Reverse modeling compensation method for concrete shrinkage deformation

Abstract

The invention provides a concrete shrinkage deformation reverse modeling compensation method, and belongs to the technical field of wind tunnel construction, and the method comprises the steps: building a pouring unit adjacent relation topological graph, optimizing a pouring time sequence through a graph coloring algorithm, and building a heat-humidity-force multi-physics field coupling finite element model; kriging interpolation is adopted to reconstruct complete temperature field distribution and extract a temperature deviation field, Kalman filtering inversion is used to correct hydration heat source parameters, a moving boundary identification algorithm is used to update dynamic boundary conditions, an incremental iteration algorithm is used to solve coupling response, and a Green function stress response matrix is used to quickly calculate stress field distribution. When the stress value of the key position exceeds the threshold value, reverse compensation calculation is started to generate a template adjustment instruction, actually measured data is fed back to the model to form closed-loop iteration, and the technical problem of wind tunnel structure construction deformation control failure caused by insufficient concrete shrinkage deformation prediction precision is solved.

Description

Technical Field

[0001] This invention belongs to the field of wind tunnel construction technology, and specifically relates to a method for reverse modeling and compensation of concrete shrinkage deformation. Background Technology

[0002] In the concrete construction of large wind tunnel structures, shrinkage deformation control is a key technology for ensuring structural accuracy. Traditional methods use finite element method (FEM) simulation to predict shrinkage deformation, and adjust the formwork elevation based on the prediction results to achieve deformation compensation. However, the prediction model typically uses fixed boundary conditions and material parameters for a single forward calculation, and during construction, partial temperature and deformation data are obtained manually for empirical judgment. Traditional techniques have significant drawbacks: first, the fixed parameters and boundary conditions of the FEM model fail to reflect the dynamic effects of support and environmental changes during construction; second, the internal temperature field of the concrete relies on a limited number of measuring points for estimation, resulting in low accuracy in reconstructing the internal temperature distribution of large-volume components; and third, the multi-physics coupling effect of temperature, humidity, and stress fields is simplified to a single-field analysis, neglecting the nonlinear interactions between these fields. In existing technologies, due to the fixed model parameters and the lack of a dynamic feedback correction mechanism based on measured data, there is a significant deviation between the predicted shrinkage deformation and the actual deformation. When construction progresses to the later stages, accumulated errors cause structural deformation to exceed the allowable range, rendering compensation measures ineffective and failing to meet the high-precision requirements of wind tunnel structures. In other words, existing technologies suffer from insufficient accuracy in predicting concrete shrinkage deformation, leading to the failure of deformation control during wind tunnel construction. Summary of the Invention

[0003] In view of this, the present invention provides a method for reverse modeling and compensation of concrete shrinkage deformation, which can solve the technical problem of insufficient prediction accuracy of concrete shrinkage deformation in the prior art, leading to the failure of deformation control during wind tunnel construction.

[0004] This invention is implemented as follows: A method for reverse modeling and compensation of concrete shrinkage deformation is provided. A distributed fiber optic temperature sensor network and humidity sensor array are deployed in the concrete pouring area of ​​a wind tunnel structure. Displacement sensors and strain sensors are installed at key constraint locations to construct a multi-physics monitoring system. The wind tunnel structure is divided into several pouring units according to construction zones, and a topological graph of the adjacent relationships between these units is established. A graph coloring algorithm is used to assign pouring sequence numbers and optimize the temperature difference between adjacent pouring units. A coupled finite element model of thermal-humidity-mechanical multi-physics fields is established. Quasi-continuous temperature data is collected, and a complete temperature field distribution is reconstructed using a three-dimensional spatial Kriging interpolation method. The temperature deviation field is extracted, and Kalman filtering is used to invert and correct the hydration heat source parameters. A moving boundary recognition algorithm is used to identify dynamic boundary condition changes and input them into the model. An incremental iterative algorithm and a time step adaptive adjustment strategy are used to solve the coupled response and output the predicted shrinkage deformation field. The stress field distribution is calculated using the Green's function stress response matrix. When the stress value at a key location exceeds a threshold, reverse compensation calculation is initiated to generate a template adjustment command. The pre-deformation adjustment amount is superimposed onto the template design elevation to form the compensated template installation elevation. Measured data is fed back to the model and iterated cyclically.

[0005] The distributed fiber optic temperature sensor network employs fiber Bragg grating technology or Raman scattering technology, with a measurement point spacing of 0.5m to 1m.

[0006] In the adjacent relationship topology diagram of the casting units, nodes represent casting units, edges connect adjacent casting units, and the weight of the edges represents the contact area of ​​adjacent surfaces.

[0007] The graph coloring algorithm assigns a color to each node in the graph so that adjacent nodes have different colors, and different colors represent different pouring batches.

[0008] Specifically, the step of optimizing the temperature difference between adjacent pouring units involves calculating the initial temperature difference between adjacent pouring units based on the pouring sequence number. When the initial temperature difference exceeds 15℃, the pouring sequence number is adjusted and the graph coloring is re-optimized until the temperature difference between all adjacent pouring units meets the constraint requirement of not exceeding 15℃, at which point the optimal pouring scheme is generated.

[0009] In the thermo-humidity-mechanical multiphysics coupled finite element model, the temperature field control equation describes the hydration heat conduction process, the humidity field control equation describes the moisture diffusion process, and the stress field control equation describes the shrinkage deformation response. The three field equations are related through a coupling coefficient matrix.

[0010] The three-dimensional spatial kriging interpolation method is based on regional variable theory and performs unbiased optimal estimation of unknown locations by calculating the spatial correlation and variogram of known measurement point data.

[0011] The temperature deviation field refers to the difference distribution between the complete temperature field distribution and the forward modeling temperature field calculated by the coupled finite element model of the thermo-humidity-mechanical multiphysics field at various points in space.

[0012] The Kalman filter inversion correction calculates the hydration heat source intensity and thermal conductivity coefficient in the thermo-humidity-mechanical multiphysics coupled finite element model by minimizing the weighted sum of squares of the temperature deviation field.

[0013] The moving boundary recognition algorithm dynamically identifies and updates boundary condition changes in the thermo-humidity-mechanical multiphysics coupled finite element model based on real-time sensor monitoring data, including constraint release caused by support removal and constraint application caused by template installation.

[0014] The incremental iterative algorithm discretizes the time domain into several time steps, and iteratively solves the temperature field, humidity field, and stress field within each time step until the field variables satisfy the coupling equilibrium condition.

[0015] The adaptive time step adjustment strategy dynamically adjusts the time increment step size based on the convergence speed and accuracy requirements of the coupled computation. When the iteration residual is less than... Convergence is determined at that time.

[0016] Specifically, the Green's function stress response matrix pre-calculates the Green's function for the basic contraction condition and stores it in matrix form. The predicted contraction deformation field is decomposed into a linear combination of the basic contraction conditions, and the stress field distribution is calculated through matrix multiplication.

[0017] The stress values ​​at key locations refer to the stress components extracted from the stress concentration area of ​​the structure, the location of large cross-sectional changes, and the intersection of multiple constraints, including the maximum principal stress and the maximum tensile stress.

[0018] Specifically, the step of initiating reverse compensation calculation involves calculating the prestress magnitude and pre-deformation adjustment amount when the tensile stress in the stress value at the critical location exceeds 80% of the tensile strength of the concrete.

[0019] Specifically, the iterative step involves feeding back the measured shrinkage deformation data and measured stress data of the current casting unit to the thermo-humid-mechanical multiphysics coupled finite element model and updating the material parameters and boundary conditions before the construction of the next casting unit.

[0020] This invention proposes a reverse modeling and compensation method for concrete shrinkage deformation. It establishes a coupled finite element model of heat, humidity, and stress, employs distributed fiber optic temperature measurement technology to acquire quasi-continuous temperature data, and reconstructs the complete temperature field distribution using three-dimensional Kriging interpolation. Based on the temperature deviation field, the model parameters are corrected using Kalman filtering inversion. Simultaneously, a moving boundary identification algorithm is used to dynamically update boundary conditions. An incremental iterative algorithm is used to solve the coupled response, and the stress field distribution is quickly calculated using the Green's function stress response matrix. When the stress value at a key location exceeds a threshold, reverse compensation calculation is initiated to generate template adjustment instructions. Before the construction of the next pouring unit, measured data is fed back to the model to achieve iterative optimization. This invention accurately describes the nonlinear interaction between temperature, humidity, and stress through a coupled multi-physics model, overcoming the limitations of traditional single-field analysis. Distributed fiber optic temperature measurement and Kriging interpolation reconstruction solve the problem of blind spots in the internal temperature field measurement of large-volume concrete. Kalman filtering inversion correction and the moving boundary identification algorithm achieve dynamic updating of model parameters and boundary conditions. The feedback of measured data forms a closed-loop control, eliminating the cumulative effect of prediction errors. In summary, this invention achieves high-precision prediction and real-time compensation control of concrete shrinkage deformation by constructing a dynamic feedback correction multiphysics coupling model system, thus solving the technical problem mentioned in the background art of insufficient prediction accuracy of concrete shrinkage deformation leading to failure of deformation control during wind tunnel construction. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method involved in the present invention.

[0022] Figure 2 The temperature and stress field distribution cloud maps are shown in the embodiment.

[0023] Figure 3 This is a schematic diagram of the reverse compensation adjustment of the template elevation in the embodiment.

[0024] Figure 4 The temperature, humidity, and stress time history curves in the embodiment are shown.

[0025] Figure 5 This is a scatter plot comparing the reverse compensation effects in the embodiments. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0027] like Figure 1 As shown, the present invention provides a method for reverse modeling and compensation of concrete shrinkage deformation, comprising:

[0028] S01. Deploy a distributed fiber optic temperature sensor network and humidity sensor array in the concrete pouring area of ​​the wind tunnel structure. At the same time, install displacement sensors and strain sensors at key constraint locations to build a multi-physics field monitoring system. Divide the wind tunnel structure into several pouring units according to the construction zones and establish a topology diagram of the adjacent relationships of the pouring units.

[0029] S02. Using a graph coloring algorithm, assign pouring sequence numbers to each node in the adjacent relationship topology graph of the pouring unit. Calculate the initial temperature difference value of adjacent pouring units based on the pouring sequence number. When the initial temperature difference value exceeds 15℃, adjust the pouring sequence number and re-optimize the graph coloring until the temperature difference value of all adjacent pouring units meets the constraint requirement of not exceeding 15℃, and then generate the optimal pouring scheme.

[0030] S03. Establish a multi-physics coupled finite element model of the wind tunnel structure, collect quasi-continuous temperature data obtained by the distributed fiber optic temperature sensor network, reconstruct the complete temperature field distribution using the three-dimensional spatial kriging interpolation method, compare the complete temperature field distribution with the forward modeling temperature field calculated by the multi-physics coupled finite element model, and extract the temperature deviation field.

[0031] S04. Based on the temperature deviation field, perform Kalman filter inversion correction on the hydration heat source parameters in the thermo-humidity-mechanical multiphysics coupled finite element model, collect real-time monitoring data from the displacement sensor and the strain sensor, and use the moving boundary recognition algorithm to identify dynamic boundary condition changes, and input the dynamic boundary condition changes into the corrected thermo-humidity-mechanical multiphysics coupled finite element model.

[0032] S05. The temperature field-humidity field-stress field coupled response of the thermal-humidity-mechanical multi-physics coupled finite element model is solved by using an incremental iterative algorithm and a time step adaptive adjustment strategy. The Green's function stress response matrix is ​​pre-calculated and stored. The predicted shrinkage deformation field is decomposed into a linear combination of basic shrinkage conditions and the stress field distribution is calculated by multiplying the Green's function stress response matrix.

[0033] S06. Extract the stress values ​​at key locations in the stress field distribution. When the tensile stress in the stress values ​​at key locations exceeds 80% of the tensile strength of the concrete, start the reverse compensation calculation, calculate the prestress magnitude and pre-deformation adjustment amount, generate the template adjustment command based on the prestress magnitude and the pre-deformation adjustment amount, and superimpose the pre-deformation adjustment amount onto the template design elevation to form the compensated template installation elevation.

[0034] S07. Before the construction of the next pouring unit, the measured shrinkage deformation data and measured stress data of the current pouring unit are fed back to the thermo-humidity-mechanical multiphysics coupled finite element model and the material parameters and boundary conditions are updated. Steps S03 to S06 are executed in a loop to realize dynamic reverse modeling compensation of the entire construction process.

[0035] Among them, the distributed fiber optic temperature sensor network refers to a fiber optic sensing system that uses fiber Bragg grating technology or Raman scattering technology to achieve quasi-continuous temperature measurement along the length of the fiber, with a measurement point spacing of 0.5m to 1m, covering key areas inside the concrete. Quasi-continuous temperature data refers to temperature measurement data where the measurement point density is high enough that the temperature changes between adjacent measurement points can be effectively interpolated and reconstructed. The multi-physics monitoring system refers to an integrated sensor system capable of simultaneously monitoring the temperature field, humidity field, displacement field, and strain field, used to acquire the spatiotemporal distribution information of multiple physical quantities during concrete shrinkage.

[0036] The topological graph of adjacency relationships between casting units refers to a mathematical model describing the spatial adjacency relationships between casting units using graph theory. Nodes represent independent casting blocks, edges connect adjacent casting blocks, and the weight of an edge represents the contact area of ​​adjacent faces. Graph coloring algorithm is a mathematical method that assigns a color to each node in the graph such that adjacent nodes have different colors. In casting sequence optimization, different colors represent different casting batches, ensuring sufficient time intervals between adjacent casting units. Casting sequence number refers to the construction order number assigned to each casting unit according to the graph coloring algorithm. Initial temperature difference value refers to the temperature difference between adjacent casting units calculated based on hydration heat simulation after the initial assignment of casting sequence numbers. Optimal casting scheme refers to the construction sequence arrangement scheme that satisfies the temperature difference constraint and minimizes the number of casting batches.

[0037] The thermo-humidity-mechanical multiphysics coupled finite element model refers to a numerical calculation model that simultaneously considers the three physical processes of temperature field conduction, humidity field diffusion, and stress field deformation, as well as their interactions. The temperature field governing equation is based on Fourier's law of heat conduction and considers the hydration heat source term; the humidity field governing equation is based on Fick's law of diffusion and considers the effect of temperature on the diffusion coefficient; and the stress field governing equation is based on the elasticity equilibrium equation and considers temperature contraction strain and drying contraction strain.

[0038] Three-dimensional spatial Kriging interpolation is a spatial interpolation technique based on regionalized variable theory. It calculates the spatial correlation and variogram of known measurement point data to provide unbiased optimal estimation of unknown locations, making it suitable for reconstructing the three-dimensional temperature field of non-uniformly distributed measurement points. The complete temperature field distribution refers to the temperature value distribution at all spatial locations within the concrete obtained through interpolation reconstruction. The forward temperature field refers to the temperature field distribution directly calculated by a coupled finite element model of thermo-humidity-mechanical multiphysics. The temperature deviation field refers to the difference between the complete temperature field distribution and the forward temperature field at various points in space.

[0039] Kalman filter inversion correction refers to using the Kalman filter algorithm to process monitoring data with measurement noise. By minimizing the weighted sum of squares of the temperature deviation field, it inversely calculates the optimal estimates of parameters such as hydration heat source intensity and thermal conductivity coefficient in the coupled thermal-humidity-mechanical multiphysics finite element model. Hydration heat source parameters describe the exothermic characteristics of concrete hydration reaction, including hydration heat release rate, degree of hydration, and adiabatic temperature rise. Moving boundary recognition algorithm dynamically identifies and updates boundary condition changes in the coupled thermal-humidity-mechanical multiphysics finite element model based on real-time sensor monitoring data. This includes constraint release due to support removal, constraint application due to formwork installation, and third-type boundary condition updates due to changes in environmental temperature and humidity. Dynamic boundary condition changes refer to the real-time changes in boundary constraints and environmental conditions as construction progresses.

[0040] Incremental iterative algorithms refer to numerical methods that discretize the time domain into several time steps, iteratively solving the temperature, humidity, and stress fields within each time step until the field variables satisfy the coupling equilibrium condition. Adaptive time step adjustment strategies refer to control methods that dynamically adjust the time increment step size based on the convergence speed and accuracy requirements of the coupled calculation. The step size is decreased when the iteration fails to converge, and increased when the convergence speed is too fast to improve computational efficiency. Predicted shrinkage deformation fields refer to the spatial distribution predictions of concrete shrinkage deformation at future moments obtained through calculations using a coupled finite element model of thermo-humidity-mechanical multiphysics fields.

[0041] The Green's function stress response matrix refers to the stress response solution caused by a unit point load or unit shrinkage strain at various points in space in elasticity. The Green's function for the basic working condition is pre-calculated and stored in matrix form. During actual stress calculation, matrix multiplication is performed using the superposition principle. The basic shrinkage working condition refers to decomposing the complex actual shrinkage field into several simplified typical shrinkage modes, including standard working condition types such as uniform shrinkage, linear gradient shrinkage, and localized concentrated shrinkage. The stress field distribution refers to the stress tensor distribution at various points inside the concrete structure, calculated from the Green's function stress response matrix.

[0042] Critical stress values ​​refer to stress components extracted from locations prone to cracking, such as stress concentration areas, large cross-sectional changes, and intersections of multiple constraints. These components include the maximum principal stress and the maximum tensile stress. Reverse compensation calculation refers to the process of calculating the required prestress or pre-deformation during construction based on the predicted shrinkage deformation field and stress field distribution. Prestress magnitude refers to the amount of prestress applied to the structure to counteract shrinkage stress. Pre-deformation adjustment refers to the amount of reverse deformation pre-set during formwork installation to offset the effects of subsequent shrinkage deformation, ensuring the structure reaches its designed geometric dimensions after shrinkage.

[0043] The formwork adjustment command refers to the correction value for the formwork installation elevation and the adjustment parameters for the support position generated based on the prestress magnitude and pre-deformation adjustment amount, used to guide on-site construction operations. The formwork design elevation refers to the final elevation value of the concrete structure as required by the design drawings. The compensated formwork installation elevation refers to the actual formwork installation elevation obtained by adding the pre-deformation adjustment amount to the formwork design elevation. Measured shrinkage deformation data refers to the concrete shrinkage deformation values ​​actually measured by displacement sensors. Measured stress data refers to the internal stress values ​​of the concrete actually measured and calculated by strain sensors. Dynamic reverse modeling compensation refers to a closed-loop control process throughout the entire construction process, continuously collecting monitoring data, updating model parameters, predicting deformation, and adjusting construction measures.

[0044] The specific implementation methods of the above steps are described in detail below.

[0045] The specific implementation of step S01 involves first determining the sensor deployment density based on the geometric dimensions of the wind tunnel structure and the range of the concrete pouring area. Distributed fiber optic temperature sensors are then laid inside the concrete at intervals of 0.5m to 1m. Fiber Bragg grating technology or Raman scattering technology is used to achieve quasi-continuous temperature measurement along the fiber length, ensuring a sufficiently high density of measurement points for subsequent interpolation and reconstruction of the complete temperature field. Simultaneously, humidity sensor arrays are deployed on the concrete surface and at key depths within the concrete to monitor the spatiotemporal evolution of the humidity field inside the concrete. The deployment density of humidity sensors is determined based on the concrete thickness. For large-volume concrete exceeding 1m in thickness, humidity sensors are deployed at at least three depth levels: surface, middle, and bottom. Displacement sensors and strain sensors are installed at key constraint locations such as the interface between the structure and rock mass, construction joints, and areas with large cross-sectional changes. The displacement sensors utilize excitation... Optical displacement gauges or resistive displacement gauges with a measurement accuracy of no less than 0.01 mm are used. Strain sensors employ vibrating wire strain gauges or resistive strain gauges with a measurement accuracy of no less than 1 microstrain. All sensors record data in real time and transmit it to the central monitoring platform through a data acquisition system. A multi-physics field monitoring system capable of simultaneously monitoring temperature, humidity, displacement, and strain fields is constructed. Then, based on the construction zoning scheme of the wind tunnel structure, the entire structure is divided into several independent casting units. The volume of each casting unit is determined according to construction capacity and temperature control requirements. Generally, the volume of a single casting is controlled between 50 and 200 cubic meters. A topological graph describing the spatial adjacency relationship of each casting unit is established. This topological graph is represented by graph theory. Nodes in the graph represent independent casting units, and edges connect adjacent casting units with contact surfaces. The weight of the edges is set to the area of ​​the adjacent contact surfaces. This topological graph provides a mathematical basis for subsequent casting sequence optimization.

[0046] The specific implementation of step S02 involves using a graph coloring algorithm to assign pouring sequence numbers to the topology graph of adjacent relationships of pouring units. The basic principle of the graph coloring algorithm is to assign a color label to each node in the graph, such that any two adjacent nodes have different color labels. In the pouring sequence optimization, different colors represent different pouring batches. Pouring units of the same color can be constructed simultaneously, while pouring units of different colors have time intervals. This algorithm ensures that there is sufficient time interval between adjacent pouring units to reduce temperature differences. After initially assigning the pouring sequence numbers, the temperature history of each pouring unit is calculated using a concrete hydration heat model. The hydration heat model considers the evolution of cement hydration degree over time and the adiabatic temperature. The rise curve is used to calculate the initial temperature difference between adjacent pouring units at the moment of pouring new concrete. This initial temperature difference reflects the magnitude of the temperature gradient between the old and new concrete. When the initial temperature difference between any pair of adjacent pouring units exceeds 15℃, it indicates that there is a risk of temperature stress cracking in the pouring sequence. The pouring sequence number needs to be adjusted. The specific adjustment method is to increase the time interval between the two pouring units or change the pouring order. The graph coloring optimization process is re-executed. Through iterative calculation, the temperature difference between all adjacent pouring units meets the constraint condition of not exceeding 15℃. At the same time, the total number of pouring batches is minimized to improve construction efficiency. Finally, the optimal pouring scheme that meets the temperature control requirements and has the highest construction efficiency is generated.

[0047] The specific implementation of step S03 involves establishing a coupled finite element model of the wind tunnel structure, encompassing thermal, humidity, and mechanical fields. This model simultaneously considers the temperature field conduction process, humidity field diffusion process, and stress field deformation process, as well as their interactions. The temperature field control equation is based on Fourier's law of heat conduction and incorporates a concrete hydration heat source term. The humidity field control equation is based on Fick's law of diffusion and considers the influence of temperature on the diffusion coefficient. The stress field control equation is based on the elasticity equilibrium equation and couples temperature shrinkage strain and drying shrinkage strain. The model inputs the thermal and mechanical parameters of the concrete and boundary conditions. Quasi-continuous temperature data is acquired from a distributed fiber optic temperature sensor network using a multi-physics monitoring system. This data is spatially non-uniformly distributed, and the measurement points do not coincide with the finite element node positions. Three-dimensional Kriging interpolation is then used to analyze the data. The Kriging interpolation method performs spatial interpolation on the temperature data of these discrete measuring points. Based on the theory of regionalized variables, the Kriging interpolation method calculates the spatial autocorrelation of the measuring point data and constructs a variogram model to perform unbiased optimal estimation of the temperature values ​​at all node locations in the finite element model. This reconstructs a complete temperature field distribution covering all spatial locations inside the concrete. Simultaneously, a forward modeling calculation is performed using an established thermo-humidity-mechanical multiphysics coupled finite element model. After inputting initial and boundary conditions, the forward temperature field distribution is calculated. The complete temperature field distribution reconstructed by Kriging interpolation is compared point-by-point with the forward temperature field distribution in space, and the difference between the two is extracted to form a temperature deviation field. This temperature deviation field reflects the deviation between the finite element model parameters and the actual situation, providing a basis for subsequent model parameter inversion and correction.

[0048] The specific implementation of step S04 involves using the Kalman filter algorithm to inversely correct the hydration heat source parameters in the thermo-humidity-mechanical multiphysics coupled finite element model based on the temperature deviation field extracted in step S03. The Kalman filter algorithm is a recursive optimal estimation method capable of processing monitoring data with measurement noise and providing optimal parameter estimates. By minimizing the weighted sum of squares of the temperature deviation field, the correction values ​​for parameters such as hydration heat source intensity, thermal conductivity, and hydration heat release rate are calculated in reverse, making the model calculation results more consistent with the measured data. Simultaneously, real-time monitoring data is collected from displacement and strain sensors, and a moving boundary recognition algorithm is used to dynamically identify changes in boundary conditions. Based on the abrupt changes in sensor data, the algorithm identifies constraint release caused by support removal, constraint application caused by template installation, and third-type boundary condition updates caused by changes in environmental temperature and humidity. The core of the moving boundary identification algorithm is to detect the time and location of constraint changes through time series analysis of monitoring data. When a step change occurs in displacement sensor data, it is determined as constraint release; when a sudden change occurs in strain sensor data, it is determined as new constraint application. The identified dynamic boundary condition change information is promptly input into the thermo-humidity-mechanical multiphysics coupled finite element model with corrected parameters to update the boundary condition settings of the model and ensure that the model can truly reflect the actual constraint state during the construction process.

[0049] The specific implementation of step S05 involves using an incremental iterative algorithm to solve the modified thermo-humidity-mechanical multiphysics coupled finite element model. The time domain is discretized into several time steps, the initial value of which is determined based on the concrete hydration reaction rate and the temperature field change rate, typically set to 0.5 to 2 hours. Within each time step, the temperature field equation, humidity field equation, and stress field equation are solved sequentially. Due to the coupling effect between the three physical fields—temperature affects the humidity diffusion coefficient and material elastic modulus, humidity affects drying shrinkage strain, and stress state affects cracking and heat transfer boundaries—iterative solutions are required until each field variable satisfies the coupling equilibrium condition. During the iteration process, the residual values ​​of each field variable are calculated. When the iteration residual is less than... The system determines when a time step has converged and outputs the calculation results for that time step. Simultaneously, an adaptive time step adjustment strategy is employed based on the convergence speed. When the number of iterations exceeds a set threshold, indicating convergence difficulty, the time step is automatically reduced to 50% of its original size. Conversely, when the number of iterations is less than the set threshold, indicating convergence is too fast, the time step is automatically increased to 150% of its original size to improve computational efficiency. Through the aforementioned iterative calculations, the coupled response results of the temperature field, humidity field, and stress field are obtained, leading to the output of a predicted shrinkage deformation field. This predicted shrinkage deformation field provides the spatial distribution of concrete shrinkage deformation at future times. To improve the efficiency of stress field calculation, the stress response solutions caused by a unit point load or unit shrinkage strain in elasticity at various points in space are pre-calculated and stored, forming a Green's function stress response matrix. The complex actual shrinkage deformation field is decomposed into a linear combination of basic shrinkage conditions such as uniform shrinkage, linear gradient shrinkage, and localized concentrated shrinkage. The stress field distribution is quickly calculated through matrix multiplication of the Green's function stress response matrix and the shrinkage condition coefficient vector, avoiding the computational burden of repeatedly solving the elasticity equations.

[0050] The specific implementation of step S06 involves extracting stress values ​​at key locations from the stress field distribution calculated in step S05. Key locations include stress concentration areas, locations with large cross-sectional changes, and points where multiple constraints intersect, areas prone to cracking. The extracted stress components include the maximum principal stress and the maximum tensile stress. The extracted tensile stress values ​​at key locations are compared with the concrete tensile strength. Concrete tensile strength increases with age. The current tensile strength value is determined based on the concrete mix proportion and age. When the tensile stress at a key location exceeds 80% of the concrete tensile strength, it indicates a risk of cracking, and the reverse compensation calculation process is initiated. The purpose of the reverse compensation calculation is to offset the adverse effects of subsequent shrinkage through measures applied in advance during the construction phase. The required prestress is calculated, with the direction of the prestress opposite to the shrinkage stress direction, and its magnitude determined according to the stress superposition principle. To ensure that the combined stress of prestress and shrinkage stress does not exceed 50% of the tensile strength of the concrete, a pre-deformation adjustment is calculated. This pre-deformation adjustment is a reverse deformation pre-set during formwork installation to ensure that the structure reaches the designed geometric dimensions after shrinkage. The calculation of the pre-deformation adjustment considers the elastic modulus, creep coefficient, and predicted shrinkage deformation field of the concrete. Based on the deformation coordination principle, the required pre-lifting or pre-sinking amount during formwork installation is calculated. The calculated prestress and pre-deformation adjustment are converted into formwork adjustment instructions, which include correction values ​​for the formwork installation elevation and adjustment parameters for the support positions. The pre-deformation adjustment is then superimposed on the formwork design elevation specified in the design drawings to form a compensated formwork installation elevation. This installation elevation takes into account the influence of subsequent shrinkage deformation, ensuring that the final elevation of the concrete after shrinkage meets the design requirements.

[0051] The specific implementation of step S07 involves feeding back the measured shrinkage deformation data and measured stress data of the current pouring unit to the thermo-humidity-mechanical multiphysics coupled finite element model before the construction of the next pouring unit. The measured shrinkage deformation data comes from the monitoring results of displacement sensors, and the measured stress data is obtained by converting the strain values ​​measured by strain sensors into the elastic modulus. These measured data are compared with the model prediction results to identify the deviations between the material parameters and boundary condition settings in the model and the actual situation. Based on the deviation information, the model's material parameters such as the concrete elastic modulus, creep coefficient, and shrinkage coefficient are updated, while the boundary condition parameters such as support stiffness and constraint conditions are also updated. This achieves continuous optimization and self-learning of the model, updating the model parameters. After determining the number and boundary conditions, steps S03 to S06 are repeated for the next pouring unit. This involves re-collecting temperature data and reconstructing the temperature field using Kriging interpolation, extracting temperature deviations and performing Kalman filter inversion correction, solving multiphysics coupling equations and predicting shrinkage deformation, extracting stresses at key locations and performing reverse compensation calculations, and generating template adjustment instructions for the next pouring unit. Through this cyclical process, dynamic reverse modeling and compensation are achieved throughout the entire construction process. After each pouring unit is completed, the measured data is fed back to the model and the parameters are updated, so that the model's prediction accuracy continuously improves as construction progresses, and the compensation measures become more and more accurate. Ultimately, this achieves precise control of concrete shrinkage deformation and high-precision assurance of structural geometric dimensions.

[0052] It should be noted that the key technical ideas of this invention include the following aspects. The first key technical idea is to use a graph coloring algorithm to optimize the pouring sequence and iteratively adjust it in combination with temperature difference constraints. Traditional methods usually determine the pouring sequence based on construction experience, which makes it difficult to systematically control the temperature difference between adjacent pouring units. However, this invention uses graph theory to mathematically express the spatial adjacency relationship of pouring units, uses a graph coloring algorithm to automatically generate a preliminary pouring plan, and then combines hydration heat simulation calculations to verify whether the temperature difference meets the constraints. If it does not meet the constraints, it automatically adjusts and re-optimizes. This method can maximize construction efficiency while ensuring temperature control safety and avoids the blindness of traditional trial and error methods. The second key technical approach is to construct a coupled finite element model of thermo-humidity-stress multiphysics fields and use a Kalman filter algorithm to correct model parameters in real time. Traditional methods often calculate the temperature, humidity, and stress fields independently, neglecting the interaction between these physical fields, leading to insufficient prediction accuracy. The coupled model established in this invention can realistically reflect the complete physical process of temperature affecting humidity diffusion, humidity affecting drying shrinkage, and shrinkage generating stress. Simultaneously, high-density measured data is obtained through monitoring methods such as distributed fiber optic temperature measurement. The Kalman filter algorithm is used to recursively estimate the optimal model parameters, ensuring that the model prediction results continuously approach the actual situation, significantly improving the accuracy of shrinkage deformation prediction. The third key technical approach is to combine the Green's function stress response matrix with inverse compensation calculation. Traditional compensation methods can usually only qualitatively determine whether measures need to be taken, making it difficult to quantitatively calculate the compensation amount. This invention pre-calculates and stores the Green's function stress response of basic working conditions, decomposes the complex shrinkage field into a linear combination of basic working conditions, and uses the superposition principle to quickly calculate the stress distribution. Based on this, inverse compensation calculation is performed to quantitatively determine the magnitude of prestress and the amount of pre-deformation adjustment, transforming compensation measures from empirical judgment to precise calculation. The synergistic effect of the three key technical approaches mentioned above is to form a complete closed-loop control system from pouring sequence optimization, multi-physics field coupling prediction, real-time parameter correction to reverse compensation calculation. Compared with traditional methods, this invention not only improves the prediction accuracy and control effect of individual links, but more importantly, it realizes information transmission and collaborative optimization between links. Through dynamic feedback and model update mechanisms, the entire construction process is kept in a state of continuous optimization, fundamentally changing the traditional passive control mode of post-event remediation, realizing active prediction and accurate compensation, significantly reducing the risk of concrete cracking and ensuring the accuracy of structural geometric dimensions.

[0053] It should be noted that this invention also solves the following technical problem: This invention addresses the technical problem of cracking at joints caused by excessive temperature differences between adjacent pouring units during the pouring of large-volume concrete. In traditional construction, the pouring sequence is mainly determined based on construction convenience, without fully considering the impact of temperature field distribution caused by hydration heat on adjacent blocks. When there is a large temperature difference between adjacent pouring units, the constraint stress generated by the temperature gradient concentrates at the joint, increasing the risk of cracking. This invention establishes a topological graph of the adjacent relationships between pouring units, transforming the pouring sequence optimization problem into a graph coloring problem. A graph coloring algorithm is used to allocate pouring batches to each node to ensure sufficient time intervals between adjacent units. After initially allocating the pouring sequence numbers, the initial temperature difference value of adjacent units is calculated. When the temperature difference exceeds the constraint requirement of 15℃, graph coloring optimization is performed again to adjust the pouring sequence. Through iterative optimization, an optimal pouring scheme that meets the temperature difference constraint is generated, controlling the temperature field coordination of adjacent pouring units from the source, effectively reducing stress concentration and cracking risk at the joint, and ensuring the integrity of the overall structure.

[0054] This invention also solves the technical problem of low computational efficiency in shrinkage stress field calculation, which cannot support real-time decision-making at construction sites. Traditional finite element methods for calculating stress fields require mesh generation and stiffness matrix assembly, which takes too long when multiple iterations are needed to optimize compensation parameters, failing to meet the requirements of rapid response at construction sites. This invention introduces a fast solution method for the Green's function stress response matrix. It uses the Green's function of elasticity to represent the stress response caused by a unit shrinkage load, pre-calculates and stores the Green's function matrix of basic shrinkage conditions, and decomposes the actual predicted shrinkage deformation field into a linear combination of basic shrinkage conditions such as uniform shrinkage, linear gradient shrinkage, and localized concentrated shrinkage. In actual calculations, only matrix multiplication is needed to quickly obtain the stress field distribution, avoiding repetitive finite element mesh generation and stiffness matrix assembly processes. This improves computational efficiency by tens of times, enabling inverse compensation calculations to be completed in real-time at the construction site and generating template adjustment commands, providing timely computational support for dynamic construction decisions.

[0055] Specifically, the principle of this invention is as follows: The core technical problem solved by this invention lies in establishing a closed-loop feedback control mechanism from monitoring to prediction to compensation. The prediction error of traditional methods stems from the deviation between the model and actual operating conditions. This invention acquires dense temperature data through a distributed fiber optic temperature measurement network, uses Kriging interpolation to reconstruct a three-dimensional temperature field to eliminate measurement blind spots, compares the reconstructed temperature field with the model's forward modeling results to extract the deviation field, and uses a Kalman filter algorithm to reverse-correct the hydration heat source parameters to make the model approximate the true state. Simultaneously, a moving boundary recognition algorithm captures changes in support and environment to achieve dynamic updates of boundary conditions, ensuring that the model always reflects the current actual operating conditions. The thermo-humidity-mechanical multiphysics coupling model describes the interaction between various fields through a coupling coefficient matrix. The temperature field affects the humidity diffusion coefficient; temperature and humidity changes cause thermal strain and contraction strain, respectively. An incremental iterative algorithm solves the coupling equilibrium equations within each time step, ensuring accurate capture of nonlinear effects. The Green's function stress response matrix pre-stores the stress solutions for basic working conditions. Through the superposition principle, the complex contraction field is decomposed into combinations of basic working conditions, and matrix multiplication is used to quickly calculate the stress distribution, avoiding repetitive finite element solutions and improving computational efficiency. This allows inverse compensation calculations to be completed in real-time on the construction site. When the stress at a critical location exceeds a threshold, inverse compensation calculation is triggered. Based on the predicted future deformation, the pre-deformation adjustment is calculated and superimposed on the template elevation, ensuring the structure reaches the design position after contraction. After each pouring unit is constructed, measured data is fed back to the model to update parameters. The iterative process continuously corrects the model and eliminates accumulated errors. Therefore, this invention can maintain high-precision prediction and effective compensation throughout the entire construction process.

[0056] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0057] The specific implementation of step S01 is as follows: A distributed fiber optic temperature sensor network is deployed within the concrete pouring area of ​​the wind tunnel structure according to the spatial grid division principle. The spacing between fiber optic measuring points is set to 0.5m to 1m to ensure coverage of key areas inside the concrete. Simultaneously, a humidity sensor array is deployed on the structural surface and inside, with a spacing of 2m to 5m. Displacement and strain sensors are installed at key constraint locations such as the connection points between the structure and supports, and locations with large cross-sectional changes. This constructs a multi-physics field monitoring system capable of simultaneously monitoring the temperature, humidity, displacement, and strain fields. Based on the construction zoning requirements, the wind tunnel structure is divided into several independent pouring units, with each pouring unit's volume controlled to be within 50 cubic meters. Up to 200 Establish a topological graph of the adjacency relationships of casting units. In the graph, nodes represent independent casting blocks, edges connect adjacent casting blocks, and the weights of the edges are... Indicates casting unit and casting unit The contact area between adjacent surfaces, in units of .

[0058] The specific implementation of step S02 is to use a graph coloring algorithm to assign pouring sequence numbers to each node in the adjacent relationship topology graph of the pouring unit, let the first node be the first node. The timing number of each pouring unit is as follows: This indicates the pouring batch number of the pouring unit in the construction sequence, numbered [number]. The timing number of each pouring unit is as follows: This indicates the pouring batch number of the pouring unit in the construction sequence, and the adjacent pouring units. and The formula for calculating the initial temperature difference between them is as follows:

[0059] ;

[0060] In the formula, For casting unit and The normalized initial temperature difference between them; For casting unit The peak temperature of hydration heat, in °C; For casting unit The peak temperature of hydration heat, in °C; This refers to the ambient temperature, expressed in °C. For reference temperature, a value of 50℃ is used. Peak temperature of hydration heat. and Based on the concrete mix proportions and pouring time, the heat of hydration is estimated using an empirical formula, which is expressed as follows:

[0061] ;

[0062] In the formula, This is the peak temperature of hydration heat, in °C. This refers to the ambient temperature, expressed in °C. For reference temperature, the value is 50℃; The heat of hydration released per unit mass of concrete, in units of The experience value is 250 to 350. ; This refers to the density of concrete, in units of... The value is usually 2400. ; Specific heat capacity of concrete, unit: The empirical value is 0.97. ; For reference density, the value is taken as 2400. ; This is the hydration rate coefficient, in units of... The empirical value is 0.03 to 0.08. ; This refers to the time it takes to reach the peak value after pouring, measured in hours (h). When the difference exceeds 0.3, i.e., the actual temperature difference exceeds 15℃, the pouring sequence number is adjusted and the diagram coloring optimization is performed again. The optimization objective function is:

[0063] ;

[0064] In the formula, To normalize and optimize the objective function value; This refers to the number of pouring batches; This represents the total number of casting units; The penalty coefficient has an empirical value of 5. This is the set of all edges in the topological graph; Represents adjacent casting units in the topology diagram and The edge; This indicates that the normalized temperature difference exceeds the scalar value when... This item is 0 when it is less than or equal to 0.3. The optimization is repeated until the temperature difference of all adjacent pouring units meets the constraint requirements, and then the optimal pouring scheme is generated.

[0065] The specific implementation of step S03 is as follows: A coupled finite element model of the wind tunnel structure's thermo-hygroscopic multi-physics field is established. The temperature field control equation is based on Fourier's law of heat conduction, the humidity field control equation is based on Fick's law of diffusion, and the stress field control equation is based on the elasticity equilibrium equation. Quasi-continuous temperature data is acquired from a distributed fiber optic temperature sensor network. Let the... The coordinates of the measuring points are: The unit is meters (m), and the corresponding temperature measurement value is... The unit is °C. The complete temperature field distribution is reconstructed using the three-dimensional Kriging interpolation method, with interpolation points... The formula for calculating the estimated temperature at a certain location is as follows:

[0066] ;

[0067] In the formula, Spatial location The normalized temperature estimate at that location; The number of measurement points participating in the interpolation; For the first Kriging interpolation weighting coefficients for each measurement point; For the first Temperature measurements at each measuring point, in °C; This is the average temperature at all measuring points, in °C. The standard deviation of temperature at all measuring points, in °C; , , These are the spatial coordinates of the interpolation points, in meters (m). Kriging interpolation weighting coefficients. The Kriging equations are obtained by solving the system of equations, which are expressed as follows:

[0068] ;

[0069] In the formula, For the first Kriging interpolation weighting coefficients for each measurement point; For the first The first measuring point and the first Normalized semivariogram values ​​among the measurement points; For the first The first measuring point and the first The normalized spatial distance between the measurement points is calculated using the following formula: ,in , , For the first The spatial coordinates of each measuring point are in meters. For reference length, the value is 10m; For Lagrange multipliers; interpolation point To the Normalized semivariogram values ​​for each measurement point; For the interpolation point to the th The normalized spatial distance between each measuring point is calculated using the following formula: Semivariogram Using a spherical model, the calculation formula is expressed as follows:

[0070] ,when ;

[0071] ,when ;

[0072] In the formula, The value is the normalized semivariogram. Normalized spatial distance; This is the nugget value, representing the effect of measurement error and micro-variation, with an empirical value of 0.1 to 0.3; The sill value represents the magnitude of spatial variation, with an empirical value ranging from 0.5 to 1.0. The normalized range represents the scope of spatial correlation, and the calculation formula is as follows: ,in The actual range is given in meters (m), with empirical values ​​ranging from 5m to 20m. The Kriging interpolation weighting coefficients satisfy the unbiasedness condition. The reconstructed complete temperature field distribution is compared with the forward-modeled temperature field calculated by the thermo-hygroscopic multiphysics coupled finite element model. The formula for calculating the temperature deviation field is as follows:

[0073] ;

[0074] In the formula, Spatial location The normalized temperature deviation value at that location; The observed temperature field reconstructed by interpolation, in °C; This is the forward-modeled temperature field calculated using the finite element model, in °C. For reference temperature, the value is 50℃.

[0075] The specific implementation of step S04 is to perform Kalman filter inversion correction on the hydration heat source parameters in the thermo-hygroscopic multi-physics coupled finite element model based on the temperature deviation field. The calculation formula for the hydration heat source intensity correction is as follows:

[0076] ;

[0077] In the formula, This is the normalized correction for the intensity of the hydration heat source; The Kalman gain coefficient is obtained by minimizing the estimation error covariance matrix, and its empirical value is 0.6 to 0.9. This is the normalized temperature deviation value; This is the time gradient weighting coefficient, with an empirical value of 0.3; This is the normalized temperature deviation time derivative, in units of... ; For reference time, the value is 10 hours. The reference angular frequency is set to 0.1. Real-time monitoring data from displacement and strain sensors are collected, and a moving boundary recognition algorithm is used to identify changes in dynamic boundary conditions. The formula for calculating the boundary constraint state discrimination index is as follows:

[0078] ;

[0079] In the formula, This is a normalized boundary constraint state discrimination index; These are measured stress values, in MPa. The stress values ​​predicted by the model are in MPa. The reference stress value is 2 MPa. This is the displacement weighting coefficient, with an empirical value of 0.5; These are measured displacement values, in mm. The displacement values ​​are predicted for the model, in mm. The reference displacement value is 5mm. When When the threshold of 0.3 is exceeded, it is determined that the boundary conditions have changed, and the dynamic boundary condition changes are input into the corrected thermo-hygroscopic multiphysics coupled finite element model.

[0080] The specific implementation of step S05 is to solve the coupled response of the temperature field, humidity field, and stress field of the thermo-hygroscopic multiphysics coupled finite element model using an incremental iterative algorithm and a time step adaptive adjustment strategy. In this iteration, the formula for calculating the iterative residual of each field variable is as follows:

[0081] ;

[0082] In the formula, For the first The normalized residual of the next iteration; For the first The iteration of the ... Temperature values ​​at individual temperature field nodes, in °C; For the first The iteration of the ... Temperature values ​​at individual temperature field nodes, in °C; For the first The iteration of the ... Humidity values ​​at each humidity field node; For the first The iteration of the ... Humidity values ​​at each humidity field node; For the first The iteration of the ... The stress values ​​at each stress field node are expressed in MPa. For the first The iteration of the ... The stress values ​​at each stress field node are expressed in MPa. For reference temperature, the value is 50℃; For reference humidity, the value is 0.5; The reference stress is set at 2 MPa. This represents the number of nodes in the temperature field. This represents the number of nodes in the humidity field. This represents the number of nodes in the stress field. When... Less than The iteration convergence is determined and the predicted shrinkage deformation field is output. An adaptive time step adjustment strategy adjusts the time step based on the iteration convergence speed: when the number of iterations is less than 5 for three consecutive steps, the time step increases to 1.5 times the original value; when the number of iterations exceeds 20, the time step decreases to 0.5 times the original value. The Green's function stress response matrix is ​​pre-calculated and stored. Green's function stress response matrix elements Indicates the first The unit contraction strain at the location is at the... The normalized stress response caused by each location is calculated using the following formula: ,in For the first The unit contraction strain at the location is at the... The actual stress response caused at each location, in MPa, will be used to predict the shrinkage deformation field. The stress field distribution is decomposed into a linear combination of basic contraction conditions and calculated using matrix multiplication. The calculation formula is as follows:

[0083] ;

[0084] In the formula, For the first Normalized stress values ​​at each location; These are the elements of the Green's function stress response matrix; For the first The shrinkage strain value at each location, dimensionless; This represents the number of nodes in the contracted strain field.

[0085] The specific implementation of step S06 is to extract the stress values ​​at key locations in the stress field distribution, such as stress concentration areas, locations of large cross-sectional changes, and points where multiple constraints intersect. Let the maximum principal stress at these key locations be... The tensile strength of concrete is The formula for calculating the tensile stress criterion is as follows:

[0086] ;

[0087] In the formula, This is the normalized tensile stress ratio. The maximum principal stress at the critical location is expressed in MPa. This refers to the tensile strength of concrete, measured in MPa. It is determined based on the concrete strength grade, with an empirical value ranging from 1.5 to 3.5 MPa. When the value exceeds 0.8, reverse compensation calculation is initiated. The formula for calculating the pre-deformation adjustment amount is as follows:

[0088] ;

[0089] In the formula, This is the normalized pre-deformation adjustment amount; The compensation coefficient has an empirical value of 1.1 to 1.3. Dimensionless for predicting contraction strain; The length of the structural feature is expressed in meters (m). The reference length is 10m. Based on the pre-deformation adjustment amount, a template adjustment instruction is generated. The formula for calculating the compensated template installation elevation is as follows:

[0090] ;

[0091] In the formula, The adjusted template installation elevation is shown in meters (m). The elevation is specified for the template design, in meters (m). This is the normalized pre-deformation adjustment amount; For reference length, the value is 10m.

[0092] The specific implementation of step S07 is as follows: before the construction of the next casting unit, the measured shrinkage deformation data and measured stress data of the current casting unit are fed back to the thermo-hygroscopic multiphysics coupled finite element model. The material parameters are updated using a weighted average method, and the updated parameter calculation formula is expressed as follows:

[0093] ;

[0094] In the formula, For the updated material parameters; These are the original parameters of the model; These are parameters obtained by inversion from measured data; This is a weighting coefficient, with an empirical value of 0.7; various material parameters , , The dimensions depend on the type of parameter, including the elastic modulus (unit: GPa), Poisson's ratio (dimensionless), and the coefficient of thermal expansion (unit: ). The unit of thermal conductivity is The boundary conditions are updated in real time according to the actual construction progress. Steps S03 to S06 are executed cyclically to achieve dynamic reverse modeling and compensation for the entire construction process. The model is updated and parameters are corrected once after the construction of each pouring unit is completed, so as to ensure that the prediction accuracy of subsequent pouring units continues to improve.

[0095] To better understand and implement this invention, the following is a specific application scenario of the invention, Example 2: Based on the structural characteristics and construction conditions of the wind tunnel, the technical team divided the entire structure into 42 casting units, each with a volume of 350 to 500 cubic meters. The technical team deployed a distributed fiber optic temperature sensor network within each pouring unit. Employing Raman scattering technology, the fiber optic sensing system used 156 fibers, each 80m long, with a 0.8m spacing between measuring points, forming a three-dimensional temperature monitoring network covering the interior of the concrete. Simultaneously, 72 humidity sensors were installed on the concrete surface and at key depths within the concrete, measuring relative humidity from 30% to 100% with an accuracy of ±2%. Twenty-eight displacement sensors with a measurement accuracy of 0.01mm and 35 strain sensors with a measurement range of ±3000με were installed at critical constraint locations on the structure, constructing a complete multiphysics monitoring system.

[0096] The technical team established a topology diagram of the adjacency relationships of 42 casting units. The weight of each edge in the diagram was determined based on the contact area between adjacent surfaces, with the contact area ranging from 15 to 45. The graph coloring algorithm was used to optimize the pouring sequence. The initial coloring scheme allocated 42 pouring units to 8 pouring batches. According to hydration heat simulation calculations, the initial temperature difference between adjacent units in the third batch and the second batch reached 19℃, exceeding the constraint requirement of 15℃. The technical team adjusted the timing of pouring units numbered 17, 18, and 21 in the third batch, postponing them to the fourth batch. After re-optimizing the graph coloring, the temperature difference between all adjacent pouring units was controlled within 12℃, generating an optimal pouring scheme containing 9 pouring batches, with an interval of 72 to 96 hours between adjacent batches.

[0097] During the construction of the first batch of pouring units, the technical team established a coupled finite element model of the wind tunnel structure, encompassing thermal, humidity, and mechanical fields. The model used eight-node hexahedral elements, resulting in a mesh of 86,000 elements. The cement hydration heat source in the temperature field control equations was described using an exponential function, with the initial hydration heat release rate set at 8.5%. The characteristic time for the hydration reaction is 18 hours. The humidity diffusion coefficient in the humidity field governing equation varies with temperature; the baseline diffusion coefficient at 20℃ is... In the stress field governing equation, the elastic modulus of concrete develops with age, reaching 75% of the design value at 7 days. The coefficient of thermal expansion is taken as The drying shrinkage coefficient is taken as .

[0098] Forty-eight hours after pouring, quasi-continuous temperature data collected by a distributed fiber optic temperature sensor network showed that the highest internal temperature of the concrete occurred at a depth of 0.6m from the surface, with a peak temperature of 68℃, while the surface temperature was 43℃. The technical team used a three-dimensional spatial kriging interpolation method to calculate the spatial correlation variability function, setting the range parameter to 1.2m, the nugget value to 0.15, and the sill value to 0.92, thus reconstructing the complete temperature field distribution. Figure 2 As shown, the complete temperature field distribution is compared with the forward temperature field calculated by the thermo-humidity-mechanical multi-physics coupled finite element model, and the temperature deviation field is extracted. In the deviation field, the temperature deviation of 68% of the region is within ±2℃, but there are local deviations of 5 to 7℃ near the boundary of the casting unit.

[0099] Based on the temperature deviation field data, the technical team updated the model parameters using a Kalman filter inversion correction method. The hydration heat release rate was corrected to 7.8. The thermal conductivity coefficient starts from an initial value of 2.3. Revised to 2.5 The convective heat transfer coefficient is from 12 Revised to 14 Real-time monitoring data from displacement and strain sensors indicated that the bottom formwork of the first batch of cast-in-place units began to be removed at 72 hours. The technical team used a moving boundary recognition algorithm to identify the boundary condition changes caused by the release of constraints. This dynamic boundary condition change was input into the modified thermo-humidity-mechanical multiphysics coupled finite element model, and the original fixed constraint boundaries in the model were converted into free boundaries.

[0100] The technical team employed an incremental iterative algorithm to solve the coupled thermal-humidity-stress multiphysics finite element model, setting the initial time step to 1 hour. An alternating iterative solution strategy was used for the temperature, humidity, and stress fields. In the calculations for the first batch of cast units at 7 days of age, the residual for the temperature field decreased to [value missing] after 5 iterations. After 7 iterations of the humidity field, the residual decreased to After six iterations of the stress field, the residual decreased to ,satisfy The convergence criterion was determined. Based on the adaptive time step adjustment strategy, the time step was adjusted to 2 hours in subsequent calculations due to the faster convergence speed, improving computational efficiency. The predicted shrinkage deformation field output by the model shows that the predicted shrinkage deformation at the center of the top surface of the first batch of cast elements at 28 days of age is 2.8 mm, and at the edge is 1.5 mm.

[0101] The technical team pre-calculated the Green's function stress response matrix for 108 basic shrinkage conditions, including 36 uniform shrinkage conditions, 42 linear gradient shrinkage conditions, and 30 locally concentrated shrinkage conditions. The predicted shrinkage deformation field was decomposed into a linear combination of these basic conditions, and the decomposition coefficients are shown in Table 1.

[0102] Table 1. Basic Decomposition Coefficients for Predicting the Contraction Deformation Field

[0103]

[0104] The stress field distribution is calculated using the Green's function stress response matrix multiplication, and the calculation time is only 8% of that of the traditional direct finite element method. Figure 4 As shown, the stress field distribution data indicates that the planned pouring location for the second batch of units is at the interface with the already poured first batch of units, and the predicted maximum principal stress at 28 days is 2.1. The design value for the 28-day tensile strength of the concrete at this location is 2.5. The stress value at the critical location accounts for 84% of the tensile strength, exceeding the control threshold of 80%.

[0105] The technical team immediately initiated the reverse compensation calculation program. Based on the predicted shrinkage deformation field and stress field distribution, they calculated that the prestress required to be applied during the formwork installation of the second batch of pouring units was 0.5. This is achieved through prestressed steel mesh. Simultaneously, the pre-deformation adjustment was calculated: 3.2mm at the center of the top surface of the second batch of pouring units, 1.8mm at the edges, and 1.5mm in the horizontal direction of the side formwork. Figure 3 As shown, the technical team generated a template adjustment instruction, which superimposed the pre-deformation adjustment amount onto the template design elevation. The template design elevation at the center of the top surface of the second batch of pouring units was 23.500m, and the compensated template installation elevation was adjusted to 23.5032m. The edge position was adjusted from the design elevation of 23.480m to 23.4818m.

[0106] After the completion of the second batch of poured units, measured shrinkage deformation data obtained from displacement sensors showed that at 28 days, the actual shrinkage deformation at the center of the top surface was 2.6 mm, a deviation of 7% from the predicted value of 2.8 mm. Measured stress data from strain sensors indicated that the maximum principal stress at the interface was 1.8 mm. The shrinkage rate was below 72% of the tensile strength of concrete, and no risk of cracking was observed. The technical team fed these measured data back into the thermo-humidity-mechanical multiphysics coupled finite element model, updating the shrinkage coefficient in the material parameters from the initial value. Revised to Meanwhile, the environmental humidity parameter in the boundary conditions was updated, revised from the initial assumption of 65% to the measured average of 58%.

[0107] In the subsequent construction of the third to ninth batches of pouring units, the technical team cyclically executed the complete process of temperature field reconstruction, model inversion correction, coupled field solution, and inverse compensation calculation. The compensation parameters for each batch of pouring units are shown in Table 2.

[0108] Table 2 Reverse Compensation Parameters for Each Batch of Casting Units

[0109]

[0110] After the completion of all 42 pouring units, the technical team conducted a comprehensive measurement of the wind tunnel structure. The overall geometric deviation of the structure was controlled within ±2.5mm, meeting the design requirement of ±3mm. No visible cracks exceeding 0.05mm in width were found in the entire structure. Figure 5 As shown, throughout the entire construction process, the dynamic reverse modeling compensation method achieved closed-loop control from monitoring data acquisition to model parameter updates, deformation prediction, and construction measure adjustments. The parameter transfer between batches ensured the continuous improvement of compensation accuracy.

[0111] The technological advancements of this invention compared to traditional concrete shrinkage control methods are mainly reflected in three key innovations. First, traditional methods rely on empirical formulas to estimate shrinkage deformation, failing to accurately reflect the complex coupling effects of temperature, humidity, and stress fields during actual construction. This invention, however, establishes a coupled finite element model of heat, humidity, and stress fields, describing the mutual influence of hydration heat conduction, humidity diffusion, and stress evolution at the physical mechanism level, enabling accurate prediction of shrinkage behavior under different construction conditions. Second, traditional methods address existing cracks through reactive measures, essentially a passive response. This invention, through pre-calculation and rapid superposition of the Green's function stress response matrix, achieves efficient prediction of future stress states, proactively adjusting construction parameters through reverse compensation calculations before cracks occur, keeping shrinkage stress within a safe range. Third, traditional methods treat each pouring unit as an independent construction object, ignoring the mutual influence between adjacent units. This invention optimizes the pouring sequence through a graph coloring algorithm and updates the model with measured data from preceding units at each construction stage, achieving dynamic iterative optimization throughout the entire construction process. This allows subsequent pouring units to benefit from previous construction experience, with compensation accuracy continuously improving as construction progresses.

[0112] It should be noted that the variables involved in this invention are explained in detail in Tables 3 and 4.

[0113] Table 3. Variable Explanation Table (Part 1)

[0114]

[0115] Table 4. Variable Explanation Table (Part Two)

[0116]

[0117] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for reverse modeling and compensation of concrete shrinkage deformation, characterized in that, A distributed fiber optic temperature sensor network and humidity sensor array were deployed in the concrete pouring area of ​​the wind tunnel structure. Displacement sensors and strain sensors were installed at key constraint locations to construct a multi-physics monitoring system. The wind tunnel structure was divided into several pouring units according to the construction zones, and a topology diagram of the adjacent relationships of the pouring units was established. A graph coloring algorithm was used to assign pouring sequence numbers and optimize the temperature difference between adjacent pouring units. A coupled finite element model of thermal-humidity-mechanical multi-physics was established. Quasi-continuous temperature data was collected, and the complete temperature field distribution was reconstructed using the three-dimensional spatial Kriging interpolation method. The temperature deviation field was extracted, and Kalman filtering was used to invert and correct the hydration heat source parameters. A moving boundary recognition algorithm was used to identify dynamic boundary condition changes and input them into the model. An incremental iteration algorithm and a time step adaptive adjustment strategy were used to solve the coupled response and output the predicted shrinkage deformation field. The stress field distribution was calculated using the Green's function stress response matrix. When the stress value at a key location exceeds the threshold, a reverse compensation calculation was initiated to generate a template adjustment command. The pre-deformation adjustment amount was superimposed on the template design elevation to form the compensated template installation elevation. The measured data was fed back to the model and iterated cyclically.

2. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 1, characterized in that, The distributed fiber optic temperature sensor network employs fiber Bragg grating technology or Raman scattering technology, with a measurement point spacing of 0.5m to 1m.

3. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 2, characterized in that, In the topology diagram of the adjacency relationship of the casting units, nodes represent casting units, edges connect adjacent casting units, and the weight of the edge represents the contact area of ​​adjacent surfaces.

4. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 3, characterized in that, The graph coloring algorithm assigns a color to each node in the graph so that adjacent nodes have different colors, and different colors represent different pouring batches.

5. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 4, characterized in that, The step of optimizing the temperature difference between adjacent pouring units specifically involves calculating the initial temperature difference between adjacent pouring units based on the pouring sequence number. When the initial temperature difference exceeds 15℃, the pouring sequence number is adjusted and the graph coloring optimization is performed again until the temperature difference between all adjacent pouring units meets the constraint requirement of not exceeding 15℃, and then the optimal pouring scheme is generated.

6. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 5, characterized in that, In the thermo-humidity-mechanical multiphysics coupled finite element model, the temperature field control equation describes the hydration heat conduction process, the humidity field control equation describes the moisture diffusion process, and the stress field control equation describes the shrinkage deformation response. The three field equations are related through a coupling coefficient matrix.

7. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 6, characterized in that, The three-dimensional spatial kriging interpolation method is based on the theory of regionalized variables. It performs unbiased optimal estimation of unknown locations by calculating the spatial correlation and variogram of known measurement point data.

8. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 7, characterized in that, The temperature deviation field refers to the difference distribution of the temperature field at various points in space between the complete temperature field distribution and the forward modeling temperature field calculated by the coupled finite element model of thermo-humidity-mechanical multiphysics.

9. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 8, characterized in that, The Kalman filter inversion correction calculates the hydration heat source intensity and thermal conductivity coefficient in the thermo-humidity-mechanical multiphysics coupled finite element model by minimizing the weighted sum of squares of the temperature deviation field.

10. The method for reverse modeling and compensation of concrete shrinkage deformation according to claim 9, characterized in that, The moving boundary recognition algorithm dynamically identifies and updates boundary condition changes in the thermo-humidity-mechanical multiphysics coupled finite element model based on real-time sensor monitoring data, including constraint release caused by support removal and constraint application caused by template installation.

Images