A method and system for safety early warning in steel structure construction based on the Internet of Things

Abstract

This invention discloses a method and system for early warning of steel structure construction safety based on the Internet of Things (IoT), belonging to the field of structural health monitoring. The method includes: acquiring first-state data and second-state data of the monitored structure; injecting topological incremental features corresponding to the placement of new components into the foundation structure model based on the second-state data to generate a dynamic structural model that evolves in real time with the construction process; determining a dynamic safety boundary threshold corresponding to the current construction process state based on the dynamic structural model; comparing the first-state data with the dynamic safety boundary threshold, and outputting an early warning command when the first-state data deviates from the dynamic safety boundary threshold. This invention eliminates the systematic mismatch between static thresholds and time-varying structures, reducing false alarms and missed detections.

Description

Technical Field

[0001] This invention relates to the field of structural health monitoring, and in particular to a method and system for early warning of safety during steel structure construction based on the Internet of Things. Background Technology

[0002] Before the steel structure is closed into a ring, it is in an unclosed construction phase. Its stress system continuously changes with the hoisting and placement of each new component and the unloading of each temporary support: the exposed windward area, structural eccentricity, and overall stiffness all evolve in real time with each process, making it a typical highly nonlinear time-varying dynamic system. However, existing construction safety monitoring technologies generally use a static, fixed structural model corresponding to the final completed state and fixed safety thresholds to fit the aforementioned time-varying structure. This approach easily leads to a systematic mismatch between the structural model and the actual structural state, resulting in frequent false alarms and missed judgments. Summary of the Invention

[0003] One of the objectives of this invention is to provide an IoT-based method for early warning of safety during steel structure construction, in order to solve the problems of systematic mismatch and false alarms caused by fitting time-varying unclosed structures with static fixed models and thresholds in the prior art, as well as the monitoring blind spots caused by sensor disconnection.

[0004] This invention is achieved through the following technical solution: a method for early warning of steel structure construction safety based on the Internet of Things, comprising the following steps: acquiring first state data and second state data of the monitored structure, wherein the first state data indicates the real-time mechanical response of each monitored node in the monitored structure, and the second state data indicates the placement status of new components in the monitored structure; based on the second state data, injecting the topological incremental features corresponding to the placement of the new components into the foundation structure model to generate a dynamic structure model that evolves in real time with the construction process, wherein the foundation structure model represents the overall mechanical state of the monitored structure at the end of the previous construction process; based on the dynamic structure model, determining a dynamic safety boundary threshold corresponding to the current construction process status; comparing the first state data with the dynamic safety boundary threshold, and outputting an early warning command when it is determined that the first state data deviates from the dynamic safety boundary threshold.

[0005] Furthermore, the first state data includes at least one of the strain, displacement, tilt angle, axial force, or acceleration of the monitored node.

[0006] Furthermore, the first state data is collected by a set of sensing devices deployed at key nodes and temporary support systems of the monitored structure, the set of sensing devices including at least one of strain gauges, vibrating wire strain sensors, inclinometers, accelerometers or axial force gauges; the processing device includes an edge computing gateway.

[0007] Further, acquiring the second state data of the monitored structure includes: generating valid second state data only when the new component simultaneously meets the spatial convergence condition, the motion convergence condition, and the identity verification condition; wherein, the spatial convergence condition is that the deviation between the current spatial position and the target installation position of the new component enters a preset tolerance range; the motion convergence condition is that the amount of motion of the new component within a set time window is lower than the static determination threshold; the identity verification condition is that the identification code of the new component matches the preset process list, and the contact trigger signal corresponding to the new component is triggered.

[0008] Furthermore, the amount of motion of the new component within a set time window in the motion convergence condition is the time average of the motion rate of the new component within the set time window; when the time average is lower than the stationary determination threshold, the motion convergence condition is determined to be met; the current spatial position of the new component is continuously acquired by the first positioning device, the identification code is read by the second identification device, and the second identification device is triggered to read when both the spatial convergence condition and the motion convergence condition are met; the first positioning device includes an ultra-wideband positioning device, the second identification device includes a radio frequency identification device, and the preset process list includes a building information model process list.

[0009] Further, the generation of the dynamic structural model includes: extracting the physical dimensions and material properties of the new component based on the second state data, generating a first sub-matrix characterizing the mechanical properties of the new component, and extracting the mapping vector of the new component in the global coordinate system, wherein the mapping vector is used to establish the correspondence between the local degrees of freedom of the new component and the global nodes; extracting the displacement constraints of the currently retained temporary support nodes in the monitored structure, and generating a constraint vector set; aligning the first sub-matrix to the corresponding dimension of the basic structural model through the mapping vector, as the topological incremental feature; and superimposing the aligned first sub-matrix and the constraint vector set onto the basic structural model without performing a global inversion on the basic structural model to obtain the dynamic structural model.

[0010] Further, without performing a global inversion on the basic structure model, the aligned first submatrix and the constraint vector set are superimposed onto the basic structure model, including: dividing the nodes in the basic structure model into an active node set affected by the placement of the new component and a dormant node set not affected by the placement of the new component; reusing the factor cache obtained from the dormant node set in the previous construction process to determine the cohesive equivalent stiffness of the active node set after aggregating the mechanical contribution of the dormant node set; solving the state of the active node set based on the cohesive equivalent stiffness to obtain the dynamic structure model; wherein, the factor cache is the result of decomposing the corresponding part of the dormant node set and is in a read-only reuse state in the current construction process.

[0011] Furthermore, the processing device includes an edge computing gateway, and the steel structure construction safety early warning method further includes: dividing the basic structure model into memory sub-blocks of fixed size according to the physical construction area; locating the affected target memory sub-block according to the spatial location of the new component indicated by the second state data; when determining the cohesive equivalent stiffness, only the target memory sub-block and the adjacent memory sub-blocks sharing the boundary node with the target memory sub-block are loaded into the fast cache of the edge computing gateway for incremental calculation, while the remaining memory sub-blocks are in a read-only locked state; wherein, the target memory sub-block and the adjacent memory sub-block loaded into the fast cache correspond to the active node set, and the remaining memory sub-blocks in the read-only locked state correspond to the dormant node set.

[0012] Furthermore, the topological incremental feature also includes the local change corresponding to the unloading of the temporary support node. When the dismantling confirmation signal of the target temporary support node is obtained, the constraint reaction force corresponding to the target temporary support node is extracted from the constraint vector set; the constraint reaction force is applied to the dynamic structure model as a reverse equivalent node load; the row and column corresponding to the target temporary support node are stripped from the dynamic structure model to obtain the updated dynamic structure model.

[0013] Furthermore, the dynamic safety boundary threshold includes: collecting environmental load parameters of the current construction site, converting the environmental load parameters into dynamic load vectors applied to each exposed node of the dynamic structural model; extracting the eccentric load values ​​of the semi-rigid connection nodes in the monitored structure; applying the dynamic load vectors and the eccentric load values ​​as inputs to the dynamic structural model, determining the theoretical stress extreme values ​​and theoretical deformation extreme values ​​of each monitored node under the current construction process state, and setting the theoretical stress extreme values ​​and the theoretical deformation extreme values ​​as the dynamic safety boundary threshold.

[0014] Furthermore, the environmental load parameters include the real-time wind speed and real-time wind direction at the current construction site; the dynamic load vector is determined based on the real-time wind speed, the real-time wind direction, and the exposed windward surface area of ​​the monitored structure under the current construction procedure state, and the exposed windward surface area is updated as the new component is placed.

[0015] Furthermore, the semi-rigid connection node is a connection node in the monitored structure that has not yet been permanently fixed, and the permanent fixation includes full-section welding or final tightening of high-strength bolts; the eccentric load value is determined based on the eccentric moment of the semi-rigid connection node and the spatial position of the node.

[0016] Further, when it is determined that the first state data deviates from the dynamic safety boundary threshold, an early warning command is output, including: when it is determined that the first state data deviates from the dynamic safety boundary threshold and satisfies the cascade instability judgment condition, the early warning command is output; wherein, determining whether the cascade instability judgment condition is satisfied includes: identifying the node whose first state data deviates from the dynamic safety boundary threshold among the monitored nodes as the target node; setting the stiffness parameter of the target node to zero in the dynamic structural model to generate a damaged topology model; determining the load redistribution coefficient for transferring the load originally borne by the target node to the adjacent nodes of the target node according to the damaged topology model; and determining that the cascade instability judgment condition is satisfied when the expected stress value of any of the adjacent nodes determined according to the load redistribution coefficient exceeds the corresponding yield extreme value.

[0017] Furthermore, in the dynamic structural model, the stiffness parameter of the target node is set to zero to generate a damaged topology model, including: subtracting the stiffness contribution of the target node in the row and column directions from the dynamic structural model to obtain the damaged topology model characterizing the residual structural state of the target node after it leaves the load-bearing system.

[0018] Furthermore, the adjacent nodes include first-order adjacent nodes directly connected to the target node, and second-order adjacent nodes indirectly connected to the target node via the first-order adjacent nodes; the load redistribution coefficient is determined based on the residual connection stiffness between each of the adjacent nodes and the target node, and the spatial distance between each of the adjacent nodes and the target node, wherein the residual connection stiffness is taken from the damaged topology model, the spatial distance is determined in the form of inverse square ratio, and the load redistribution coefficient corresponding to each of the adjacent nodes is normalized so that the sum of the load increments obtained by each of the adjacent nodes is equal to the load released by the target node.

[0019] Furthermore, the expected stress value of any of the adjacent nodes is the sum of the first state data of the adjacent node at the current moment and the strain increment of the adjacent node after constitutive transformation based on the load redistribution coefficient; when the number of adjacent nodes whose expected stress value exceeds the corresponding yield extreme value reaches at least one, it is determined that the cascade instability judgment condition is met and the warning command is output.

[0020] Furthermore, the steel structure construction safety early warning method further includes: when the data stream heartbeat of the target sensing device is detected to be lost, the monitored node corresponding to the target sensing device is identified as a failed node, and the spatial adjacency node matrix of the failed node is extracted in the dynamic structural model, wherein the spatial adjacency node matrix represents the mechanical transmission relationship between the failed node and the healthy node with normal data; the first state data of the healthy node is obtained, and a reverse node force is applied to the failed node, and the virtual state data of the failed node is determined by reverse calculation based on the balance relationship of the dynamic structural model; the virtual state data replaces the missing first state data of the failed node and participates in the comparison between the first state data and the dynamic safety boundary threshold.

[0021] Furthermore, when no direct external load is applied to the failed node, the reverse node force is zero, and the virtual state data of the failed node is uniquely determined by the first state data of the healthy node through the mechanical transmission relationship represented by the spatial adjacent node matrix.

[0022] Another aspect of the present invention provides an Internet of Things (IoT)-based steel structure construction safety early warning system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements any of the IoT-based steel structure construction safety early warning methods described above.

[0023] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0024] This invention establishes a dynamic structural model and dynamic safety boundary threshold by performing deterministic algebraic reorganization on the foundation structure model based on hoisting and positioning status data. This model evolves in real time with construction procedures and strictly corresponds to the current procedure status, eliminating the systematic mismatch between static thresholds and time-varying structures, and reducing false alarms and missed detections. Furthermore, by performing dimensionality reduction algebraic solutions only on the active node set affected by the procedure and reusing dormant zone factor caching, the solution scale is decoupled from the global structural scale, making real-time dynamic solutions feasible even on computationally limited edge devices. True positioning is determined by a combination of spatial convergence, motion convergence, and identity confirmation, fundamentally filtering out invalid high-load calculations caused by false positioning signals. Through pure algebraic deduction of damaged topology degradation and load redistribution, cascade instability is deterministically predicted within milliseconds, eliminating safety blind spots in cascade collapse. Virtual strain of missing nodes is inferred from the equilibrium relationship based on the dynamic structural model, ensuring that interpolation results meet mechanical equilibrium constraints even in scenarios where any sensor fails, maintaining the blind-spot-free operation of the safety monitoring system. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0026] Figure 1 This is a flowchart of the method provided in Embodiment 1 of the present invention.

[0027] Figure 2 This is a timing comparison diagram of the three-channel signals provided in Embodiment 1 of the present invention.

[0028] Figure 3 This is a comparison chart of cache hit rate and update time for the memory sub-block strategy provided in Embodiment 1 of the present invention.

[0029] Figure 4 This is a comparison chart of computational scales provided in Embodiment 1 of the present invention.

[0030] Figure 5 This is a comparison curve of edge gateway memory usage as the process progresses, provided in Embodiment 1 of the present invention.

[0031] Figure 6 This is a graph showing the percentage of active nodes provided in Embodiment 1 of the present invention.

[0032] Figure 7 The evolution curve provided in Embodiment 1 of the present invention.

[0033] Figure 8 This is a polar coordinate comparison diagram provided in Embodiment 1 of the present invention.

[0034] Figure 9 This is a comparison diagram of the entire process provided in Embodiment 1 of the present invention.

[0035] Figure 10 The thermal diagram of stress redistribution at key structural nodes provided in Embodiment 1 of the present invention.

[0036] Figure 11 This is a schematic diagram of the triggering time of the shutdown warning command provided in Embodiment 1 of the present invention.

[0037] Figure 12 This is a comparison chart of the early warning timeline provided in Embodiment 1 of the present invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0039] Example 1

[0040] This embodiment discloses a method for early warning of steel structure construction safety based on the Internet of Things. Figure 1 The flowchart of the method in this embodiment is shown. As can be seen from the flowchart, this embodiment includes the following steps:

[0041] Step 1: Collect real-time physical status data by deploying IoT sensing devices at key nodes of the steel structure and temporary support system, and obtain the hoisting and positioning status data of new components through RFID and ultra-wideband check-in devices.

[0042] In this context, the Internet of Things (IoT) sensing device set refers to a group of heterogeneous sensing devices deployed to comprehensively characterize the mechanical state of a steel structure during its unclosed phase. For example, the IoT sensing device set may include strain gauges, vibrating wire strain sensors, inclinometers, accelerometers, axial force gauges, etc., deployed at beam-column joints, truss chords, and temporary support columns.

[0043] Critical nodes are connection points that bear the main loads on the structural force transmission path, and whose failure will significantly affect the overall stability. For example, critical nodes can be beam-column rigid joints in frame structures, bolt-ball joints in space frame structures, and the connection between the core tube and the cantilever in tower structures.

[0044] A temporary support system refers to an auxiliary support structure temporarily installed before the steel structure is closed into a loop to bear construction loads and maintain the geometric stability of the structure. For example, a temporary support system can be a lattice-type temporary support column, guy rope, temporary tie rod, formwork, etc., which will be gradually removed or unloaded as the construction progresses.

[0045] Real-time physical state data refers to a set of physical quantities used to fully describe the current mechanical response of each monitored node in a steel structure. This real-time physical state data includes, but is not limited to, node strain. These include nodal displacements, nodal tilt angles, nodal axial forces, and nodal accelerations. It is understood that all of these physical quantities are expressed in continuous time... The index is reported in real time by the set of sensing devices, forming the input for subsequent security comparison logic.

[0046] Lifting and positioning status data refers to a set of composite information used to deterministically determine whether a new component has been absolutely and stably installed and positioned. It includes at least the component's real-time three-dimensional spatial coordinates, radio frequency identification (RFID) encoding, and contact-type limit switch trigger signals at the structural connection interfaces. It is understandable that the core value of lifting and positioning status data lies in providing an absolutely certain triggering basis for subsequent computationally intensive topology matrix reorganization operations, rather than simply a position arrival signal.

[0047] In this embodiment, in order to completely eliminate signal jitter during the hoisting operation at the mathematical level and achieve deterministic true positioning determination, thereby fundamentally eliminating invalid high-load calculations caused by false triggering of false positioning signals such as wind load amplitude and rope elastic oscillation in the edge gateway, the following physical timing verification sub-steps can be included to obtain the hoisting positioning status data of the new component:

[0048] Sub-step 1.1: Continuously acquire the real-time three-dimensional spatial coordinates of the hoisting components using ultra-wideband (UWB) tags;

[0049] Sub-step 1.2: When the real-time three-dimensional spatial coordinates enter the preset tolerance envelope of the target installation position, and the rate of change of coordinate fluctuation is lower than the static threshold within the set time window, trigger the radio frequency identification (RFID) reading command.

[0050] Sub-step 1.3: Compare the read RFID code with the pre-loaded BIM process list. If the codes match and the contact limit switch at the structural connection interface is triggered, then generate valid hoisting and positioning status data.

[0051] It is important to note that, in this embodiment, determining whether a component is truly and stably in place requires the simultaneous fulfillment of three independent and indispensable conditions, rather than triggering the process solely based on meeting the required precision at a single instant:

[0052] Firstly, there is the convergence determination of spatial position, meaning that the current three-dimensional spatial coordinates of the component have entered the tolerance envelope of the target installation position. The physical quantity corresponding to this determination is the current coordinates output in real time by the UWB positioning system. With the target theoretical coordinates Euclidean distance between If and only if the distance is less than the radius of the installation tolerance envelope. When (as specified in the engineering specifications), the spatial convergence condition is met.

[0053] Secondly, the time integral convergence determination of the motion state is based on the condition that the component's velocity has decreased to near-stationary levels within a sustained time window. It is important to emphasize that this embodiment deliberately uses the time average of the velocity within the time window, rather than a single instantaneous velocity, as the criterion. This is because, during wind-driven oscillation, the component's velocity may be zero even at its instantaneous value near the zero-crossing point. If only instantaneous velocity is used for determination, it is highly likely that a oscillating component that happens to pass the target position will be mistakenly identified as already in place. Therefore, a sliding time integration window is introduced. Calculate the time average of the component's motion rate within the window. Only when the average rate is below the static determination threshold The motion convergence condition is only met when the system's allowable small residual sway rate is within a certain limit. This design mathematically forces the component to remain stationary for a certain period of time, thereby fundamentally filtering out spurious positioning signals caused by instantaneous sway.

[0054] Thirdly, there is hardware verification of physical identity, meaning that the contact-type limit switch at the installation site has been physically triggered, and the RFID electronic tag code completely matches the preset component identity in the digital twin model (BIM process list). This determination is made through Boolean indicator variables. The expression is 1 if and only if the code matches and the limit switch is physically triggered, otherwise it is 0.

[0055] To implement the logical AND gate function of the above three conditions at the pure algebraic level, this embodiment quantifies the three criteria through mathematical functions and constructs a joint triggering functional by multiplication. If any condition is not met, the overall result is zero, that is, no subsequent calculation is triggered.

[0056] For example, in this embodiment, continuous time is set. Lower hoisting components The real-time three-dimensional spatial coordinates are The theoretical coordinates of the target installation location are: The radius of the tolerance envelope is Then the joint triggering functional It can be calculated using the following formula:

[0057]

[0058]

[0059] in, This is the Heaviside step function, which takes the value 1 when its independent variable is greater than zero, and 0 otherwise. This function acts as a hard decision gate, mapping continuous distance or velocity quantities to a binary state that is either 0 or 1; the first term... The second term corresponds to the spatial location convergence criterion. The third term corresponds to the convergence criterion for time integrals in the motion state. Corresponding physical identity hardware verification. Figure 2 The timing comparison diagram of the three-channel signals of the joint triggering functional in this embodiment is shown.

[0060] Understandably, the purpose of this formula is to move the signal jitter filtering logic from the software judgment level to the mathematical structure itself through the product structure of the three terms. If any one of the three terms is 0, then... Subsequent topology reorganization operations will never be triggered; only when all three terms are simultaneously 1, i.e. Only when this absolutely certain physical condition is met can legitimate hoisting and positioning status data be generated and subsequent calculations be triggered, fundamentally eliminating invalid high-load calculations caused by false triggers in the edge gateway.

[0061] Step 2: Based on the hoisting and positioning status data, extract the topology incremental features corresponding to the current construction process from the edge computing gateway, inject the topology incremental features into the unclosed foundation structure matrix for deterministic algebraic recombination, generate a real-time evolving dynamic topology stress model, and calculate the dynamic safety boundary threshold of the current unclosed stage in combination with environmental load parameters.

[0062] The unclosed structural matrix refers to the global stiffness matrix that characterizes the overall mechanical state of the steel structure at the end of the previous construction process. It can be understood that during the unclosed construction phase, the dimensions and non-zero element distribution of the global stiffness matrix change with each completed component hoisting and placement. Therefore, this matrix is ​​essentially a time-varying variable that evolves continuously with the construction process, rather than a fixed constant.

[0063] The topological incremental feature refers to the local stiffness change caused by the placement of the new component (or the unloading of temporary supports) and needs to be superimposed on the unclosed structural matrix of the foundation.

[0064] Deterministic algebraic reorganization refers to the process of updating the global stiffness matrix to the current process state by accurately identifying the set of locally active nodes affected by the current process and performing algebraic operations on them, without performing global matrix inversion.

[0065] The dynamic topology stress model refers to the global stiffness matrix and its solution result obtained after deterministic algebraic recombination, which strictly corresponds to the current construction process state at this unique moment.

[0066] It is important to note that existing technologies for monitoring the safety of steel structure construction have a fundamental flaw: they attempt to fit a highly nonlinear, time-varying dynamic system that continuously evolves with each construction process using a static, constant state space. Mathematically, this implies a systematic mismatch between the stiffness matrix and the actual structural state, inevitably leading to false alarms and missed detections. This step addresses this flaw by establishing a dynamic topological stress model that evolves in real time with each construction process.

[0067] In this embodiment, in order to complete the updating and solving of the structural stiffness matrix caused by each component placement in real time under the limited computing resources of the edge gateway, and to fundamentally solve the problem of edge gateway memory overflow (OOM) or computation timeout caused by re-performing the complete matrix inversion of the entire ultra-large-dimensional global stiffness matrix after each update (whose computational complexity increases cubically with the matrix dimension), the process of generating the real-time evolving dynamic topology force model in this step may include the following sub-steps:

[0068] Sub-step 2.1: Analyze the hoisting and positioning status data, extract the physical dimensions and material properties of the newly positioned component, generate the local stiffness matrix of the newly positioned component, and extract its connection node mapping vector in the global coordinate system;

[0069] Sub-step 2.2: Extract the displacement constraints of the currently retained temporary support nodes and generate a set of temporary support constraint vectors;

[0070] Sub-step 2.3: Align the local stiffness matrix to the corresponding dimension of the foundation unclosed structure matrix by connecting the node mapping vectors;

[0071] Sub-step 2.4: Using the block matrix elimination method, without performing global matrix inversion, the aligned local stiffness matrix and the temporary support constraint vector set are superimposed on the unclosed foundation structure matrix, and the updated dynamic topological stress model is output.

[0072] The local stiffness matrix refers to the element stiffness matrix that describes only the mechanical properties of a single newly positioned component, denoted as . Its elements are determined by the physical dimensions of the component (such as the moment of inertia and length of the cross section) and material properties (such as the elastic modulus).

[0073] The node mapping vector (also known as the node mapping Boolean matrix) is a transformation matrix responsible for establishing a one-to-one correspondence between the local degree-of-freedom numbers of a component and the global node numbers, denoted as . Understandably, since each component is defined in its own local coordinate system during manufacturing, while the overall structural analysis must be performed in the global coordinate system, this mapping matrix is ​​necessary to achieve alignment from local to global.

[0074] The temporary support constraint vector set refers to the set consisting of the displacement constraints of all currently retained temporary support nodes, denoted as . Its physical meaning is the equivalent constraint reaction force provided by the temporary support to the structure.

[0075] Having established the above elements, the core insight of this embodiment lies in the fact that steel structures exhibit high sparsity, with each process affecting only a local set of nodes. The stress state of the vast majority of nodes does not change significantly after the component is in place. Therefore, it is unnecessary to resolve the entire matrix; it is sufficient to accurately identify the set of active nodes affected by the current component and strictly limit the computational scale to this extremely small local subset.

[0076] Specifically, when the joint triggering functional When triggered, new component Injection system. First, a Boolean matrix is ​​mapped through nodes. The local element stiffness is assembled into a global incremental stiffness matrix. Thus, the updated system equilibrium equations are obtained.

[0077] For example, in this embodiment, let the previous process (state) be... The global stiffness matrix of ) is The updated system equilibrium equation can then be expressed as follows:

[0078]

[0079] in, The global incremental stiffness matrix introduced in this process is the mathematical implementation of aligning the local stiffness matrix to the corresponding dimension of the unclosed structure matrix in sub-step 2.3 by connecting the node mapping vector; This represents the global node displacement vector to be determined in the current process. This represents the external load vector for the current process. The equivalent nodal reaction force vector provided for temporary supports is given by the temporary support constraint vector set. Sure.

[0080] Understandably, when this process involves removing temporary supports rather than adding new components, One item takes a negative value, and By subtracting the stiffness contribution of the corresponding support from the global matrix, the equation can uniformly express the two types of topological evolution: component addition and support unloading.

[0081] After establishing the incremental update equation, this embodiment further introduces a block matrix elimination strategy to achieve dimensionality reduction. Specifically, all nodes are divided into two mutually exclusive sets according to whether they are affected by the newly added component: the set of affected active nodes (subscript...). ) and the set of unaffected dormant nodes (subscript) Based on this, the updated global matrix is ​​reorganized into memory blocks, resulting in the following system of block linear equations:

[0082]

[0083] in, For the self-coupled stiffness sub-block of the active node set; For the set of dormant nodes, there is a self-coupled stiffness sub-block; and This is a coupled sub-block between the active and dormant regions, describing the mechanical interaction between them; , These are the displacement vectors of the active and dormant nodes, respectively. This is the equivalent node load subvector corresponding to the active node set. The equivalent node load subvector corresponding to the set of dormant nodes.

[0084] It should be noted that, due to the dormant node set corresponding submatrix No changes occurred in this process; the sparse factor decomposition result calculated in the previous process (i.e., for...) remains the same. The factor matrix cached after performing sparse Cholesky decomposition can be directly cached and reused without recalculation. Based on this, the Schur complement matrix is ​​utilized. By extracting the active region from the complete equation through algebraic elimination, the solution can be completed without touching the huge dormant region. This is the essence of step S2.4, which is to solve the problem without performing global matrix inversion.

[0085] For example, in this embodiment, the Schur complement matrix and the active node displacement vector can be obtained by solving the following formula:

[0086]

[0087]

[0088] in, The Schur complement matrix, whose physical meaning is the equivalent stiffness of the active node set after incorporating the mechanical contributions of the dormant region; In actual implementation, it corresponds to... The factor matrix cached after performing sparse Cholesky decomposition is in a read-only locked state and does not need to be recalculated for each operation.

[0089] Understandable The dimension of the matrix is ​​strictly equal to the size of the active node set, and much smaller than the dimension of the global matrix. Therefore, for The computational cost of performing inversion (or solving linear equations) is extremely low. The algorithm rigorously guarantees from a mathematical structure that the actual matrix size involved in the solution by the edge gateway during each process update is determined solely by the size of the active node set, decoupling it from the global structure size. This fundamentally solves the memory overflow and computation timeout problems caused by excessively large matrix dimensions in existing technologies, making real-time dynamic finite element solution engineering feasible on resource-constrained edge devices.

[0090] In this embodiment, to further adapt to hardware scenarios where edge computing gateways have limited memory resources, and to ensure that the aforementioned block elimination strategy is also implemented at the physical memory management level, the processing procedure in the edge computing gateway in this step may include:

[0091] The unclosed structure matrix is ​​divided into fixed-size memory sub-blocks according to the physical construction area; the affected target memory sub-blocks are located based on the three-dimensional spatial coordinates contained in the hoisting and positioning status data; during algebraic reorganization, only the target memory sub-blocks and their adjacent memory sub-blocks shared with boundary nodes are loaded into the edge gateway's fast cache (CPUCache) for incremental calculation, while the remaining unaffected memory sub-blocks are in a read-only locked state.

[0092] In this context, a memory sub-block refers to a fixed-size storage unit obtained by dividing the large global stiffness matrix according to physical construction areas (e.g., by floor, by structural zone, or by hoisting work surface). It is understandable that, due to the spatial locality of steel structures, physically adjacent nodes often form continuous non-zero element regions in the stiffness matrix. Therefore, dividing the memory sub-block according to physical construction areas allows nodes affected by the same construction process to fall into the same sub-block as much as possible, thereby maximizing cache hit rate. Figure 3 A comparison chart of cache hit rates for the memory sub-block strategy in this embodiment is shown; Figure 4 A comparison chart of the computational scale of Schur complement block elimination in this embodiment is shown; Figure 5 A graph showing the comparison of edge gateway memory usage as the process progresses in this embodiment is provided. Figure 6 The diagram shows the percentage of active node sets in steel structure projects of different scales in this embodiment.

[0093] It is understandable that the logic for processing this memory sub-block corresponds completely to the aforementioned block elimination strategy at the mathematical and engineering levels: the target memory sub-block loaded into the CPU cache for incremental computation and its boundary adjacent sub-blocks correspond precisely to the aforementioned self-coupled sub-blocks of the active node set. and active-dormant coupling sub-block , The remaining memory sub-blocks in read-only lock state correspond to the sleep area sub-blocks that have undergone factorization and cache reuse. Thus, the active-dormant partitioning at the algorithm level is physically mapped to the read-write-read-only partitioning at the memory level, allowing the low-computing-power advantage of Schur complement dimensionality reduction to be fully realized on edge gateways with limited hardware resources.

[0094] In this embodiment, considering that the topological incremental feature includes not only the addition of components but also the reverse evolution process of unloading temporary support systems, the processing logic in this step may include:

[0095] When a process confirmation signal for the dismantling of a specific temporary support is obtained, the constraint reaction force of the temporary support in the constraint vector set of the temporary support is extracted; the constraint reaction force is applied as a reverse equivalent nodal load to the existing dynamic topology force model; then the Gaussian elimination method is used to extract the row and column corresponding to the temporary support from the dynamic topology force model to complete the topology reverse evolution.

[0096] Among them, constraint reaction force refers to the supporting force provided by temporary supports to the structure before they are removed. It can be understood that the removal of temporary supports is mechanically equivalent to a sudden unloading process: the reaction force originally borne by the supports disappears at the moment of removal, and the load that was originally balanced by them must be borne by the remaining structure.

[0097] It should be noted that this reverse evolution logic is mathematically strictly consistent with the aforementioned forward incremental update equation: the constraint reaction force is applied as a reverse equivalent nodal load, corresponding to the equation in the previous equation. One term takes a negative value; while using Gaussian elimination to remove the row and column corresponding to the temporary support from the dynamic topology stress model corresponds to subtracting the stiffness contribution of the support from the global stiffness matrix. Thus, the two types of topology evolution, namely component addition (forward) and support unloading (reverse), are incorporated into the same deterministic algebraic reorganization framework, enabling the dynamic topology stress model to track the entire construction process bidirectionally and continuously.

[0098] In this embodiment, in order to establish a real-time environmental-structural load model that is strictly coupled with the current process status, thereby overcoming the systematic deviation caused by the existing technology of using a fixed threshold of the final completed state, the process of calculating the dynamic safety boundary threshold of the current unclosed stage in combination with environmental load parameters in this step may include the following sub-steps:

[0099] Sub-step 2.5: Collect real-time wind speed and direction data at the current construction site and convert them into dynamic wind pressure vectors applied to each exposed node of the dynamic topology stress model;

[0100] Sub-step 2.6: Extract the temporary eccentric load value of the semi-rigid connection nodes in the current unclosed structure that have not been permanently fixed (welded or final tightened with high-strength bolts);

[0101] Sub-step 2.7: Use the dynamic wind pressure vector and the temporary eccentric load value as input column vectors, perform matrix multiplication with the updated dynamic topology stress model, calculate the theoretical stress extreme value and deformation extreme value of each node under the current topology state, and set them as the dynamic safety boundary threshold.

[0102] It is important to note that the stress state of an unclosed steel structure during construction differs fundamentally from that of a completed structure: its exposed windward surface area, structural eccentricity, and overall stiffness all change in real time with each construction phase. The dominant dynamic load on the structure during construction is wind load, the magnitude of which is determined by two time-varying factors: first, the real-time changing wind speed and direction, which directly affect the magnitude of dynamic pressure and the direction of load distribution; and second, the actual exposed windward surface area of ​​the structure under the current construction phase, which is updated in real time as components are added or removed.

[0103] The dynamic wind pressure vector refers to the instantaneous wind load column vector applied to each exposed node of the structure, which is determined by the real-time wind speed, wind direction, and the current exposed windward surface area.

[0104] Semi-rigid connection nodes refer to connection nodes in a transitional state where full-section welding or final tightening of high-strength bolts has not been completed during the construction phase, and the connection stiffness is between that of an ideal hinged connection and an ideal rigid connection. Understandably, because the connection has not reached the final design state, such nodes will generate additional eccentric moments, which must be included as additional loads.

[0105] For example, in this embodiment, the current process Real-time dynamic wind pressure vector and total load vector They can be calculated using the following formulas:

[0106]

[0107]

[0108] in, This is the standard Bernoulli dynamic pressure formula. air density, The wind speed is collected in real time by the weather station; This is a vector of shape coefficients, reflecting the aerodynamic characteristics of various parts of the structure; For the current process The actual exposed windward surface area vector of the unclosed structure is updated each time a component is placed. This is the Hadamard element-wise product symbol, used to multiply the body shape coefficient by the corresponding windward area of ​​each part; Based on real-time wind direction angle The wind pressure distribution mapping matrix decomposes the resultant force into the directions of each node; This refers to the temporary eccentric moment caused by the semi-rigid connection during the current process. The node space vector, together with the eccentric moment, is used to calculate the equivalent nodal additional load. Figure 7 The graph shows the evolution of the real-time dynamic wind pressure vector with wind speed, wind direction angle and process number in this embodiment; Figure 8 The figure shows a polar coordinate comparison of the wind pressure distribution mapping matrix under different processes in this embodiment as the wind direction angle changes. As can be seen from the figure, as the process progresses, the change in structural form causes the most dangerous wind direction and the direction of the maximum load to change continuously, rather than remain fixed.

[0109] After establishing the real-time total load vector, it is substituted into the structural displacement field obtained by the aforementioned block elimination model. Combined with the finite element constitutive relation, the nodal displacements can be transformed into nodal strains, thereby obtaining the dynamic safety boundary threshold strictly corresponding to the current process. This transformation corresponds to the mathematical essence of performing matrix multiplication between the input column vector and the updated dynamic topological force model in sub-step 2.7.

[0110] For example, in this embodiment, the current process The dynamic strain extreme value matrix (i.e., the safety boundary threshold). It can be calculated using the following formula:

[0111]

[0112] in, is the strain-displacement transformation matrix for finite element strain, whose elements are composed of the spatial partial derivatives of the element geometry function, which maps nodal displacements to strains of each element. The constitutive elastic matrix of steel reflects the linear elastic relationship between stress and strain. This refers to the node displacement vector obtained by the aforementioned block elimination model under the current process. In the engineering implementation, a global inversion is not actually performed; instead, the solution is obtained by reusing the aforementioned Schur complement dimensionality reduction result. Figure 9 A full-process comparison diagram of the dynamic strain safety boundary in this embodiment and the static threshold of the prior art is shown; Figure 10 The thermal diagram of stress redistribution at key structural nodes before and after unloading of the temporary support in this embodiment is shown.

[0113] It should be noted that all key variables in this formula are strictly labeled with process indexes. This means the safety boundary threshold It is a dynamic quantity that updates gradually with each process, rather than a fixed constant. This mathematically directly answers the root cause of false alarms in existing technologies: existing technologies are equivalent to fixed usage. and The model calculates the safety boundaries of all construction stages using the final completed state parameters, while this model strictly calculates the corresponding safety boundaries using the process state parameters at the current moment, thus completely eliminating the systematic mismatch between static thresholds and time-varying structures.

[0114] Step 3: Compare the real-time physical state data with the dynamic safety boundary threshold. If it is determined that the real-time physical state data deviates from the dynamic safety boundary threshold and meets the cascade instability judgment condition, then generate and output a shutdown warning command.

[0115] Among them, the cascading instability judgment criteria refer to a set of deterministic criteria used to determine whether excessive strain in local components or temporary supports will trigger a chain reaction of collapses in the overall structure. Understandably, the problem in steel structure construction safety monitoring is that excessive strain in local components or temporary supports is often not an isolated safety event, but may be the starting point of a cascading collapse. When a node fails, the load it originally bore will transfer to adjacent nodes. If the adjacent nodes exceed their yield limits as a result, the load will continue to spread to more distant nodes, ultimately triggering a domino-like overall overturning.

[0116] A shutdown warning command is a control signal that the system immediately outputs when the cascading instability judgment condition is met, requiring the on-site hoisting operation to be stopped and emergency response to be initiated. It is denoted as... .

[0117] It should be noted that traditional node-by-node threshold comparison algorithms cannot predict load transfer paths or provide early warnings before cascading failures occur, resulting in a fundamental safety blind spot. Therefore, this embodiment designs an algorithm capable of deterministically deduce load transfer paths and predict whether cascading instability will occur within milliseconds, without relying on time-consuming full nonlinear elastoplastic time history analysis.

[0118] In this embodiment, in order to perform a deterministic prediction of the unique disaster mode of steel structure construction—the collapse of the entire framework due to the failure of a temporary support—within milliseconds using pure algebraic operations, the process of determining whether the cascading instability judgment condition is met in this step includes the following three sub-steps:

[0119] Sub-step 3.1: When the real-time physical state data of a target node exceeds the dynamic safety boundary threshold, the stiffness parameter of the target node is forced to zero in the dynamic topology force model to generate a residual topology matrix; where the residual topology matrix refers to the stiffness matrix that reflects the residual structural state of the target node after it has left the load-bearing system, which is constructed by applying a local zeroing operation to the global stiffness matrix.

[0120] Specifically, when the strain data collected in real time by the IoT sensing matrix Target node detected Meets the conditions for exceeding limits Immediately, the loss of the node's load-bearing capacity is mathematically simulated to provide a correct structural state basis for subsequent load redistribution simulations.

[0121] For example, in this embodiment, the damaged topology matrix It can be constructed using the following formula:

[0122]

[0123] in, For a mask matrix, it is only in the first... The diagonal elements are all 1, and the rest are all 0; this operation subtracts the nodes from the global matrix. The stiffness contribution in the corresponding row and column directions is simulated at the algebraic level. The residual structural state after exiting the load-bearing system is the mathematical implementation of forcibly setting the stiffness parameter of the target node to zero in step S3.1.

[0124] Sub-step 3.2: Based on the residual topology matrix, calculate the load redistribution coefficients for the load originally borne by the target node that is transferred to its first- and second-order adjacent nodes; the deterministic calculation of the load redistribution coefficients includes:

[0125] node After failure, the load it originally bore It will be released and placed into the same connected component set. The transition between adjacent nodes in the connected component set. It should be noted that the connected component set... It includes the target node The first-order adjacent nodes that are directly connected also include the second-order adjacent nodes that are indirectly connected through the first-order nodes, thus covering the near-field propagation path of load transfer.

[0126] The load distribution ratio is determined by two physical factors: adjacent nodes. With failure node Residual connection stiffness between (The greater the stiffness, the stronger the load absorption capacity), and the spatial distance between the two. (The closer the distance, the more direct the load transfer). By constructing these two factors into a normalized weighting coefficient, the load increment absorbed by each adjacent node can be deterministically calculated.

[0127] For example, in this embodiment, the load weighting is... and adjacent nodes The obtained load increment It can be calculated using the following formula:

[0128]

[0129]

[0130] in, For nodes in the damaged topology matrix With nodes The connection stiffness elements between them (i.e., extracted from the residual topology matrix constructed in sub-step 3.1) This ensures that the redistribution occurs on the node. (Performed under the correct structural condition that has failed). For nodes With nodes The spatial geometric distance between them; The spatial distance penalty factor, expressed as an inverse square of the distance, reflects the mechanical law that loads are preferentially transferred to nearest neighbor nodes; the denominator... For sets Sum all adjacent nodes within the range to ensure weight. The normalization condition must be met, meaning the sum of the load increments absorbed by all adjacent nodes must be exactly equal to the total load released by the failed node. .

[0131] Understandable The algorithm achieves a dual-weighted distribution calculation of the failure load by the remaining nodes, based on both stiffness ratio and distance proximity. This design, which normalizes the distribution by multiplying the residual linear stiffness by the inverse square of the spatial distance, allows the determination of the load transfer path to be completed deterministically at the pure algebraic level without any complete nonlinear time history solution. This is the core of the algorithm's millisecond-level ultra-fast prediction capability.

[0132] Sub-step 3.3: If the expected stress value of any adjacent node calculated based on the load redistribution coefficient exceeds its corresponding yield extreme value, then the current structural state is determined to have a domino-like collapse risk, and the cascading instability judgment condition is met.

[0133] After obtaining the load increments of each adjacent node, these increments are superimposed onto the current measured strain of that node to obtain the expected stress level after absorbing the transferred load. This expected stress is then compared with the yield strength of the steel, and the instability cascade index is defined by accumulating the number of nodes exceeding the limit. .

[0134] For example, in this embodiment, adjacent nodes Expected force and cascade instability index It can be calculated using the following formula:

[0135]

[0136]

[0137] in, For nodes The measured strain value at the current moment; The mapping function that converts the load increment into the equivalent strain increment is given by the constitutive relation; This represents the yield stress limit of the steel, i.e., the extreme yield value corresponding to each node. It is still the Heaviside step function, which contributes 1 when the expected stress exceeds the yield limit, and 0 otherwise. That is, a set The number of nodes expected to be crushed by load transfer. Figure 11 The diagram illustrates the evolution of the cascade instability index Γ over time and the triggering time of the shutdown warning command in this embodiment.

[0138] It should be noted that when When at least one adjacent node exceeds the yield limit due to load transfer, the current structural state is deemed to have a domino-like collapse risk, and the cascading instability judgment condition is met. The system immediately generates and outputs a work stoppage warning command. Understandably, this criterion, by linking two levels of conditions—local overrun and overrun of adjacent nodes after load transfer—provides a rapid and reliable hard logic interception specifically for the unique disaster mode in steel structure construction where the failure of a temporary support leads to the collapse of the entire framework.

[0139] A deterministic data interpolation recovery mechanism for network partitioning or sensor hardware failure scenarios.

[0140] In this embodiment, considering that strong electromagnetic interference at the construction site, metal components blocking wireless signals, and network partitioning failures may all cause some node sensors to fail to report data for a certain period of time, in order to avoid interrupting the safety monitoring of missing nodes and causing information blind spots in the cascaded early warning logic, and to avoid the risk of interpolation distortion introduced by general statistical methods such as Kalman filtering or neural network mean interpolation, which may not necessarily satisfy the mechanical equilibrium constraints of the current unclosed state, a deterministic data interpolation recovery mechanism can also be included between steps 1 and 3:

[0141] When the heartbeat packet of the target IoT sensing device is lost, the spatial adjacency node matrix of the failed node where the device is located is extracted from the dynamic topology force model; the real-time physical state data of each normal node in the spatial adjacency node matrix is ​​obtained; by applying reverse node force in the dynamic topology force model, the virtual strain data of the failed node is calculated by back-calculating using the equilibrium equation; and the virtual strain data is used to replace the missing real-time physical state data for comparison in step 3.

[0142] It is worth noting that the fundamental solution of this mechanism lies in the fact that this system has already established a complete and real-time updated dynamic topological stiffness matrix in step 2 above. We can use this known and defined structural mechanics model to rigorously deduce the virtual displacement and strain that satisfy physical equilibrium at the missing node by reverse algebraic solution, starting from the known displacement of the healthy node, without any statistical assumptions.

[0143] Specifically, in the state The global node displacement vector is divided into two subsets according to the sensor's operating state: the known displacement vectors corresponding to the set of healthy nodes where the sensor is operating normally. And the displacement vector to be determined for the set of failure nodes where the sensor is offline. Based on this, the global static equilibrium equations are rearranged into blocks according to the same node partitions, resulting in the following block-based linear equation system:

[0144]

[0145] Extracting the second line of the block equation yields a result containing only unknowns. Independent equations:

[0146]

[0147] Regarding the equation By performing an algebraic solution, the virtual physical state of the missing node can be obtained by reverse balancing.

[0148] For example, in this embodiment, the virtual displacement vector of the missing node It can be calculated using the following formula:

[0149]

[0150] in, For self-coupled stiffness sub-blocks between healthy nodes; and This is a mutually coupled sub-block between healthy nodes and failed nodes, describing the mechanical transmission relationship between them; For self-coupling stiffness sub-blocks between failed nodes; This is the external force vector at the failure node.

[0151] Understandably, since the number of disconnected nodes is usually much smaller than the total number of nodes, The dimension is extremely small, and the computational cost of inverting it is negligible. Furthermore, in most monitoring scenarios, the offline node does not have a direct external load applied, in which case... The formula is further simplified to That is, the virtual displacement of the missing node is uniquely determined by the known displacement of the healthy node through the mechanical transmission relationship.

[0152] After obtaining the virtual displacement Then, through the strain-displacement transformation matrix Further differentiation yields the corresponding virtual strain data:

[0153]

[0154] The virtual strain data is then seamlessly injected into the cascaded early warning logic in step S3, participating in the over-limit comparison at the missing node, ensuring that the entire safety monitoring system does not have any safety blind spots in extreme scenarios where any number of sensors are offline.

[0155] It should be noted that the essential difference between this mechanism and statistical interpolation methods lies in the fact that virtual strain... Instead of being a statistical estimate of historical data, the interpolation is a physically calculated value uniquely determined through inverse algebraic solution under the constraints of the currently defined dynamic topological stiffness matrix. This mathematically guarantees that the interpolation result 100% satisfies the mechanical equilibrium constraints of the current unclosed structure, without introducing any additional statistical uncertainty. This allows the system to maintain a safety judgment capability equivalent to normal operating conditions even when facing extreme communication failures. Figure 12 The diagram shows a comparison of the early warning timelines of this scheme and the traditional scheme in a simulated temporary support sudden overload accident in this embodiment.

[0156] Example 2

[0157] This embodiment discloses an Internet of Things-based steel structure construction safety early warning system.

[0158] Specifically, the IoT-based steel structure construction safety early warning system can be integrated into electronic devices, such as edge computing gateways, terminals, and servers. The edge computing gateway can be an industrial-grade embedded computing unit deployed at the construction site, such as the tower crane base, temporary power distribution room, or core tube of the structure. The terminal can be an industrial tablet, handheld terminal, or personal computer used by on-site supervisors. The server can be a single server, a server cluster consisting of multiple servers, or a cloud-edge integrated architecture where cloud servers and edge gateways collaborate. When the electronic device is running, it implements the IoT-based steel structure construction safety early warning method as described in Example 1.

[0159] In this embodiment, the IoT-based steel structure construction safety early warning system can also be integrated into multiple electronic devices. For example, the edge computing gateway deployed on site can undertake topology reorganization and cascading early warning calculation with extremely high real-time requirements, while the cloud server can undertake non-real-time BIM process list management and historical data archiving. The two work together to realize the IoT-based steel structure construction safety early warning method in Embodiment 1 of this application.

[0160] In this embodiment, considering the common problems of network partitioning, strong electromagnetic interference and unstable communication links in steel structure construction sites, the core computing power is moved to the edge computing gateway for local execution, which can maintain the continuity of security monitoring capabilities even in extreme working conditions where the connection with the cloud is lost.

[0161] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for early warning of steel structure construction safety based on the Internet of Things, characterized in that, The steel structure construction safety early warning method includes: Acquire the first and second state data of the monitored structure, wherein, The first state data indicates the real-time mechanical response of each monitored node in the monitored structure, and the second state data indicates the placement status of the new component in the monitored structure. Based on the second state data, the topological incremental features corresponding to the placement of the new component are injected into the basic structure model to generate a dynamic structural model that evolves in real time with the construction process. The basic structural model represents the overall mechanical state of the monitored structure at the end of the previous construction process; Based on the dynamic structure model, determine the dynamic safety boundary threshold corresponding to the current construction process status; The first state data is compared with the dynamic safety boundary threshold. When it is determined that the first state data deviates from the dynamic safety boundary threshold, an early warning command is output.

2. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 1, characterized in that, The acquisition of the second state data of the monitored structure includes: The second state data is generated only when the new component simultaneously satisfies the spatial convergence condition, the motion convergence condition, and the identity verification condition. The spatial convergence condition is that the deviation between the current spatial position of the new component and the target installation position falls within a preset tolerance range. The motion convergence condition is that the amount of motion of the new component within a set time window is lower than the static determination threshold. The identity verification condition is that the identification code of the new component matches the preset process list. And the contact trigger signal corresponding to the new component is triggered.

3. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 2, characterized in that, The motion amount of the new component within the set time window in the motion convergence condition is the time average of the motion rate of the new component within the set time window. When the average time value is lower than the static determination threshold, the motion convergence condition is determined to be met; The current spatial position of the new component is continuously collected by the first positioning device. The identification code is obtained by reading it from the second identification device. Furthermore, the second identification device is triggered to read when both the spatial convergence condition and the motion convergence condition are met; The first positioning device includes an ultra-wideband positioning device, the second identification device includes a radio frequency identification device, and the preset process list includes a building information model process list.

4. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 1, characterized in that, The generation of the dynamic structure model includes: Based on the second state data, the physical dimensions and material properties of the new component are extracted, a first sub-matrix characterizing the mechanical properties of the new component is generated, and the mapping vector of the new component in the global coordinate system is extracted. The mapping vector is used to establish the correspondence between the local degrees of freedom and global nodes of the new component. Extract the displacement constraints of the currently retained temporary support nodes in the monitored structure and generate a constraint vector set; The first sub-matrix is ​​aligned to the corresponding dimension of the basic structure model using the mapping vector, serving as the topological incremental feature; Without performing a global inversion on the basic structure model, the aligned first submatrix and the constraint vector set are superimposed on the basic structure model to obtain the dynamic structure model; Superimposing the aligned first submatrix and the constraint vector set onto the basic structure model without performing a global inversion on the basic structure model includes: The nodes in the basic structure model are divided into an active node set affected by the placement of the new component and a dormant node set not affected by the placement of the new component. The factor cache obtained from the dormant node set in the previous construction process is reused to determine the cohesive equivalent stiffness of the active node set after aggregating the mechanical contribution of the dormant node set. The dynamic structural model is obtained by solving the state of the active node set based on the condensed equivalent stiffness. The factor cache is the result of decomposing the corresponding part of the dormant node set and caching it, which is in a read-only reuse state during this construction process.

5. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 1, characterized in that, The dynamic security boundary threshold includes: Collect the environmental load parameters of the current construction site and convert the environmental load parameters into dynamic load vectors applied to each exposed node of the dynamic structural model; Extract the eccentric load value of the semi-rigid connection node in the monitored structure; The dynamic load vector and the eccentric load value are applied as inputs to the dynamic structural model to determine the theoretical stress extreme value and theoretical deformation extreme value of each monitored node under the current construction process state. The theoretical stress extreme value and the theoretical deformation extreme value are then set as the dynamic safety boundary threshold.

6. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 1, characterized in that, The step of outputting a warning instruction when it is determined that the first state data deviates from the dynamic safety boundary threshold includes: When it is determined that the first state data deviates from the dynamic safety boundary threshold and meets the cascade instability judgment condition, the warning command is output; The determination of whether the cascade instability criteria are met includes: The nodes whose first state data deviates from the dynamic safety boundary threshold among the monitored nodes are identified as target nodes; In the dynamic structural model, the stiffness parameter of the target node is set to zero to generate a residual topology model; Based on the damaged topology model, determine the load redistribution coefficient for transferring the load originally borne by the target node to the adjacent nodes of the target node; When the expected stress value of any of the adjacent nodes, determined according to the load redistribution coefficient, exceeds the corresponding yield extreme value, the cascade instability judgment condition is determined to be satisfied.

7. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 6, characterized in that, In the dynamic structural model, the stiffness parameter of the target node is set to zero to generate a residual topology model, including: Subtracting the stiffness contributions of the target node in the row and column directions from the dynamic structural model yields the residual topology model characterizing the residual structural state after the target node leaves the load-bearing system. The adjacent nodes include first-order adjacent nodes that are directly connected to the target node, and second-order adjacent nodes that are indirectly connected to the target node through the first-order adjacent nodes. The load redistribution coefficient is determined based on the residual connection stiffness between each of the adjacent nodes and the target node, and the spatial distance between each of the adjacent nodes and the target node, wherein... The residual connection stiffness is taken from the damaged topology model, and the spatial distance is determined using an inverse square ratio. Furthermore, the load redistribution coefficients corresponding to each of the adjacent nodes are normalized so that the sum of the load increments obtained by each of the adjacent nodes is equal to the load released by the target node.

8. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 7, characterized in that, The expected stress value of any of the adjacent nodes is the sum of the first state data of the adjacent node at the current moment and the strain increment of the adjacent node after constitutive transformation based on the load redistribution coefficient. When the number of nodes in the adjacent nodes whose expected stress value exceeds the corresponding yield extreme value reaches at least one, the cascade instability judgment condition is determined to be met and the warning command is output.

9. The method for early warning of steel structure construction safety based on the Internet of Things according to claim 1, characterized in that, The steel structure construction safety early warning method also includes: When the heartbeat of the data stream of the target sensing device is lost, the monitored node corresponding to the target sensing device is identified as a failed node, and the spatial adjacency node matrix of the failed node is extracted in the dynamic structure model. The spatial adjacency node matrix represents the mechanical transmission relationship between the failed node and the healthy node with normal data. The first state data of the healthy node is obtained, and a reverse node force is applied to the failed node. The virtual state data of the failed node is determined by reverse calculation based on the balance relationship of the dynamic structure model. The virtual state data replaces the missing first state data of the failed node and is used to compare the first state data with the dynamic security boundary threshold. When no direct external load is applied to the failed node, the force of the reverse node is zero, and the virtual state data of the failed node is uniquely determined by the mechanical transmission relationship represented by the first state data of the healthy node through the spatial adjacency node matrix.

10. A steel structure construction safety early warning system based on the Internet of Things, characterized in that, The steel structure construction safety early warning system includes: processor; A memory storing a computer program, which, when executed by a processor, implements the IoT-based steel structure construction safety early warning method as described in any one of claims 1 to 9.

Images