of high-risk sensitive nodes, and schemes that do not meet the effective coverage conditions are eliminated. Therefore, the finite element sensitivity analysis results can continuously participate in the monitoring point optimization, rather than serving as a one-time preliminary analysis result. This helps reduce the possibility of overlooking high-risk sensitive nodes during the optimization process due to simply pursuing the number of sensors or deployment costs, and reduces unnecessary monitoring resource allocation while ensuring effective coverage of risk-sensitive locations. Specific implementations further optimize candidate schemes based on the number of sensors, deployment costs, and the Pareto front.

[0019] This invention performs anomaly filtering and signal reconstruction on real-time monitoring data collected from target monitoring points. By decomposing, identifying, and eliminating environmental noise components from different signal components, it generates structural response data for subsequent prediction. This reduces the impact of environmental disturbances such as wind loads, mechanical vibrations, and vehicle operation on structural state data at the construction site, while retaining effective response information characterizing the actual stress and deformation state of formwork and scaffolding, thus improving the reliability of the input data used in subsequent structural state analysis.

[0020] This invention utilizes the mechanical equilibrium relationships, upright stability relationships, and structural deformation relationships characterized by the finite element stress simulation model to form physical constraints, and then predicts the future structural state based on the structural response data under these physical constraints. Furthermore, by substituting the network output into the corresponding physical constraint equations to form a physical residual loss, which, together with the data fitting loss, participates in network parameter optimization, the prediction process not only utilizes the data patterns of historical monitoring data and simulation samples but is also constrained by the actual mechanical relationships of the formwork and scaffolding. This helps reduce the possibility of purely data-driven prediction results deviating from the laws of structural mechanics, and enhances the stability and physical consistency of structural state predictions under different construction conditions. Attached Figure Description

[0021] Figure 1This is a schematic diagram of the overall process of the template scaffolding safety risk assessment method of the present invention; Figure 2 This is a schematic diagram of the process for establishing a finite element stress simulation model and determining high-risk sensitive nodes based on strain energy sensitivity analysis in this invention. Figure 3 This is a schematic diagram of the process of determining candidate monitoring points based on high-risk sensitive nodes and optimizing the layout of target monitoring points in this invention. Figure 4 This is a schematic diagram of the process for filtering abnormal data and reconstructing signals from real-time monitoring data in this invention. Figure 5 This is a schematic diagram of the process for predicting structural state based on the fusion of physical constraints and dynamic temporal features in this invention. Figure 6 This is a schematic diagram of the process in this invention for determining the safety risk level and outputting the safety risk assessment result based on the structural state prediction result. Detailed Implementation

[0022] To clearly illustrate the technical features of this solution, the following detailed implementation method will be used to explain the solution.

[0023] See Figures 1 to 6 As shown, this embodiment is a method for assessing the safety risks of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction, including the following steps: S1. Obtain the structural parameters, material parameters, connection parameters, and construction condition parameters of the formwork scaffold, and establish a finite element stress simulation model based on the parameters. S2. Based on the finite element stress simulation model, determine the strain energy of each member under the benchmark construction condition and the target construction condition, calculate the strain energy change rate, and determine the high-risk sensitive nodes according to the strain energy change rate and the preset sensitive judgment conditions. S3. Based on the spatial location of the high-risk sensitive nodes and the physical quantities to be monitored, determine the candidate monitoring points, optimize the points with the effective coverage of the high-risk sensitive nodes as the layout constraint, return the obtained candidate layout scheme to the finite element stress simulation model to verify the coverage of the high-risk sensitive nodes, and determine the target monitoring point layout scheme based on the verification results. S4. Collect real-time monitoring data according to the target monitoring point layout scheme, and obtain structural response data after abnormal data filtering and signal reconstruction. S5. Based on the finite element stress simulation model, determine the physical constraints including mechanical equilibrium relationship, pole stability relationship and structural deformation relationship. Under the physical constraints, predict the structural state of the structural response data within a future preset time range to obtain the structural state prediction result. Determine the safety risk level based on the comparison result of the structural state prediction result and the preset safety control conditions, and output the safety risk assessment result.

[0024] Further, step S1 includes: S11. Establish a geometric structure model of the template scaffolding based on the structural parameters, perform finite element discretization on the geometric structure model, and discretize the rods and template panels in the template scaffolding into corresponding finite element elements to obtain a finite element discretization model composed of multiple finite element elements. S12. Determine the material properties of each finite element unit according to the material parameters, and assign the material properties to the corresponding finite element unit; determine the connection relationship and corresponding connection stiffness between the corresponding finite element units according to the connection parameters, and set the connection relationship and connection stiffness in the finite element discrete model; S13. Determine the boundary conditions of the finite element discrete model based on the structural parameters and connection parameters, and determine the load conditions and load application sequence corresponding to different construction conditions based on the construction condition parameters. S14. Configure the boundary conditions in the finite element discrete model after completing the material properties, connection relationships and connection stiffness settings, and configure the load conditions corresponding to different construction conditions according to the load application sequence to form the finite element stress simulation model.

[0025] The structural parameters are used to determine the geometric dimensions, spatial positions, and arrangement relationships of each structural object in the formwork scaffold. After completing the finite element discretization, the members and formwork panels are converted into corresponding finite element elements, and material properties, connection relationships, boundary conditions, and load conditions are further configured in the same finite element discretization model.

[0026] The boundary conditions for the model must consider the connection method between the scaffolding and the main structure. For ground-supported scaffolding, the bottom uprights are considered fixed-end constraints; for cantilevered scaffolding, the base of the cantilever is considered a fixed end, while the cantilever end is free. The coupler connection between the horizontal bars and the uprights is simplified to a semi-rigid connection, and the connection stiffness is determined based on the results of the coupler anti-slip test.

[0027] The material parameters are used to determine the material properties of each finite element element, and the connection parameters are used to determine the connection method and connection stiffness between the corresponding finite element elements; the boundary conditions are determined according to the formwork scaffold structure and its actual support and connection status with the main structure.

[0028] Construction loads are applied using a sequential superposition method. Dead loads include the self-weight of the scaffolding, the self-weight of the formwork, and the weight of newly poured concrete; live loads include the weight of construction workers, the weight of construction equipment, the concrete vibration load, and the wind load. The load application sequence simulates the actual construction process, first applying the dead load, and then applying the live load layer by layer.

[0029] Step S14 unifies the material properties, connection relationships, connection stiffness, boundary conditions, and load conditions corresponding to different construction conditions into the finite element discrete model, forming a finite element stress simulation model capable of solving the stress under different construction conditions. This model also serves as the structural calculation basis for subsequent strain energy sensitivity analysis, candidate layout scheme coverage verification, and determination of physical constraint conditions.

[0030] The bending resistance of beams and slabs is checked using the limit state design method, and the check formula is as follows:

[0031] Where M is the design bending moment, f_y is the yield strength of the steel reinforcement, A_s is the area of ​​the tensile reinforcement, h_0 is the effective height, and a_s' is the distance from the resultant point of the compression reinforcement to the edge of the section. The maximum bending moment is calculated using finite element analysis results, extracting the bending moment envelope diagram of each beam / slab element to determine the location and value of the maximum bending moment. The verification results are output in the form of a safety factor; a safety factor greater than 1.0 indicates that the bearing capacity requirements are met.

[0032] The stability verification of uprights is a core component of the safety assessment of formwork scaffolding. This invention employs a stability theory for compression members that considers initial defects, and the verification formula is as follows:

[0033] Where N is the design value of axial pressure, φ is the stability coefficient, A is the cross-sectional area of ​​the pole, and f is the design strength of the steel.

[0034] The stability coefficient φ is determined by referring to a table based on the slenderness ratio λ of the upright. The slenderness ratio relationship is as follows:

[0035] Where μ is the calculated length coefficient, l is the column step distance, and i is the radius of gyration. The value of the calculated length coefficient μ needs to take into account the constraint effect of the horizontal bar. This invention uses a modified Euler formula for iterative calculation, taking into account the nonlinear behavior of the column and the influence of initial bending.

[0036] The above mechanical verification relationships can be used in conjunction with the finite element calculation results; among them, the stability relationship of the upright can be used as the specific implementation relationship for establishing the stability constraint equation of the upright in step S51, and the stress and deformation verification results of the beam and plate can be used to characterize the mechanical state of the corresponding structural object under load.

[0037] Further, in step S2, the calculation of the strain energy change rate of each member includes: S21. Determine the reference load conditions corresponding to the reference construction condition based on the construction condition parameters, and determine the target load conditions corresponding to the target construction condition based on the reference load conditions and the incremental construction load generated during construction. S22. Apply the reference load conditions and target load conditions to the finite element stress simulation model to solve the stress, and determine the reference strain energy of each member under the reference construction condition and the target strain energy of the same member under the target construction condition. S23. Using the reference strain energy of each member as a comparison benchmark, determine the strain energy change rate of the corresponding member based on the change of the target strain energy of the same member relative to the reference strain energy.

[0038] The baseline construction condition is used to establish a comparative benchmark for strain energy variation analysis. The target construction condition is the condition to be analyzed formed by superimposing incremental construction loads generated during subsequent construction on the baseline construction condition. For any member to be analyzed, its baseline strain energy and target strain energy are obtained in the same finite element stress simulation model, while maintaining the correspondence between member identification, member type, and model spatial position. The strain energy variation rate of the member is determined based on the change in target strain energy relative to baseline strain energy.

[0039] Specifically, for any member to be analyzed, the reference strain energy of the member under the benchmark construction condition is used as a reference quantity. The target strain energy of the member under the target construction condition is compared with the reference strain energy to determine the strain energy change rate corresponding to the member. In one specific embodiment, the strain energy change rate is determined according to the following formula:

[0040] In the formula, This represents the rate of change of strain energy corresponding to the member being analyzed. This indicates the reference strain energy of the member under the reference construction conditions; This represents the target strain energy of the member under the target construction condition after the application of the corresponding incremental construction load. By using the reference strain energy of the same member as the normalization benchmark, the degree of variation in strain energy of different members can be characterized by relative change.

[0041] The absolute value is used to characterize the magnitude of strain energy change caused by the target construction condition relative to the reference construction condition, and the strain energy change rate is used to evaluate the sensitivity of the corresponding member to changes in the construction condition. The above calculation is performed based on the reference strain energy and target strain energy corresponding to each member to obtain the strain energy change rate for each member, and the obtained strain energy change rate is used as the data input for subsequent sensitivity determination.

[0042] Further, in step S2, after calculating the strain energy change rate of each member, the step of determining high-risk sensitive nodes includes: S24. According to the structural function of each member in the formwork scaffold, the members are divided into multiple member types, and the corresponding strain energy change rate thresholds are determined for different member types. Each strain energy change rate threshold is used as the preset sensitive judgment condition for the corresponding member type. S25. For each member, determine the corresponding strain energy change rate threshold according to the member type, compare the strain energy change rate of the member with the corresponding strain energy change rate threshold, and obtain the sensitivity judgment result corresponding to the member. S26. Based on the sensitivity determination result, determine the sensitive members that meet the corresponding preset sensitivity determination conditions, and determine the high-risk sensitive nodes based on the spatial position of the sensitive members in the finite element stress simulation model.

[0043] Specifically, strain energy change rate thresholds can be set according to the structural functions undertaken by members such as uprights, horizontal bars, and scissor braces. For each member, the threshold corresponding to its member type is compared. Members that meet the corresponding preset sensitivity criteria are identified as sensitive members. Then, high-risk sensitive nodes are determined based on the spatial location of the sensitive members in the finite element stress simulation model. As an optional implementation, the strain energy change rate and sensitivity determination results of each member can be mapped to the spatial location of the finite element model to form a structural sensitivity heatmap, displaying the sensitivity of different structural locations.

[0044] Furthermore, in a preferred embodiment, a three-dimensional finite element stress simulation model can be used as a carrier to map the strain energy change rate corresponding to each member according to its spatial position in the finite element model, forming a structural sensitivity heat map of the formwork scaffold. Different display states are used to characterize members with different levels of sensitivity. Through the structural sensitivity heat map, the spatial distribution of high-risk sensitive nodes and sensitive members in the formwork scaffold can be displayed intuitively, and auxiliary information can be provided for subsequent monitoring of physical quantities and spatial matching of monitoring points.

[0045] Furthermore, when the finite element stress simulation model of the formwork scaffold is periodically updated and the strain energy change rate of the members is recalculated based on the updated construction condition parameters during construction, sensitive members can be further confirmed based on the strain energy change rate judgment results obtained from multiple consecutive calculation cycles, thereby reducing the impact of a single construction load disturbance on the identification results of sensitive nodes. In one specific embodiment, a member can be marked as a continuously sensitive member when its strain energy change rate meets the corresponding sensitivity judgment condition for three consecutive cycles.

[0046] Further, in step S3, the step of determining candidate monitoring points includes: S31. Determine the spatial location of each of the high-risk sensitive nodes from the finite element stress simulation model to obtain the node location information of each of the high-risk sensitive nodes. S32. Determine the sensor type for monitoring the corresponding physical quantity to be monitored based on the physical quantity to be monitored corresponding to each of the high-risk sensitive nodes; S33. For each of the high-risk sensitive nodes, determine at least one candidate monitoring point based on the node location information corresponding to the high-risk sensitive node; associate the corresponding sensor type with the candidate monitoring point based on the physical quantity to be monitored corresponding to the high-risk sensitive node.

[0047] Step S31 extracts node position information by utilizing the correspondence between high-risk sensitive nodes and the corresponding node or rod positions in the finite element stress simulation model. The physical quantity to be monitored can be determined based on the stress characteristics and safety monitoring requirements of the structural parts corresponding to the high-risk sensitive nodes, such as axial force, tilt angle, horizontal displacement, vertical displacement, or stress, and the type of sensor capable of collecting the corresponding physical quantity is determined accordingly.

[0048] For the same high-risk sensitive node, one or more candidate monitoring points can be determined based on the node's spatial location, on-site installation conditions, and the physical quantity to be monitored. Corresponding sensor types are then associated with these candidate monitoring points, forming a set of candidate monitoring points that includes their spatial location, the physical quantity that can be collected, and the sensor type. In one specific implementation, the axial force of the upright can be obtained using pressure or strain sensors, the tilt of the frame can be obtained using tilt sensors, and horizontal and vertical displacements can be obtained using displacement sensors.

[0049] Furthermore, in a preferred embodiment, when determining candidate monitoring points and their corresponding sensor types, a sensor measurement sensitivity matrix can be established to characterize the response characteristics of different types of sensors to changes in the corresponding structural state when set at different candidate monitoring points, and the matching relationship between sensor types and candidate monitoring points can be determined based on the measurement sensitivity matrix.

[0050] Furthermore, in step S3, the step of determining the target monitoring point deployment scheme includes: S34. Based on the spatial location of each high-risk sensitive node and the corresponding physical quantity to be monitored, as well as the sensor type associated with each candidate monitoring point, determine the coverage correspondence between each high-risk sensitive node and the candidate monitoring point that can collect its corresponding physical quantity to be monitored; take the condition that each high-risk sensitive node corresponds to at least one candidate monitoring point with the coverage correspondence in the candidate deployment scheme as an effective coverage condition, and use the effective coverage condition as the deployment constraint for the point optimization; S35. Based on the candidate monitoring points, a genetic algorithm is used to generate multiple candidate deployment schemes; according to the candidate monitoring points and their associated sensor types included in each candidate deployment scheme, the number of sensors and sensor deployment cost corresponding to the candidate deployment scheme are determined, and the number of sensors and sensor deployment cost are used as optimization targets, combined with Pareto front analysis to iteratively optimize the candidate deployment schemes. S36. In each iteration, each candidate deployment scheme is returned to the finite element stress simulation model. Based on the spatial location of the candidate monitoring points included in each candidate deployment scheme and the coverage correspondence, the coverage of the corresponding candidate deployment scheme to all the high-risk sensitive nodes is verified, and the corresponding coverage verification result is obtained. S37. Eliminate candidate deployment schemes that do not meet the effective coverage conditions based on the coverage verification results, and continue iterative optimization based on candidate deployment schemes that meet the effective coverage conditions until the iteration converges. S38. Select candidate deployment schemes that meet the effective coverage conditions from the candidate deployment schemes after iterative convergence, and determine the target monitoring point deployment scheme from the candidate deployment schemes that are at the Pareto front.

[0051] The coverage correspondence takes into account both the spatial location of the candidate monitoring points and whether the associated sensor types can collect the physical quantities to be monitored from the corresponding high-risk sensitive nodes. Effective coverage conditions do not require a unique one-to-one correspondence between high-risk sensitive nodes and monitoring points, but rather require that each high-risk sensitive node be covered by at least one candidate monitoring point with a coverage correspondence, allowing the coverage requirements for high-risk sensitive nodes to directly enter the point optimization process.

[0052] The genetic algorithm generates candidate deployment schemes based on a set of candidate monitoring points. Each candidate scheme includes selected candidate monitoring points and their associated sensor types, thereby determining the number of sensors and the sensor deployment cost. Pareto front analysis is used to screen candidate schemes among the above optimization objectives. In each iteration, the candidate deployment schemes are returned to the same finite element stress simulation model, which serves as a unified carrier for spatial location. The coverage correspondence between the candidate monitoring points and high-risk sensitive nodes is verified. Candidate schemes that do not meet the effective coverage conditions are eliminated, while schemes that meet the conditions continue to iterate until convergence, at which point the target monitoring point deployment scheme is determined.

[0053] When construction conditions change and high-risk sensitive nodes are updated, the process of determining and optimizing candidate monitoring points can be re-executed based on the updated high-risk sensitive nodes, and subsequent monitoring configurations can be adjusted accordingly.

[0054] Furthermore, in step S4, the structural response data obtained after abnormal data filtering and signal reconstruction includes: S41. Perform wavelet transform on the original monitoring signal in the real-time monitoring data to decompose the original monitoring signal into multiple sub-band signals of different frequencies; S42. Perform intrinsic mode decomposition on the sub-band signal to obtain multiple intrinsic components corresponding to the original monitoring signal; S43. Input the intrinsic components into a pre-trained classifier for identification, determine the intrinsic components that characterize the real structural response as structural response components, and determine the intrinsic components that characterize environmental noise as noise components. S44. Remove the noise component from the intrinsic components and reconstruct the signal from the remaining structural response components to obtain the structural response data.

[0055] Sensors deployed according to the target monitoring point layout plan continuously collect real-time monitoring data such as axial force, tilt angle, displacement, and stress. The raw monitoring signals may simultaneously contain the actual structural response of the formwork and scaffolding, as well as noise generated by environmental disturbances such as wind load, mechanical vibration, and vehicle passage. Therefore, data processing steps S41 to S44 are performed before proceeding to the subsequent prediction process.

[0056] In this process, wavelet transform decomposes the original monitoring signal into multiple sub-band signals according to frequency characteristics, and intrinsic mode decomposition further extracts multiple intrinsic components from the wavelet-transformed signal. A pre-trained classifier then identifies each intrinsic component, determining the structural response component and noise component respectively. After removing the noise component, the remaining structural response components are reconstructed to obtain structural response data that corresponds to the corresponding monitoring point, sensor type, and physical quantity to be monitored. This structural response data is then used as the prediction input for step S5.

[0057] It should be noted that the abnormal data filtering in this embodiment does not delete all monitoring signals that deviate from the normal state as abnormal data. Instead, it mainly identifies and removes environmental noise that does not characterize the actual structural state of the formwork and scaffolding. For the real response signals that can characterize the dangerous stress state, abnormal deformation, or structural state change trend of the formwork and scaffolding, they are still retained as structural response components and participate in subsequent signal reconstruction.

[0058] Further, in step S5, the physical constraints are determined based on the finite element stress simulation model, and the structural state is predicted within a future preset time range using the physical constraints on the structural response data, including: S51. Based on the mechanical equilibrium relationship, pole stability relationship and structural deformation relationship represented by the finite element force simulation model, establish mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations respectively, and use the mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations as the physical constraint conditions. S52. Obtain historical monitoring data of existing formwork and scaffolding projects, and generate simulation sample data corresponding to different construction conditions based on the finite element stress simulation model. Combine the historical monitoring data and the simulation sample data to form a training sample set for training the physical information neural network. S53. Construct the physical information neural network and train it based on the training sample set to obtain the network output corresponding to the training sample set; substitute the output quantities corresponding to each physical constraint equation in the network output into the corresponding physical constraint equation to obtain the equation residuals corresponding to each physical constraint equation, and determine the corresponding physical residual loss based on the equation residuals; determine the data fitting loss based on the deviation between the network output and the corresponding response data in the training sample set. S54. The data fitting loss and each of the physical residual losses are weighted and combined according to their respective preset weights to obtain a composite loss function, wherein the sum of the weights corresponding to each of the physical residual losses is greater than the weight corresponding to the data fitting loss. S55. Adjust the network parameters of the physical information neural network with the composite loss function as the optimization objective to obtain the trained physical information neural network; input the structural response data into the trained physical information neural network to obtain the mechanical characteristics constrained by the physical constraints.

[0059] Step S51 determines the physical constraints from the structural mechanical relationships represented by the finite element stress simulation model. The mechanical equilibrium relationship is used to constrain the stress state under the corresponding construction conditions and load conditions; the column stability relationship is used to characterize the stability state of the column under axial compression and corresponding constraint conditions; the structural deformation relationship is used to characterize the relationship between the deformation states such as displacement, deflection, and tilt of the members, beams, slabs, or frames and their stress states.

[0060] The training sample set consists of historical monitoring data from existing projects and simulation sample data generated by the finite element stress simulation model under different construction conditions. During the training calculation, the output quantities corresponding to each physical constraint equation in the network output are substituted into the corresponding equations to obtain the equation residuals, which form the physical residual loss. At the same time, the data fitting loss is formed based on the deviation between the network output and the corresponding response data in the training sample set. The physical residual losses and the data fitting losses are combined according to preset weights to form a composite loss function, and the network parameters are adjusted with this composite loss function as the optimization objective.

[0061] After training is complete, the structural response data obtained in step S44 is input into the trained physical information neural network, which outputs mechanical characteristics constrained by mechanical equilibrium, column stability, and structural deformation relationships. These mechanical characteristics serve as the input for fusion with dynamic temporal characteristics in step S59.

[0062] In one specific implementation, the training of the physical information neural network employs a combination of supervised learning and physical constraints. The network structure uses a deep fully connected network, containing 5 hidden layers, each with 256 neurons, and the ReLU activation function is used.

[0063] The training data comes from two sources: first, historical monitoring data, including sensor data collected from existing projects and accident case data; and second, synthetic data generated based on finite element analysis, which is used to expand the training sample size.

[0064] The construction of physical constraint terms encodes the mechanical equilibrium constraint equations, pole stability constraint equations, and structural deformation constraint equations into the loss calculation process of the physical information neural network, and forms corresponding physical residual losses based on the residuals of the corresponding equations. In one specific implementation, the material constitutive relation and boundary conditions can also be combined to form corresponding residual terms. During training, the physical loss and data fitting loss are weighted and summed, and the backpropagation of the total loss simultaneously optimizes the network parameters, so that the learned solution both fits the observed data and satisfies the physical laws.

[0065] The training hyperparameters were set as follows: learning rate of 0.001, Adam optimizer, batch size of 64, and 500 training epochs. The physics loss weight was set to 10 times the data loss to ensure the dominance of physical constraints.

[0066] Further, in step S5, predicting the structural state within a future preset time range based on the mechanical characteristics and the structural response data includes: S56. Organize the structural response data according to the acquisition time sequence of the structural response data, and divide the time-series organized structural response data using a sliding window method to obtain structural response time sequence segments. S57. Input the structural response time series segment into a long short-term memory network, and extract the temporal correlation information between different time steps through the long short-term memory network to obtain the first temporal feature; S58. Input the first temporal feature into a temporal convolutional network, and perform multi-scale temporal convolution processing on the first temporal feature through the temporal convolutional network to obtain dynamic temporal features. S59. Configure learnable fusion weight parameters corresponding to the mechanical features and the dynamic temporal features respectively, and perform weighted fusion of the mechanical features and the dynamic temporal features based on the learnable fusion weight parameters to obtain fused prediction features; S60. Based on the fusion prediction features, determine the predicted values ​​of the structural state parameters corresponding to the structural response data within the future preset time range, and construct the structural state prediction result from the predicted values ​​of the structural state parameters.

[0067] Steps S56 to S58 complete the sliding window partitioning, LSTM temporal correlation information extraction, and TCN multi-scale temporal convolution processing according to the acquisition sequence of the structural response data to obtain dynamic temporal features. Step S59 configures learnable fusion weight parameters for the mechanical features output in step S55 and the dynamic temporal features output in step S58, and adjusts the contribution of the two types of features to the fused prediction features according to the training results, so that the fused prediction features simultaneously contain physical and mechanical constraint information and temporal evolution information.

[0068] In one specific implementation, the changing trends of key structural parameters can be predicted over the next 10 to 15 minutes. Structural state parameters may include parameters related to the safety status of the formwork scaffolding, such as axial force, displacement, tilt angle, and stress. The predicted values ​​of each structural state parameter within a preset future time range constitute the structural state prediction result.

[0069] Specifically, the LSTM-TCN fusion model is used for time series prediction tasks. The LSTM layer has 128 hidden units and adopts a double-layer stacked structure to enhance feature extraction capabilities; the TCN layer has 4 convolutional blocks, each containing a dilated causal convolutional layer and residual connections, with the dilation factor growing exponentially at 1, 2, 4, and 8 to cover time dependencies at different scales.

[0070] In a specific model training and time series data preprocessing method, the input data preprocessing includes time series normalization, which maps physical parameters of different dimensions to the interval between 0 and 1; sliding window sampling, with the window size set to 180 time steps (corresponding to 3 minutes of monitoring data) and the step size set to 6 time steps; and outlier handling, which replaces data points that exceed 3 times the standard deviation with the window mean.

[0071] The model training employs a temporal cross-validation strategy, dividing the data into training, validation, and test sets in a 6:2:2 ratio. An early stopping mechanism is used during training: training stops when the validation set loss no longer decreases after 30 consecutive epochs, and the model with the smallest validation set loss is selected as the final model.

[0072] Evaluation metrics for the prediction results include: Root Mean Square Error (RMSE), which assesses the deviation between the predicted and actual values; Mean Absolute Percentage Error (MAPE), which assesses the relative prediction accuracy; and Directional Accuracy, which assesses the correctness of the predicted trend.

[0073] Further, in step S5, the step of outputting the security risk assessment result includes: S61. Determine one or more structural state parameters for safety risk assessment from the structural state prediction results, and extract the prediction results corresponding to each structural state parameter to obtain structural state prediction information corresponding to each structural state parameter. S62. Based on the preset safety control conditions, determine the risk classification judgment conditions corresponding to each of the structural state parameters, wherein the risk classification judgment conditions include parameter state judgment conditions corresponding to different risk levels. S63. Match the structural state prediction information corresponding to each of the structural state parameters with the corresponding risk classification judgment conditions to obtain the parameter risk judgment result corresponding to each of the structural state parameters, and determine the safety risk level based on each of the parameter risk judgment results. S64. Associate the safety risk level, the structural state prediction information corresponding to each of the structural state parameters, and the parameter risk judgment results to form and output the safety risk assessment result.

[0074] Step S61 selects one or more structural state parameters corresponding to the specific engineering safety control requirements from the structural state prediction results, and extracts the corresponding future predicted values ​​or predicted change trends; Step S62 determines the parameter state judgment conditions corresponding to different risk levels based on the design calculation values, standard control values, or other safety control conditions corresponding to each structural state parameter. For each structural state parameter participating in risk judgment, its structural state prediction information is matched with the corresponding risk classification judgment conditions to obtain the parameter risk judgment result; when multiple structural state parameters are used, the safety risk level is determined by comprehensively considering the risk judgment results of each parameter according to the risk judgment rules corresponding to the preset safety control conditions.

[0075] In one specific implementation, risk levels can be divided into blue, yellow, orange, and red alerts according to the degree of risk. A blue alert corresponds to a monitored parameter reaching 70% to 80% of the design calculation value or a change rate exceeding 5% compared to the previous period; a yellow alert corresponds to reaching 80% to 90% of the design calculation value or a change rate exceeding 10%; an orange alert corresponds to reaching 90% to 100% of the design calculation value or a change rate exceeding 20%; and a red alert corresponds to a monitored parameter exceeding the design calculation value, a change rate exceeding 30%, or a prediction indicating that the structure will become unstable within 10 minutes. The above classification conditions are used as a specific parameter configuration.

[0076] The safety risk assessment results correlate the safety risk level, the structural state prediction information corresponding to each structural state parameter, and the parameter risk judgment results, and can be output to the safety management platform. Based on the risk level, further application actions such as record archiving, information push notifications, SMS messages, or on-site audible and visual alarms can be executed.

[0077] The preset safety control conditions can be dynamically adjusted according to construction progress, load changes, environmental conditions, and structural evolution trends. For example, the axial force determination conditions can be adjusted based on the concrete pouring progress, and the relevant load effects can be corrected by combining real-time wind speed and temperature, or the risk determination threshold can be adjusted based on the continuous deterioration trend reflected by the structural state prediction results; the adjusted safety control conditions serve as the basis for risk classification determination in the current period of steps S62 to S63.

[0078] When the safety risk assessment results indicate that the structure is in a high-risk state or the mechanical calculation does not meet the control requirements, the Sobol index and Morris screening method can be used to analyze the influence of parameters such as pole spacing, step distance, scissor bracing arrangement, node connection stiffness, and foundation bearing capacity. Then, the sequential quadratic programming algorithm is used to determine the combination of adjustment parameters and form suggestions such as scissor bracing densification, pole step distance adjustment, or node connection strengthening. The adjusted scheme continues to undergo mechanical calculation.

[0079] The technical features of this invention not described can be implemented by or using existing technology, and will not be repeated here. Of course, the above description is not a limitation of this invention, and this invention is not limited to the examples above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of this invention should also be within the protection scope of this invention.

Claims

1. A method for assessing the safety risks of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction, characterized in that, Includes the following steps: S1. Obtain the structural parameters, material parameters, connection parameters, and construction condition parameters of the formwork scaffold, and establish a finite element stress simulation model based on the structural parameters, material parameters, connection parameters, and construction condition parameters. S2. Based on the finite element stress simulation model, determine the strain energy of each member under the benchmark construction condition and the target construction condition, calculate the strain energy change rate of each member, and determine the high-risk sensitive nodes according to the strain energy change rate and the preset sensitive judgment conditions. S3. Based on the spatial location of the high-risk sensitive nodes and the physical quantities to be monitored, determine the candidate monitoring points, optimize the points with the effective coverage of the high-risk sensitive nodes as the layout constraint, return the obtained candidate layout scheme to the finite element stress simulation model to verify the coverage of the high-risk sensitive nodes, and determine the target monitoring point layout scheme based on the verification results. S4. Collect real-time monitoring data according to the target monitoring point layout scheme, and obtain structural response data after abnormal data filtering and signal reconstruction. S5. Based on the finite element stress simulation model, determine the physical constraints including mechanical equilibrium relationship, pole stability relationship and structural deformation relationship. Under the physical constraints, predict the structural state of the structural response data within a future preset time range to obtain the structural state prediction result. Determine the safety risk level based on the comparison result of the structural state prediction result and the preset safety control conditions, and output the safety risk assessment result.

2. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, Step S1 includes: S11. Establish a geometric structure model of the template scaffolding based on the structural parameters, perform finite element discretization on the geometric structure model, and discretize the rods and template panels in the template scaffolding into corresponding finite element elements to obtain a finite element discretization model composed of multiple finite element elements. S12. Determine the material properties of each finite element unit according to the material parameters, and assign the material properties to the corresponding finite element unit; determine the connection relationship and corresponding connection stiffness between the corresponding finite element units according to the connection parameters, and set the connection relationship and connection stiffness in the finite element discrete model; S13. Determine the boundary conditions of the finite element discrete model based on the structural parameters and connection parameters, and determine the load conditions and load application sequence corresponding to different construction conditions based on the construction condition parameters. S14. Configure the boundary conditions in the finite element discrete model after completing the material properties, connection relationships and connection stiffness settings, and configure the load conditions corresponding to different construction conditions according to the load application sequence to form the finite element stress simulation model.

3. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S2, the calculation steps for the strain energy change rate of each member include: S21. Determine the reference load conditions corresponding to the reference construction condition based on the construction condition parameters, and determine the target load conditions corresponding to the target construction condition based on the reference load conditions and the incremental construction load generated during construction. S22. Apply the reference load conditions and target load conditions to the finite element stress simulation model to solve the stress, and determine the reference strain energy of each member under the reference construction condition and the target strain energy of the same member under the target construction condition. S23. Using the reference strain energy of each member as a comparison benchmark, determine the strain energy change rate of the corresponding member based on the change of the target strain energy of the same member relative to the reference strain energy.

4. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S2, after calculating the strain energy change rate of each member, the step of determining high-risk sensitive nodes includes: S24. According to the structural function of each member in the formwork scaffold, the members are divided into multiple member types, and the corresponding strain energy change rate thresholds are determined for different member types. Each strain energy change rate threshold is used as the preset sensitive judgment condition for the corresponding member type. S25. For each member, determine the corresponding strain energy change rate threshold according to the member type, compare the strain energy change rate of the member with the corresponding strain energy change rate threshold, and obtain the sensitivity judgment result corresponding to the member. S26. Based on the sensitivity determination result, determine the sensitive members that meet the corresponding preset sensitivity determination conditions, and determine the high-risk sensitive nodes based on the spatial position of the sensitive members in the finite element stress simulation model.

5. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S3, the step of determining candidate monitoring points includes: S31. Determine the spatial location of each of the high-risk sensitive nodes from the finite element stress simulation model to obtain the node location information of each of the high-risk sensitive nodes. S32. Determine the sensor type for monitoring the corresponding physical quantity to be monitored based on the physical quantity to be monitored corresponding to each of the high-risk sensitive nodes; S33. For each of the high-risk sensitive nodes, determine at least one candidate monitoring point based on the node location information corresponding to the high-risk sensitive node; associate the corresponding sensor type with the candidate monitoring point based on the physical quantity to be monitored corresponding to the high-risk sensitive node.

6. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 5, characterized in that, In step S3, the steps for determining the target monitoring point layout include: S34. Based on the spatial location of each high-risk sensitive node and the corresponding physical quantity to be monitored, as well as the sensor type associated with each candidate monitoring point, determine the coverage correspondence between each high-risk sensitive node and the candidate monitoring point that can collect its corresponding physical quantity to be monitored; take the condition that each high-risk sensitive node corresponds to at least one candidate monitoring point with the coverage correspondence in the candidate deployment scheme as an effective coverage condition, and use the effective coverage condition as a deployment constraint for point optimization. S35. Based on the candidate monitoring points, a genetic algorithm is used to generate multiple candidate deployment schemes; according to the candidate monitoring points and their associated sensor types included in each candidate deployment scheme, the number of sensors and sensor deployment cost corresponding to the candidate deployment scheme are determined, and the number of sensors and sensor deployment cost are used as optimization targets, combined with Pareto front analysis to iteratively optimize the candidate deployment schemes. S36. In each iteration, each candidate deployment scheme is returned to the finite element stress simulation model. Based on the spatial location of the candidate monitoring points included in each candidate deployment scheme and the coverage correspondence, the coverage of the corresponding candidate deployment scheme to all the high-risk sensitive nodes is verified, and the corresponding coverage verification result is obtained. S37. Eliminate candidate deployment schemes that do not meet the effective coverage conditions based on the coverage verification results, and continue iterative optimization based on candidate deployment schemes that meet the effective coverage conditions until the iteration converges. S38. Select candidate deployment schemes that meet the effective coverage conditions from the candidate deployment schemes after iterative convergence, and determine the target monitoring point deployment scheme from the candidate deployment schemes that are at the Pareto front.

7. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S4, the structural response data obtained after abnormal data filtering and signal reconstruction includes: S41. Perform wavelet transform on the original monitoring signal in the real-time monitoring data to decompose the original monitoring signal into multiple sub-band signals of different frequencies; S42. Perform intrinsic mode decomposition on the sub-band signal to obtain multiple intrinsic components corresponding to the original monitoring signal; S43. Input the intrinsic components into a pre-trained classifier for identification, determine the intrinsic components that characterize the real structural response as structural response components, and determine the intrinsic components that characterize environmental noise as noise components. S44. Remove the noise component from the intrinsic components and reconstruct the signal from the remaining structural response components to obtain the structural response data.

8. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S5, the physical constraints are determined based on the finite element stress simulation model, and the structural state is predicted within a preset time range using the physical constraints on the structural response data, including: S51. Based on the mechanical equilibrium relationship, pole stability relationship and structural deformation relationship represented by the finite element force simulation model, establish mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations respectively. The mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations are collectively referred to as physical constraint equations, and the mechanical equilibrium constraint equations, pole stability constraint equations and structural deformation constraint equations are used as the physical constraint conditions. S52. Obtain historical monitoring data of existing formwork and scaffolding projects, and generate simulation sample data corresponding to different construction conditions based on the finite element stress simulation model. Combine the historical monitoring data and the simulation sample data to form a training sample set for training the physical information neural network. S53. Construct the physical information neural network and train it based on the training sample set to obtain the network output corresponding to the training sample set; substitute the output quantities corresponding to each physical constraint equation in the network output into the corresponding physical constraint equation to obtain the equation residuals corresponding to each physical constraint equation, and determine the corresponding physical residual loss based on the equation residuals; determine the data fitting loss based on the deviation between the network output and the corresponding response data in the training sample set. S54. The data fitting loss and each of the physical residual losses are weighted and combined according to their respective preset weights to obtain a composite loss function, wherein the sum of the weights corresponding to each of the physical residual losses is greater than the weight corresponding to the data fitting loss. S55. Adjust the network parameters of the physical information neural network with the composite loss function as the optimization objective to obtain the trained physical information neural network; input the structural response data into the trained physical information neural network to obtain the mechanical characteristics constrained by the physical constraints.

9. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 8, characterized in that, In step S5, predicting the structural state within a preset time range based on the mechanical characteristics and the structural response data includes: S56. Organize the structural response data according to the acquisition time sequence of the structural response data, and divide the time-series organized structural response data using a sliding window method to obtain structural response time sequence segments. S57. Input the structural response time series segment into a long short-term memory network, and extract the temporal correlation information between different time steps through the long short-term memory network to obtain the first temporal feature; S58. Input the first temporal feature into a temporal convolutional network, and perform multi-scale temporal convolution processing on the first temporal feature through the temporal convolutional network to obtain dynamic temporal features. S59. Configure learnable fusion weight parameters corresponding to the mechanical features and the dynamic temporal features respectively, and perform weighted fusion of the mechanical features and the dynamic temporal features based on the learnable fusion weight parameters to obtain fused prediction features; S60. Based on the fusion prediction features, determine the predicted values ​​of the structural state parameters corresponding to the structural response data within the future preset time range, and construct the structural state prediction result from the predicted values ​​of the structural state parameters.

10. The method for assessing the safety risk of formwork scaffolding based on strain energy sensitivity analysis and physical constraint prediction as described in claim 1, characterized in that, In step S5, the step of outputting the security risk assessment result includes: S61. Determine one or more structural state parameters for safety risk assessment from the structural state prediction results, and extract the prediction results corresponding to each structural state parameter to obtain structural state prediction information corresponding to each structural state parameter. S62. Based on the preset safety control conditions, determine the risk classification judgment conditions corresponding to each of the structural state parameters, wherein the risk classification judgment conditions include parameter state judgment conditions corresponding to different risk levels. S63. Match the structural state prediction information corresponding to each of the structural state parameters with the corresponding risk classification judgment conditions to obtain the parameter risk judgment result corresponding to each of the structural state parameters, and determine the safety risk level based on each of the parameter risk judgment results. S64. Associate the safety risk level, the structural state prediction information corresponding to each of the structural state parameters, and the parameter risk judgment results to form and output the safety risk assessment result.

Images