Geopolymer deformation regulation method based on digital twinning

Abstract

The application discloses a geopolymer deformation regulation and control method based on digital twinning, relates to the technical field of digital twinning regulation and control, and comprises the following steps: locking a risk front according to a geopolymer deformation risk atlas, deducing three-track counterfactual twin branches of a current maintenance state, a current reaction phase candidate regulation and a predicted effective reaction phase candidate regulation under the same digital twinning state, and combining the risk front migration difference to form a geopolymer cross-phase regulation track; analyzing the geopolymer cross-phase regulation track by using a geopolymer phase discrimination model, verifying the risk inhibition effect after the candidate regulation crosses the reaction phase, eliminating the candidate regulation that is invalid and expands the deformation risk, generating a geopolymer robust regulation sequence, and generating a geopolymer time delay compensation regulation instruction in combination with an actual execution time delay. The application achieves the effect of improving the robustness of the geopolymer regulation scheme and the reliability of the deformation risk inhibition.

Description

Technical Field

[0001] This invention relates to the field of digital twin control technology, and in particular to a method for controlling geopolymer deformation based on digital twins. Background Technology

[0002] Geopolymers are inorganic cementitious materials formed by alkali activation of raw materials rich in silicon and aluminum components. They have been gradually applied to precast components and low-carbon building materials. During the curing and maintenance process, geopolymers are affected by chemical shrinkage, temperature and humidity changes, and constraint effects. The material ratio and curing conditions can also change the reaction process and deformation patterns. Digital twin technology can provide a technical basis for the prediction and control of geopolymer deformation by dynamically mapping physical components to digital space, combined with real-time monitoring to continuously update the material state and predict subsequent deformation.

[0003] Existing methods for controlling geopolymer deformation typically rely on age, temperature and humidity conditions, and local measurement points to determine the state of components. These methods are insufficient to reflect asynchronous deformation caused by differences in reaction processes in different regions, and also make it difficult to identify risk locations and expansion directions in advance. Furthermore, geopolymers continue to react between the calculation of control commands and their actual implementation, which may cause effective control schemes at the current reaction phase to fail when they actually take effect. In addition, the changes caused by the natural reaction of materials in the actual deformation can easily lead to misjudgments of the control effect. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for regulating geopolymer deformation based on digital twins to solve the problems of difficulty in early identification of geopolymer deformation risks and easy failure of cross-reaction phase regulation.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for regulating geopolymer deformation based on digital twins, comprising: collecting basic data of the geopolymer component to be regulated; establishing a digital twin space and associating reaction state with deformation state to generate a geopolymer reaction-deformation twin benchmark; using the geopolymer reaction-deformation twin benchmark to identify the material reaction phase, obtaining geopolymer reaction phase state data; reconstructing the full-field deformation and deducing the deformation risk to form a geopolymer deformation risk map; locking the risk frontier based on the geopolymer deformation risk map, and deducing, under the same digital twin state, maintaining the current maintenance state, current reaction phase candidate regulation, and pre-regulation. The three-track counterfactual twin branches of the candidate control for the effective reaction phase are combined with the risk frontier migration difference to form the geopolymer cross-phase control trajectory. The geopolymer phase discrimination model is used to analyze the geopolymer cross-phase control trajectory, verify the risk suppression effect of the candidate control after crossing the reaction phase, eliminate the candidate control with failure and increased deformation risk, generate the geopolymer robust control sequence, and generate the geopolymer time delay compensation control command based on the actual execution time delay. Maintenance control is implemented according to the geopolymer time delay compensation control command and the digital twin state is corrected according to the actual deformation feedback to obtain the geopolymer deformation closed-loop control record.

[0007] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the basic data of the geopolymer component to be controlled includes component geometric parameters, geopolymer mix proportion parameters, initial temperature and humidity state, initial mechanical state, curing boundary state, and allowable deformation boundary.

[0008] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the specific steps for generating the geopolymer reaction deformation twin reference are as follows: The entity calculation area is divided according to the basic data of the geopolymer component to be controlled and spatial discretized. A digital twin space is established. The initial temperature and humidity state and the initial mechanical state are written into the digital twin space and the temperature and humidity transfer boundary and mechanical constraint relationship are established to obtain the initial spatial state of the geopolymer. The reaction progress is calculated based on the initial spatial state of the geopolymer and the geopolymer ratio parameters. Chemical shrinkage and temperature and humidity deformation are determined according to the reaction progress, and the deformation state is obtained by combining the mechanical constraint relationship. Then, the reaction state and the deformation state are associated and the allowable deformation boundary is bound to generate a geopolymer reaction deformation twin reference.

[0009] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the specific steps for obtaining geopolymer reaction phase state data are as follows: Real-time temperature and humidity conditions and strain conditions are written into the geopolymer reaction deformation twin reference. The reference deformation value is calculated with the reaction progress fixed. The reaction-sensitive deformation is extracted based on the difference between the measured deformation value and the reference deformation value. The phase space is reconstructed in combination with the reaction progress change to form the geopolymer reaction deformation phase trajectory. The geopolymer reaction deformation phase trajectory is continuously segmented and the structural change positions of the relationship between the reaction progress state and the deformation response are identified. Combined with the state correction of adjacent regions, abnormal turning points are corrected. Then, the material reaction phase is determined according to the reaction progress and unidirectional locking is performed to obtain the geopolymer reaction phase state data.

[0010] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the specific steps for forming a geopolymer deformation risk map are as follows: Phase space constraint relationship is established based on geopolymer reaction phase state data, full-field deformation inversion is performed in combination with measured strain, and the deformation evolution of each entity's calculation region is deduced according to the material reaction phase to form the geopolymer predicted deformation trajectory. By using geopolymer-predicted deformation trajectories to determine the initial risk time and form a geopolymer risk arrival time field, the earliest risk location and expansion direction are identified, and a geopolymer deformation risk map is generated.

[0011] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the method involves locking the risk frontier based on the geopolymer deformation risk map, deducing three counterfactual twin branches—maintaining the current maintenance state, current reaction phase candidate control, and expected effective reaction phase candidate control—under the same digital twin state, and then combining the differences in risk frontier migration to form a geopolymer cross-phase control trajectory. The specific steps are as follows: Based on the geopolymer deformation risk map, the risk frontier is locked, the digital twin state at the locking time is extracted, and the reaction phase and deformation state are fixed to form the geopolymer risk frontier mother state; then, based on the geopolymer risk frontier mother state, candidate regulation is generated, and a homologous inference path is established to maintain the current maintenance state, the current reaction phase candidate regulation, and the expected effective reaction phase candidate regulation, generating a three-track counterfactual twin branch; Based on the three-track counterfactual twin branch, the material reaction phase and deformation state are advanced according to the unified prediction time sequence, and the prediction state of each branch is synchronized at the reaction phase transition position. The risk frontier position and expansion process of each branch are continuously tracked to form a three-track risk frontier sequence. Based on the three-track risk frontier sequence, and using the risk frontier migration process under the current maintenance state as a control, the changes in the risk frontier position, the order of risk arrival, and the direction of expansion under the current response phase candidate regulation and the expected effective response phase candidate regulation were compared to obtain the differences in risk frontier migration. By using the differences in risk frontier migration to correlate the response phase transition position and the timing of regulatory intervention for candidate regulation, and connecting the risk frontier changes before and after the candidate regulation crosses the response phase according to a unified prediction time series, a geopolymer cross-phase regulation trajectory is formed.

[0012] As a preferred embodiment of the geopolymer deformation regulation method based on digital twins described in this invention, the steps of analyzing the geopolymer cross-phase regulation trajectory using a geopolymer phase discrimination model, verifying the risk suppression effect of candidate regulation after crossing the reaction phase, eliminating candidate regulation with failure and increased deformation risk, and generating a robust geopolymer regulation sequence are as follows: Geopolymer phase discrimination model is used to analyze geopolymer cross-phase regulation trajectory. Based on the reaction phase transition position and regulation intervention time, the trajectory state of candidate regulation before and after crossing the reaction phase is aligned to generate geopolymer phase-aligned regulation trajectory. Based on the phase alignment control trajectory of geopolymers, cross-phase risk suppression constraints are established, and physically feasible disturbance scenarios are constructed by combining the reaction phase transition deviation and maintenance execution deviation. Then, counterfactual falsification inferences are performed on candidate controls to form the falsification trajectory of geopolymer candidate controls. The risk changes of each candidate regulation after crossing the reaction phase are verified by using the falsification trajectory of the candidate regulation of geopolymer. Candidate regulations that show risk suppression failure and risk frontier expansion are eliminated. Then, the cross-phase risk suppression fidelity value of the remaining candidate regulations is calculated to obtain the fidelity sequence of the candidate regulation of geopolymer. Based on the geopolymer candidate regulation fidelity sequence, candidate regulations are screened according to the expected effective response phase and the regulation order is determined by combining the cross-phase risk suppression fidelity value. Then, the candidate regulations of adjacent response phases are continuously simulated to verify the first risk time and risk frontier changes after regulation switching. Candidate regulations that generate new deformation risks are replaced to generate a geopolymer robust regulation sequence.

[0013] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the specific steps for generating the geopolymer time delay compensation control command are as follows: Based on the robust geopolymer regulation sequence, the actual execution delay of each regulation segment is read, and the arrival time of the regulation effect to the risk frontier is determined by combining the risk frontier location, thus obtaining the geopolymer regulation arrival time series; Based on the arrival time of geopolymer regulation, the reaction progress and material reaction phase when the regulation effect arrives are deduced using the geopolymer reaction deformation twin benchmark, forming the geopolymer regulation arrival state. Then, the command generation time of each regulation segment is corrected according to the geopolymer regulation arrival state, and regulation segments that exceed the target reaction phase are eliminated, generating geopolymer time delay compensation regulation commands.

[0014] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the specific steps for obtaining the closed-loop control record of geopolymer deformation are as follows: Based on the geopolymer time delay compensation control command, maintenance control is implemented and temperature and humidity status and measured deformation are collected. The actual execution time is associated to form geopolymer control execution feedback. The actual maintenance control and deformation evolution under the original maintenance state are then replayed and associated with the measured deformation to form a geopolymer dual mirror feedback trajectory. Based on the comparison of the measured deformation with the actual curing-controlled mirror deformation and the mirror deformation maintaining the original curing state according to the dual mirror feedback trajectory of the geopolymer, the control effect of the actual deformation change is determined. The digital twin state is corrected by combining the material reaction phase and the risk frontier position, and the updated deformation risk is deduced. The curing control and digital twin correction process is recorded according to the control time sequence to obtain the closed-loop control record of geopolymer deformation.

[0015] As a preferred embodiment of the geopolymer deformation control method based on digital twins described in this invention, the geopolymer phase discrimination model includes: The phase event coding layer receives the cross-phase regulation trajectory of the geopolymer, extracts the risk frontier state before and after the candidate regulation intervention according to the reaction phase transition position, and converts the reaction phase transition position, regulation intervention time and risk frontier migration state into continuous phase event codes; The cross-phase state mapping layer establishes a state mapping relationship between adjacent response phases based on the encoding of continuous phase events, maps the risk suppression state in the current response phase to the expected effective response phase, and obtains the same-scale state representation of the candidate regulation before and after crossing the response phase. The falsification constraint discrimination layer maps the differences between the same-scale state representation and the risk evolution under the current maintenance state, and uses the reaction phase transition bias and maintenance execution bias to limit the acceptable state range, and identifies the failure state that causes the first risk time to advance and the risk frontier to expand.

[0016] The cross-phase fidelity analysis layer analyzes the degree of retention of the risk suppression effect of the current response phase after it is transmitted to the expected effective response phase, based on the location and duration of the failure state in the cross-phase trajectory of the candidate regulation, and obtains the cross-phase risk suppression fidelity value of each candidate regulation. The regulation succession discrimination layer establishes the candidate regulation succession relationship between adjacent response phases based on the cross-phase risk suppression fidelity value, determines whether the end state of regulation of the previous response phase can enter the effective starting state of candidate regulation of the next response phase, and outputs the robust discrimination result of candidate regulation.

[0017] The beneficial effects of this invention are as follows: By constructing a three-track counterfactual twin branch and comparing the differences in risk frontier migration, a geopolymer cross-phase regulation trajectory is formed, realizing continuous comparison and cross-phase tracking of the risk suppression effect of candidate regulation under different reaction phases, thereby improving the accuracy of geopolymer deformation risk prediction and the adaptability of cross-phase regulation; by performing falsification analysis on the geopolymer cross-phase regulation trajectory and screening candidate regulation through a geopolymer phase discrimination model, the continuous effectiveness and anti-disturbance ability of candidate regulation after crossing the reaction phase are jointly verified, which can eliminate candidate regulation that fails to suppress risk and generates new deformation risks, thereby improving the robustness of geopolymer regulation schemes and the reliability of deformation risk suppression. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for regulating geopolymer deformation based on digital twins.

[0020] Figure 2 A flowchart for generating a geopolymer deformation risk map.

[0021] Figure 3 A flowchart for forming the cross-phase modulation trajectory of geomers.

[0022] Figure 4 A flowchart for generating geopolymer time delay compensation control commands. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for regulating geopolymer deformation based on digital twins, comprising the following steps: S1. Collect basic data of the geopolymer components to be regulated, establish a digital twin space and associate the reaction state with the deformation state to generate a geopolymer reaction-deformation twin benchmark.

[0027] Based on the component design drawings and combined with 3D scanning measurements, the component's geometric parameters, such as size, shape, and spatial position, are obtained. The amount and proportion of each raw material are obtained from the raw material weighing records and mix proportion records to form the geopolymer mix proportion parameters. The initial temperature and humidity conditions are collected by temperature and humidity sensors installed inside and on the surface of the component. The initial strain, initial displacement, and stress state of the component are collected by strain gauges, displacement sensors, and load detection devices to form the initial mechanical state. The curing boundary state is determined by collecting the curing environment temperature, humidity, and curing equipment operating parameters. The allowable deformation boundary is determined according to the component design requirements, material performance indicators, and allowable deformation control standards to form the basic data of the geopolymer component to be controlled.

[0028] Based on the geometric parameters of the geopolymer components in the basic data of the components to be regulated, the outer contour, internal structural boundary and geometric dimension change positions of the geopolymer components to be regulated are determined. According to the spatial continuity of the components, the geopolymer components to be regulated are divided into multiple solid computational regions. The finite element mesh generation method is used to spatially discretize each solid computational region, and each solid computational region is divided into interconnected finite elements. The spatial position and connection relationship of each finite element are determined according to the geometric parameters of the components. Then, the finite elements are organized according to the actual spatial position of the geopolymer components to be regulated to establish a digital twin space.

[0029] Based on the spatial location of each finite element in the digital twin space, the initial temperature and humidity state is written into the digital twin space according to the measurement locations of temperature and humidity sensors. The initial temperature and humidity state at locations without sensor placement is determined using spatial interpolation. The initial mechanical state is written into the digital twin space according to the measurement locations of strain gauges, displacement sensors, and load detection devices. The temperature and humidity transfer conditions between the component surface and the curing environment are determined in conjunction with the curing boundary state, forming a temperature and humidity transfer boundary. Based on the initial strain, initial displacement, and stress state, the deformation constraints and force transfer relationships between the calculation regions of each entity are determined, forming mechanical constraint relationships, and the initial spatial state of the geopolymer is obtained.

[0030] The initial temperature and humidity states of each finite element are read sequentially based on the initial spatial state of the geopolymer. The reaction rate level at the start of the calculation for each finite element is determined according to the geopolymer mix ratio. At each calculation moment, the reaction activity under temperature conditions is determined based on the current absolute temperature, the reaction retention under humidity conditions is determined based on the current normalized relative humidity, and the limitation of the remaining unreacted components on subsequent reactions is determined based on the degree of reaction already occurred. The reaction rate at the current calculation moment is comprehensively determined, and the process is continuously advanced and accumulated according to adjacent calculation moments to determine the reaction progress of each finite element. The reaction progress is then used to query the pre-tested calibrated relationship between reaction progress and chemical shrinkage for geopolymers with the same mix ratio to determine chemical shrinkage. Temperature and humidity deformation is determined based on the changes in current temperature relative to the initial temperature and current humidity relative to the initial humidity, combined with the pre-tested calibrated temperature and humidity deformation relationships. Chemical shrinkage and temperature and humidity deformation are used as the free deformation conditions for each finite element. The displacement and strain of each finite element are solved using the finite element method according to the established mechanical constraint relationships to obtain the deformation state.

[0031] Based on the spatial location of each finite element in the digital twin space, the reaction state and deformation state formed by the reaction progress determined by the same finite element at the same calculation time are associated, and the change process of the reaction state and deformation state is continuously saved according to the calculation time. Based on the allowable deformation boundary in the basic data of the geopolymer component to be controlled, the allowable deformation range applicable to each finite element is determined according to the spatial location of each entity's calculation region. The deformation state of each finite element is bound to the allowable deformation boundary of the calculation region of the entity where it is located, so that the reaction state, deformation state and allowable deformation boundary at each calculation time can be queried and compared in the digital twin space according to the same spatial location, generating a geopolymer reaction deformation twin benchmark.

[0032] It should be noted that the geopolymer reaction deformation twin benchmark is used to unify the reaction progress, deformation state and allowable deformation boundary in the associated digital twin space, realize the synchronous characterization of geopolymer reaction evolution and deformation response, and provide a unified calculation basis for material reaction phase identification, full-field deformation reconstruction and deformation risk inference.

[0033] The expression for calculating the reaction extent is: ; in, Indicates the first The entity computing region in the first The reaction progress at each calculation time point has a value range of 0 to 1; Indicates the number of the entity's computational region; The number representing the calculation time; Indicates the first The entity computing region in the first The reaction progress at each calculation time point; Indicates according to the first The reaction frequency factor is determined by the geopolymer ratio parameters of each entity's computational region; Represents the natural exponential function; Indicates the first The apparent activation energy of the geopolymer reaction in each entity calculation region; Represents the gas constant; Indicates the first The entity computing region in the first The absolute temperature at each calculation moment; Indicates the first The entity computing region in the first The normalized relative humidity at each calculation time, with a value ranging from 0 to 1; The reaction order parameter indicates the degree to which the remaining unreacted component affects the reaction rate; This represents the calculation time interval between two adjacent calculation moments; A correction index indicating the degree of influence of humidity on the reaction rate of geopolymers.

[0034] It should be noted that this expression achieves dimensional consistency by uniformly converting all parts involved in the calculation into dimensionless form. The reaction progress itself is a dimensionless quantity, the energy term inside the exponential function becomes dimensionless after dividing by terms of the same dimension, the reaction frequency term becomes dimensionless after multiplying by the time interval, and the remaining correction terms also adopt dimensionless form. Therefore, the multiplication of each part yields a dimensionless increment of the reaction progress, which can be directly added to the original reaction progress, ensuring that the calculated reaction progress remains consistent in dimension.

[0035] S2. Use the geopolymer reaction deformation twin benchmark to identify the material reaction phase, obtain geopolymer reaction phase state data, reconstruct the full-field deformation and deduce the deformation risk to form a geopolymer deformation risk map.

[0036] Based on the measurement positions of temperature sensors, humidity sensors, and strain gauges in the digital twin space, the real-time temperature and humidity states and strain states collected at each calculation moment are written into the geopolymer reaction deformation twin reference. Spatial interpolation is used to determine the real-time temperature and humidity states and strain states at locations without sensor placement. The reaction progress of each finite element at the current calculation moment is read and fixed. Chemical shrinkage is determined based on the fixed reaction progress. Temperature deformation is determined based on the change in real-time temperature relative to the initial temperature. Humidity deformation is determined based on the change in real-time normalized relative humidity relative to the initial normalized relative humidity. Constraint deformation is then determined in conjunction with mechanical constraint relationships, and the reference deformation value is determined based on the characteristic length of each finite element. The measured deformation is determined based on the strain state and the characteristic length of each finite element. The deformation value is defined as the difference between the measured deformation value and the reference deformation value as the reaction-sensitive deformation. For the same finite element, the reaction progress is compared at adjacent calculation times to determine the change in reaction progress at each calculation time. The change in reaction progress is used to determine the reaction advancement position in phase space, and the reaction-sensitive deformation at the same calculation time is used to determine the deformation response position in phase space. The change in reaction progress and the reaction-sensitive deformation at each calculation time are mapped together to a phase point. Then, the continuous phase points are connected sequentially according to the calculation time. The movement of the phase point along the reaction advancement direction reflects the change in reaction progress, and the movement of the phase point along the deformation response direction reflects the change in reaction-sensitive deformation, forming a geopolymer reaction deformation phase trajectory that can characterize the relationship between the reaction advancement state and the continuous change in deformation response.

[0037] The expression for calculating the datum deformation value is: ; in, Indicates the first Each entity computing region at time... The reference deformation value; Indicates the baseline status identifier; Indicates the first The characteristic length of the computational region for each entity; Indicates the first Chemical shrinkage coefficient of each entity's computational region; Indicates chemical shrinkage; Indicates the first The fixed reaction progress of each entity's computational region when calculating the baseline deformation value; Indicates the first Temperature deformation coefficient of each entity's computational region; Indicates temperature; Indicates the first Each entity computing region at time... Real-time temperature; Indicates the first The initial temperature of each entity's computational region; Indicates the first Humidity deformation coefficient of each entity's calculation area; Indicates humidity; Indicates the first Each entity computing region at time... Real-time normalized relative humidity; Indicates the first Initial normalized relative humidity for each entity's computational region; Indicates the first Each entity computing region at time... Constraint deformation strain determined based on mechanical constraint relationships; Indicates the effect of mechanical constraints.

[0038] It should be noted that before the calculation, the dimensions of each parameter were processed by unit normalization and strain unification. The temperature was uniformly converted to the Kelvin scale, the relative humidity was normalized to the range of 0 to 1, and the chemical shrinkage coefficient, temperature deformation coefficient, humidity deformation coefficient, and constraint deformation strain were determined according to the same benchmark. All items in parentheses were uniformly expressed as dimensionless strain. Then, the dimensionless strain was multiplied by the characteristic length in millimeters, thereby uniformly converting the calculation result into the benchmark deformation value in millimeters, ensuring the dimensional consistency of each item in the expression during the calculation process.

[0039] Furthermore, This value was obtained through a sealing shrinkage test on geopolymer specimens with the same proportion, combined with a fitting of the reaction progress. An exemplary value could be 4.0 × 10⁻⁶. -4 ~8.0×10 -4 The specific values ​​are determined based on the geopolymer ratio and test calibration results; The value was obtained by fitting the thermal expansion test results of geopolymer specimens with the same proportion; an exemplary value of 3×10⁻⁶ was used. -6 ~13×10 - 6 K -1 ; It was obtained by fitting a graded humidity shrinkage test under constant temperature conditions, with an example value of 1×10. -3 ~1×10 -2 .

[0040] The continuous phase points in the geopolymer reaction deformation phase trajectory are arranged according to the calculation time sequence. A piecewise linear fitting method is used to examine the relationship between the change in reaction progress and the reaction-sensitive deformation point by point along the geopolymer reaction deformation phase trajectory. Continuous phase points before and after the candidate segment position are fitted, and the slope of change and fitting deviation before and after the segment are compared. When the slope of change before and after the candidate segment position maintains the same trend across multiple consecutive calculation times, and the fitting deviation of the continuous phase points after the segmentation decreases by a set proportion compared to before the segmentation, the candidate segment position is determined as the structural change position in the relationship between the reaction advancement state and the deformation response. Based on the connection relationship of finite elements in the digital twin space, the change state of the geopolymer reaction deformation phase trajectory of the finite element where the structural change position is located is compared with that of adjacent finite elements at similar calculation times. When a single finite element exhibits... When a short-term transition occurs and adjacent finite elements do not exhibit the same change, and subsequent phase points resume their original direction of change, the short-term transition is identified as an abnormal transition and merged. When adjacent finite elements exhibit the same direction of change at similar calculation times, the corresponding structural change position is retained. The corrected structural change positions are arranged in ascending order of reaction progress. The material reaction phase is divided based on the reaction progress range corresponding to adjacent structural change positions. The current reaction progress of each finite element is compared with the corresponding reaction progress range to determine the current material reaction phase of each finite element. Unidirectional locking is performed according to the relationship that the reaction progress can only increase. When a short-term monitoring fluctuation points to the previous material reaction phase, the locked material reaction phase is maintained. When the reaction progress enters the reaction progress range corresponding to the next material reaction phase, the material reaction phase is updated to obtain the geopolymer reaction phase state data.

[0041] It should be noted that the geopolymer reaction phase state data is used to characterize the current reaction stage of each finite element, including the material reaction phase, corresponding reaction progress range, and unidirectional locking state of each finite element.

[0042] Using geopolymer reaction phase state data, the material reaction phase, reaction progress range, and unidirectional locking state of each finite element are read. Phase spatial constraints are then established by combining the spatial positions and connections of the finite elements in the digital twin space. This ensures that spatially connected finite elements with the same material reaction phase maintain deformation continuity, and at the boundaries between different material reaction phases, deformation transmission is restricted according to the corresponding deformation states in the geopolymer reaction deformation twin datum. The measured strain is written into the corresponding finite element according to the strain gauge measurement positions. The displacement and strain at locations without strain gauges are continuously adjusted using the finite element inversion method to ensure that the calculated strain matches the measured strain. To maintain consistency and satisfy both phase space constraints and mechanical constraints, the full-field deformation state at the current moment is obtained. Based on the full-field deformation state, the reaction progress is advanced hourly according to the material reaction phase and reaction progress range of each finite element. When entering the next material reaction phase, the material reaction phase is updated according to the unidirectional locking state. Chemical shrinkage is determined according to the updated reaction progress, and temperature and humidity deformation is determined according to the temperature and humidity changes. Then, combined with the mechanical constraints, the displacement and strain of each entity's calculation region at subsequent moments are continuously obtained using the finite element method. The deformation states at each calculation moment are connected in chronological order to form the geopolymer predicted deformation trajectory.

[0043] The predicted deformation trajectories of geopolymers are used to read the predicted deformation states of each finite element one by one according to the calculation time. The predicted deformation state of each finite element at each calculation time is compared with the allowable deformation boundary bound in the geopolymer reaction deformation twin datum. When the predicted deformation state first reaches the allowable deformation boundary, the corresponding calculation time is determined as the first risk time of the corresponding finite element. When two adjacent calculation times are located on both sides of the allowable deformation boundary, a linear interpolation method is used to determine the specific time when the predicted deformation state reaches the allowable deformation boundary based on the predicted deformation states of the two adjacent calculation times, and the specific time is determined as the first risk time. All the first risk times are arranged according to the spatial position of each finite element in the digital twin space, so that each finite element in the digital twin space is associated with the corresponding first risk time, forming a geopolymer risk arrival time field.

[0044] Based on the arrival time field of geopolymer risks, the initial risk time of each finite element is compared, and the spatial location of the finite element with the earliest initial risk time is determined as the earliest risk location. Then, according to the connection relationship of finite elements in the digital twin space, the initial risk time is compared step by step from the earliest risk location to adjacent finite elements. The spatial connection direction of the first risk time arranged continuously from early to late is determined as the risk expansion direction. The changes in the initial risk time of subsequent finite elements are continued to be tracked along the risk expansion direction, so that the earliest risk location, the initial risk time, and the risk expansion direction maintain spatial correlation in the digital twin space, and a geopolymer deformation risk map is generated.

[0045] It should be noted that the geopolymer deformation risk map is used to uniformly characterize the initial risk time, earliest risk location, and risk expansion direction of each finite element, thereby realizing the spatiotemporal positioning of geopolymer deformation risk in the digital twin space and providing a basis for risk frontier locking and candidate control simulation.

[0046] S3. Based on the geopolymer deformation risk map, the risk frontier is identified. Under the same digital twin state, the three-track counterfactual twin branches of maintaining the current maintenance state, current reaction phase candidate regulation, and expected effective reaction phase candidate regulation are deduced. Then, the cross-phase regulation trajectory of geopolymer is formed by combining the risk frontier migration difference. The cross-phase regulation trajectory of geopolymer is analyzed by using the geopolymer phase discrimination model to verify the risk suppression effect of candidate regulation after crossing the reaction phase. Candidate regulation with failure and increased deformation risk is eliminated to generate a geopolymer robust regulation sequence. Geopolymer time delay compensation regulation command is generated by combining the actual execution time delay.

[0047] Based on the geopolymer deformation risk map, each finite element is examined from earliest to latest according to the first risk time. The earliest first risk time is determined as the locking time. At the locking time, the predicted deformation state and allowable deformation boundary of each finite element are read. The spatial boundary between the finite element whose predicted deformation state has reached the allowable deformation boundary and the adjacent finite element whose predicted deformation state has not yet reached the allowable deformation boundary is determined as the risk frontier. Then, the spatial range of the risk frontier is continuously determined by combining the earliest risk position and risk expansion direction in the geopolymer deformation risk map. According to the locking time, the reaction progress, material reaction phase, deformation state, temperature and humidity state and curing boundary state are read from the geopolymer reaction deformation twin benchmark. The material reaction phase and deformation state at the locking time are kept unchanged and used as the common starting state for the deduction, forming the geopolymer risk frontier mother state.

[0048] The maintenance boundary state is read using the geopolymer risk frontier master state. Candidate controls are formed based on the adjustable range of maintenance environment temperature, humidity, and maintenance equipment operating parameters. Candidate controls for the current reaction phase are determined according to the current material reaction phase in the geopolymer risk frontier master state. The material reaction phase that the candidate controls will enter when they take effect is determined along the predicted deformation trajectory of the geopolymer based on the current reaction progress, the reaction progress range of the material reaction phase, and the intervention time of the candidate controls. The candidate controls for the expected effective reaction phase are also determined. The geopolymer risk frontier master state is replicated into three completely identical inference starting states. The first inference path keeps the maintenance boundary state at the lock-in time unchanged. The second inference path applies the candidate controls for the current reaction phase at the lock-in time. The third inference path applies the candidate controls for the expected effective reaction phase according to the intervention time of the candidate controls. This ensures that all three inference paths, except for the maintenance control state, start from the same reaction progress, material reaction phase, and deformation state, generating a three-track counterfactual twin branch.

[0049] For example, this invention can be applied to a geopolymer precast wall panel undergoing curing. After the geopolymer precast wall panel enters the curing stage, the edge area is affected by the ambient temperature and humidity more quickly, while the central area experiences slower moisture migration. The reaction progress and deformation development at different locations are not consistent. According to the geopolymer deformation risk map, several finite elements near the edge of the wall panel will reach the allowable deformation boundary first during the subsequent curing process. For example, it is predicted that the risk of exceeding the limit will first appear in the twelfth hour, while the adjacent finite elements are expected to reach the allowable deformation boundary after the thirteenth hour. At this time, the twelfth hour is determined as the locking moment, and the spatial boundary between the area that has reached the allowable deformation boundary and the area that has not yet reached the allowable deformation boundary is determined as the risk frontier. Subsequently, the reaction progress, material reaction phase, deformation state, temperature and humidity state, and curing boundary state of each finite element in the twelfth hour are read. The material reaction phase and deformation state in the twelfth hour are fixed, which is equivalent to taking a complete digital state "snapshot" of the geopolymer precast wall panel when the risk just begins to appear, forming the geopolymer risk frontier master state.

[0050] Three simulations are conducted simultaneously based on the same geopolymer risk frontier state. The first simulation maintains the current curing state, for example, continuing to use the current curing temperature and humidity, to determine how the risk frontier will expand without regulation. The second simulation forms candidate regulation for the current material reaction phase, for example, if the current material reaction phase is more sensitive to water loss, then the humidity is increased and the curing temperature is adjusted appropriately to observe whether the risk frontier can be delayed. The third simulation considers the time difference between the issuance of candidate regulation and its actual effect on the geopolymer precast wall panel, and forms candidate regulation for the expected effective reaction phase based on the expected material reaction phase when the candidate regulation actually takes effect, for example, if the material has already entered the next material reaction phase when the expected regulation takes effect, then the simulation is conducted using curing temperature and humidity conditions suitable for the next material reaction phase. All three scenarios start from the same reaction progress, material reaction phase, and deformation state at the twelfth hour, only changing the curing regulation state. Therefore, the differences in the time, spatial location, and expansion direction of the subsequent risk frontier can directly reflect the actual impact of different candidate regulation. This is the specific meaning of the three-track counterfactual twin branch in the actual curing process.

[0051] Based on a three-track counterfactual twin branch, the prediction proceeds using the same calculation start time, calculation time interval, and prediction termination time. At each calculation time, the temperature and humidity states under the current maintenance state, current candidate reaction phase control, and expected effective candidate reaction phase control are read. The reaction progress of each finite element is updated according to the reaction progress calculation process. Chemical shrinkage is determined based on the updated reaction progress, and temperature and humidity deformation is determined based on the changes in temperature and humidity states. Then, the deformation state of each finite element is updated using the finite element method in conjunction with mechanical constraint relationships. The updated reaction progress is compared one by one with the reaction progress range in the geopolymer reaction phase state data. When the reaction progress moves from the reaction progress range corresponding to the current material reaction phase to the reaction progress range corresponding to the next material reaction phase, the corresponding reaction progress boundary is determined as the reaction phase transition position. The values ​​of the calculation times before and after the reaction phase transition position are used to determine the reaction phase transition position. The predicted state determines the reaction progress, material reaction phase, and deformation state when each branch reaches the same reaction phase transition position, enabling the three-track counterfactual twin branches to form directly comparable predicted states at the same reaction phase transition position. At each calculation moment, the deformation state of each finite element in each branch is compared with the allowable deformation boundary. The spatial boundary between the finite element that has reached the allowable deformation boundary and the adjacent finite element that has not yet reached the allowable deformation boundary is determined as the risk frontier of the corresponding branch. The position change of the risk frontier is continuously recorded according to the connection relationship of the finite elements in the digital twin space, and the expansion process of the risk frontier is determined based on the spatial movement of the risk frontier in adjacent calculation moments. Then, the risk frontier position and expansion process corresponding to the current maintenance state, the current reaction phase candidate control, and the expected effective reaction phase candidate control are arranged according to a unified prediction time sequence to form a three-track risk frontier sequence.

[0052] Based on the three-track risk frontier sequence, the risk frontier position under the current maintenance state is read at each calculation time according to the unified prediction time series. The migration process of the risk frontier under the current maintenance state is used as a reference. At the same calculation time and the same reaction phase transition position, the risk frontier positions under the current reaction phase candidate control and the expected effective reaction phase candidate control are read. The spatial distance comparison method is used to determine the forward, backward and positional shift of the risk frontier relative to the current maintenance state. Then, the calculation time of the same finite element entering the risk frontier is compared to determine the degree of advance and delay of risk arrival relative to the current maintenance state. According to the connection relationship of finite elements in the digital twin space, the spatial movement direction of the risk frontier at adjacent calculation times is compared to determine whether the risk frontier expansion direction under the current reaction phase candidate control and the expected effective reaction phase candidate control is maintained, deflected or reversed relative to the current maintenance state. The risk frontier position change, risk arrival order and expansion direction are continuously correlated according to the unified prediction time series to obtain the risk frontier migration difference.

[0053] For example, during the curing of precast geopolymer wall panels, starting from the left edge of the wall panel and maintaining the current curing state, the risk frontier advances to a position 120 mm from the left edge at the 14th hour. After applying the current reaction phase candidate control, the risk frontier only advances to 85 mm at the 14th hour, and the risk arrival time for the same finite element is delayed from the 14th hour to the 15.2th hour. After applying the expected effective reaction phase candidate control, the risk frontier only advances to 60 mm at the 14th hour, and the risk arrival time for the same finite element is delayed to the 16.1th hour, changing from expanding towards the center of the wall panel to expanding along the edge of the wall panel. By comparing the risk frontier positions at 120 mm, 85 mm, and 60 mm, as well as the risk arrival time and expansion direction at the 14th, 15.2, and 16.1 hours, the differences in risk frontier migration can be obtained.

[0054] By utilizing the differences in risk frontier migration, the changes in the risk frontier position, the order of risk arrival, and the direction of expansion corresponding to each candidate regulation are determined. The intervention time of the candidate regulation is taken as the starting position of the risk frontier change process, and the intervention time of the candidate regulation is time-corresponded to the reaction phase transition position. The changes in the risk frontier position, the order of risk arrival, and the direction of expansion before and after the reaction phase transition position are organized according to the unified prediction time series. The risk frontier change state before the reaction phase transition position is continuously connected to the risk frontier change state after the reaction phase transition position, and the risk frontier change process under the action of candidate regulation is continuously connected along the calculation time. This ensures that the risk change of candidate regulation from intervention in the current material reaction phase, across the reaction phase transition position, to entering the next material reaction phase remains continuous, forming a geopolymer cross-phase regulation trajectory.

[0055] It should be noted that the geopolymer cross-phase regulation trajectory refers to the regulation evolution trajectory formed by continuously linking the changes in the risk frontier position, the order of risk arrival, and the direction of expansion under different material reaction phases according to the reaction phase transition process based on the candidate regulation process. It is used to characterize the risk suppression changes after the candidate regulation crosses the reaction phase.

[0056] When analyzing the cross-phase regulation trajectory of geopolymers using the geopolymer phase discrimination model, the phase event encoding layer in the model uses the intervention time and the reaction phase transition position as two key positioning points to divide the cross-phase regulation trajectory into continuous trajectory segments before intervention, after intervention to the reaction phase transition position, and after the reaction phase transition position. The changes in the risk frontier position, the order of risk arrival, and the direction of expansion in each continuous trajectory segment are continuously encoded according to the corresponding material reaction phase. This transforms the changes in the risk frontier, which were originally at different calculation times and different material reaction phases, into continuous phase event codes that can be compared according to the reaction progression process. The cross-phase state mapping layer in the geopolymer phase discrimination model uses the reaction phase transition position as a unified phase benchmark to map the cross-phase state of the geopolymers within the current material reaction phase that is at the same distance from the reaction phase transition position. A mapping relationship is established between the risk frontier state of the expected response phase and the risk frontier state of the response phase interval at the same distance from the response phase transition position. This allows the risk frontier state of candidate controls before and after crossing the response phase to no longer be directly compared according to absolute time, but rather compared on the same scale according to the control intervention position, response phase transition position, and response progress. This eliminates the time misalignment caused by different candidate controls due to different response speeds, and preserves the risk suppression changes of candidate controls before and after the response phase transition. Finally, the mapped trajectory states are continuously arranged according to a unified phase progression order to generate a geopolymer phase-aligned control trajectory. The geopolymer phase discrimination model realizes the transformation from "comparison by time" to "comparison by response phase progress", providing a unified and comparable trajectory basis for judging whether candidate controls still maintain risk suppression after crossing the response phase.

[0057] The process of constructing the geopolymer phase discrimination model is as follows: The phase event coding layer receives the geopolymer cross-phase regulation trajectory, locates the intervention time and reaction phase transition position within the trajectory, and divides the risk frontier change process corresponding to the candidate regulation into three continuous segments: before regulation intervention, after regulation intervention to the reaction phase transition position, and after the reaction phase transition position. Following a unified prediction time series, it calculates the risk frontier position changes, risk arrival sequence, and expansion direction in each continuous segment at each calculation time. Simultaneously, it records the material reaction phase and reaction progress at the corresponding calculation time and re-calculates the data according to the chronological relationship between the intervention time and the reaction phase transition position. The new arrangement of risk frontier states enables the risk frontier states at different calculation times to correspond continuously according to the reaction progression sequence. The material reaction phase, reaction progress, temporal position of distance control intervention time, reaction progression position of distance reaction phase transition position, risk frontier position change, risk arrival sequence and expansion direction of each calculation time are encoded in a fixed order. The encoding of the last calculation time before the reaction phase transition position is continuously connected with the encoding of the first calculation time after the reaction phase transition position. The risk frontier migration process from the intervention of the current material reaction phase to the entry into the next material reaction phase is converted into a continuous phase event encoding.

[0058] The cross-phase state mapping layer, based on continuous phase event codes, uses the reaction phase transition position as a common reference for adjacent reaction phases. It obtains the remaining reaction progress of each continuous phase event code in the current reaction phase from the reaction phase transition position and the advanced reaction progress of each continuous phase event code in the expected effective reaction phase from the reaction phase transition position. It then establishes a state mapping relationship for continuous phase event codes with consistent reaction progress distances, from near to far. When there are no continuous phase event codes with completely consistent reaction progress distances, a linear interpolation method is used to determine the corresponding state based on adjacent continuous phase event codes. The changes in the risk frontier position, the order of risk arrival, and the direction of expansion in the current reaction phase are mapped to the expected effective reaction phase according to the state mapping relationship. The layer compares the degree of backward shift of the risk frontier relative to maintaining the current maintenance state, the degree of delay in risk arrival, and the changes in the direction of expansion to determine the maintenance status of the risk suppression state in the current reaction phase in the expected effective reaction phase. This ensures that the risk suppression state before and after crossing the reaction phase is uniformly represented according to the reaction progress from the reaction phase transition position, resulting in a state representation of the candidate regulation before and after crossing the reaction phase at the same scale.

[0059] It should be noted that the same-scale state characterization is a state characterization formed by unifying the risk suppression state of candidate regulation before and after crossing the reaction phase to the same reaction progress scale, including changes in the position of the risk frontier, the order of risk arrival, the direction of expansion, and the corresponding reaction progress position.

[0060] The falsification constraint discrimination layer maps the state representation at the same scale to the risk evolution under the current maintenance state, corresponding item by item. At the same reaction progress position, it compares the risk frontier position, risk arrival time, and expansion direction. It identifies the risk suppression manifestations caused by the candidate control relative to the current maintenance state, such as the risk frontier shifting backward, risk arrival being delayed, and expansion direction changing. It identifies the risk deterioration manifestations, such as the risk frontier shifting forward, risk arrival being earlier, and expansion range being increased. This forms a difference mapping relationship between the candidate control before and after crossing the reaction phase. It adjusts the reaction phase transition position using the reaction phase transition deviation and adjusts the actual time of action of the candidate control and the maintenance state using the maintenance execution deviation. Within the allowable range of variation of the reaction phase transition deviation and the maintenance execution deviation, it redetermines the risk frontier state corresponding to the same reaction progress position. The state range that can still keep the risk frontier from shifting forward and the first risk time from being earlier is determined as the acceptable state range. It checks the difference mapping relationship item by item. When the candidate control has an first risk time earlier than the current maintenance state under the influence of the reaction phase transition deviation or the maintenance execution deviation, or when the risk frontier continues to expand into the non-risk area relative to the current maintenance state, the corresponding risk frontier state is identified as a failure state.

[0061] The cross-phase fidelity analysis layer marks the failure states identified by the falsification constraint discrimination layer back onto the candidate control cross-phase trajectory according to the reaction progress position. Starting from the reaction phase transition position, it determines the position of the first occurrence of the failure state along the reaction progression process of the expected effective reaction phase, and continuously counts the reaction progress range covered by the failure state from its appearance to its disappearance. Using the shift of the risk frontier, the delay in risk arrival, and the change in the expansion direction in the current reaction phase before the reaction phase transition position as the benchmark for risk suppression, it compares the same-scale state representations in the expected effective reaction phase after the reaction phase transition position segment by segment to determine how far the shift of the risk frontier, the delay in risk arrival, and the change in the expansion direction can be maintained, and determines the failure... The effective state occupies a range that is sustained in the expected effective response phase. The range of risk suppression maintained between the response phase transition position and the first occurrence of the failure state reflects the continuous maintenance capability of the risk suppression effect. The proportion of the range in the expected effective response phase where no failure state occurs to the entire cross-phase analysis range reflects the overall maintenance capability of the risk suppression effect. Then, the normalization and arithmetic average methods are used to combine the continuous maintenance capability and the overall maintenance capability, so that the candidate regulation with the later the failure state occurs and the smaller the duration range obtains a higher degree of maintenance, and the candidate regulation with the earlier the failure state occurs and the larger the duration range obtains a lower degree of maintenance, thus obtaining the cross-phase risk suppression fidelity value of each candidate regulation.

[0062] The control succession discriminant layer ranks candidate controls in adjacent reaction phases from high to low based on the cross-phase risk suppression fidelity of each candidate control. It then matches the reaction progress position, risk frontier position, risk arrival sequence, and expansion direction of the candidate control in the preceding reaction phase when it reaches the reaction phase transition position with the acceptable state range determined by the falsification constraint discriminant layer for the candidate control in the following reaction phase. If the reaction progress position at the end of the candidate control in the preceding reaction phase can enter the following reaction phase, and the risk frontier position, risk arrival sequence, and expansion direction are all within the acceptable state range of the candidate control in the following reaction phase, a candidate control succession relationship is established between the candidate control in the preceding and following reaction phases. Then, the control continues along the reaction phase of the following reaction phase... The process checks the maintenance of risk suppression corresponding to the cross-phase risk suppression fidelity value. If the risk frontier does not shift forward, the first risk time does not advance, and no failure state occurs after the transition, it is determined that the end state of the previous response phase regulation can enter the effective starting state of the candidate regulation of the next response phase. When any state exceeds the acceptable range, or the risk frontier expands, the first risk time advances, or a failure state occurs after the transition, it is determined that the end state of the previous response phase regulation cannot enter the effective starting state of the candidate regulation of the next response phase. The candidate regulation continues to be matched according to the cross-phase risk suppression fidelity value until a candidate regulation succession relationship that can continuously maintain risk suppression between adjacent response phases is determined, and the robust discrimination result of the candidate regulation is output.

[0063] It should be noted that the robustness judgment result of candidate regulation refers to determining each candidate regulation as either "able to continue to the next response phase and continuously suppress risk" or "unable to continue to the next response phase", and the judgment criteria are clearly defined as whether the risk frontier expands after continuation, whether the first risk time is advanced, and whether a failure state occurs.

[0064] The geopolymer phase discrimination model connects the phase event coding layer, cross-phase state mapping layer, falsification constraint discrimination layer, cross-phase fidelity analysis layer, and regulation succession discrimination layer using a sequential transmission and state field correspondence method. Specifically, the phase event coding layer forms continuous phase event codes based on material reaction phase, reaction progress, regulation intervention time, reaction phase transition position, risk frontier position change, risk arrival sequence, and expansion direction, and transmits these codes to the cross-phase state mapping layer in a fixed field order. The cross-phase state mapping layer uses the reaction phase transition position and reaction progress position as correspondence criteria to match the continuous phase event codes in the current reaction phase and the expected effective reaction phase item by item, forming a same-scale state representation of candidate regulation before and after the reaction phase. This same-scale state representation is then transmitted to the falsification constraint discrimination layer according to the same reaction progress position. The falsification constraint discrimination layer uses difference comparison and range determination methods to correlate the same-scale state representation with the risk evolution under the current maintenance state. The model compares the response phase transition deviation and maintenance execution deviation to determine the acceptable state range, identifies the failure state, and transmits it to the cross-phase fidelity analysis layer according to the response progress position corresponding to the failure state. The cross-phase fidelity analysis layer normalizes and arithmetically averages the first occurrence position, duration, and risk suppression effect maintenance range of the failure state to obtain the cross-phase risk suppression fidelity value, and transmits the cross-phase risk suppression fidelity value along with the corresponding candidate control to the control succession discrimination layer. The control succession discrimination layer uses state matching and sequential screening methods to compare the end state of the previous response phase control with the acceptable state range of the candidate control of the next response phase, and then combines the cross-phase risk suppression fidelity value to determine whether the risk suppression effect can be continuously maintained between adjacent response phases. Thus, the layers are connected in the order of "continuous phase event encoding - same-scale state representation - failure state - cross-phase risk suppression fidelity value - robust discrimination result of candidate control" to form the geopolymer phase discrimination model.

[0065] The training process of the geopolymer phase discrimination model is as follows: Geopolymer cross-phase regulation trajectories formed under different geopolymer ratios, curing boundary states, and material reaction phases were selected as training samples. Basic training data was formed based on the determined intervention time, reaction phase transition position, risk frontier position changes, risk arrival sequence, expansion direction, and reaction progress in each geopolymer cross-phase regulation trajectory. Following the processing method of the aforementioned phase event coding layer, the basic training data was converted into continuous phase event codes. The state mapping relationship was marked according to the actual corresponding reaction progress positions between adjacent reaction phases. Risk suppression and failure states were marked in conjunction with the risk evolution under the current curing state. The cross-phase risk suppression fidelity value was determined based on the first occurrence location, duration, and risk suppression effect range of the failure state. Simultaneously, the fidelity value was determined based on whether the previous reaction phase regulation termination state... The acceptable state range of candidate regulation in the next reaction phase is marked with robust discrimination results of candidate regulation. The training samples are divided into training data and validation data. The phase event encoding layer, cross-phase state mapping layer, falsification constraint discrimination layer, cross-phase fidelity analysis layer and regulation succession discrimination layer are trained sequentially using supervised learning method. The parameters of each layer are adjusted according to the continuous phase event encoding bias, state mapping bias, failure state discrimination bias, cross-phase risk suppression fidelity bias and candidate regulation robust discrimination bias using backpropagation method. The accuracy of state mapping, failure state identification and candidate regulation robust discrimination before and after candidate regulation crosses the reaction phase is repeatedly tested using validation data. The parameter adjustment is stopped when the validation bias in multiple rounds of training no longer decreases significantly. The trained geopolymer phase discrimination model is obtained.

[0066] Based on the geopolymer phase alignment control trajectory, the risk frontier position, initial risk time, and expansion direction of candidate control and maintaining the current maintenance state are compared according to the same reaction progress position. The state in which the risk frontier does not move forward, the initial risk time does not advance, and the risk frontier does not expand after the candidate control crosses the reaction phase is determined as the risk suppression state that can be maintained. The risk suppression state that can be maintained is continuously checked before and after the reaction phase transition position to establish cross-phase risk suppression constraints. On the basis of cross-phase risk suppression constraints, the reaction phase transition position is moved forward or backward according to the allowable change range of the reaction phase transition deviation. The actual time of the candidate control and the corresponding maintenance state are adjusted according to the allowable change range of the maintenance execution deviation. At the same time, the material reaction phase is kept to advance unidirectionally according to the reaction progress, and the adjusted maintenance state is kept within the adjustable range of the maintenance environment temperature, humidity, and maintenance equipment operating parameters. The feasible combination of different reaction phase transition positions, the actual time of the candidate control, and the maintenance state is determined as the physically feasible disturbance scenario.

[0067] For each physically feasible perturbation scenario, the reaction progress, material reaction phase, and deformation state before the intervention of candidate control in the geopolymer phase alignment control trajectory are used as common starting states. The reaction phase transition deviation and curing execution deviation corresponding to the physically feasible perturbation scenario are applied to the geopolymer reaction-deformation twin benchmark, and the deformation state under the action of candidate control is continuously deduced according to the aforementioned reaction progress calculation process, temperature and humidity deformation determination process, and finite element method. Under the same starting state, the deformation state of maintaining the current curing state is simultaneously deduced. At each calculation time, the first risk time, risk front position, and expansion direction of candidate control and maintaining the current curing state are compared. It is continuously checked whether the candidate control still meets the cross-phase risk suppression constraint under the action of reaction phase transition deviation and curing execution deviation. The risk front change states of candidate control before and after crossing the reaction phase in each physically feasible perturbation scenario are connected according to the unified prediction time sequence to form the geopolymer candidate control falsification trajectory.

[0068] Using the falsification trajectory of geopolymer candidate controls, the changes in the first risk time and risk frontier after crossing the reaction phase are examined one by one according to each physically feasible perturbation scenario. A case where the first risk time after crossing the reaction phase occurs earlier than maintaining the current maintenance state is defined as risk suppression failure. A case where the risk frontier continues to advance into the unrisk-free area relative to maintaining the current maintenance state is defined as risk frontier expansion. Candidate controls that exhibit risk suppression failure or risk frontier expansion in any physically feasible perturbation scenario are eliminated. For the remaining candidate controls, the delay in the first risk time and the risk frontier expansion before and after crossing the reaction phase are compared under the same physically feasible perturbation scenario. The distance suppression amount determines the degree of preservation of the first risk time delay effect after crossing the reaction phase and the degree of preservation of the risk frontier expansion suppression effect after crossing the reaction phase. The first risk time delay effect and the risk frontier expansion suppression suppression effect are weighted and synthesized. The lowest degree of preservation among each physically feasible perturbation scenario is used as the cross-phase risk suppression fidelity value of the corresponding candidate regulation. The candidate regulation that can stably delay the first risk time and continuously suppress the risk frontier expansion after crossing the reaction phase obtains a high cross-phase risk suppression fidelity value. The remaining candidate regulations are arranged from high to low according to the cross-phase risk suppression fidelity value to obtain the geopolymer candidate regulation fidelity sequence.

[0069] The expression for calculating the cross-phase risk suppression fidelity is: ; in, Indicates the first The cross-phase risk suppression fidelity of each candidate regulation, with a value range of 0 to 1; The number representing the physically feasible perturbation scenario; Indicates the candidate regulation number; This represents the set of physically feasible perturbation scenarios consisting of reaction phase transition deviation and maintenance execution deviation; The weighting coefficient represents the degree to which the initial risk duration is maintained, and its value ranges from 0 to 1; Indicates the first The candidate regulation in the first The initial risk time delay after crossing the reaction phase in a physically feasible perturbation scenario; Indicates crossing the reaction phase; Indicates the first The candidate regulation in the first The initial risk time delay before crossing the reaction phase in a physically feasible perturbation scenario; This represents the time correction amount used to prevent the denominator of the initial risk time delay from being zero. Indicates the first The candidate regulation in the first Under a physically feasible perturbation scenario, the risk frontier extension distance suppression amount after crossing the reaction phase; Indicates the first The candidate regulation in the first Under a physically feasible perturbation scenario, the risk frontier extension distance suppression amount across the reaction phase; This represents the distance correction amount, where the denominator of the distance suppression amount used to prevent the risk frontier from expanding is zero. Indicates the time of the first risk; Indicates the extended distance of the risk frontier; This indicates the period before crossing the reaction phase.

[0070] Based on the geopolymer candidate control fidelity sequence, the remaining candidate controls are grouped according to the expected effective reaction phase. Candidate controls with the same expected effective reaction phase are ranked from highest to lowest based on their cross-phase risk suppression fidelity. The candidate controls for each expected effective reaction phase are then determined sequentially according to the progression of the material reaction phases, forming a continuous control sequence. For candidate controls of adjacent reaction phases in the control sequence, the reaction progress, deformation state, risk front position, and first risk time when the previous candidate control reaches the reaction phase transition position are used as the state when the next candidate control begins to take effect. The reaction progress and deformation state continue to advance from the reaction phase transition position in the geopolymer reaction-deformation twin benchmark, switching to the maintenance state corresponding to the next candidate control, continuously extrapolating the control process. The initial risk time and risk frontier changes after the switch are analyzed. The initial risk time obtained from the subsequent simulation is compared with the delayed initial risk time maintained before the end of the candidate control of the previous reaction phase. At the same time, it is checked whether the risk frontier has shifted forward or whether the expansion range has increased. When the control switch causes the initial risk time to be advanced or the risk frontier continues to expand into the unrisk area, it is determined that the candidate control of the next reaction phase will generate new deformation risks. The next candidate control is selected from the same expected effective reaction phases according to the cross-phase risk suppression fidelity value from high to low to replace it. The subsequent simulation of adjacent reaction phases is carried out again until no initial risk time advances or risk frontier expansion occurs after the control switch of each adjacent reaction phase. The finally retained candidate controls are arranged according to the material reaction phase advancement order to generate a geopolymer robust control sequence.

[0071] Based on the geopolymer candidate control fidelity sequence, the remaining candidate controls are grouped according to the expected effective reaction phase. Candidate controls with the same expected effective reaction phase are ranked from highest to lowest based on their cross-phase risk suppression fidelity. The candidate controls for each expected effective reaction phase are then determined sequentially according to the progression of the material reaction phases, forming a continuous control sequence. For candidate controls of adjacent reaction phases in the control sequence, the reaction progress, deformation state, risk front position, and first risk time when the previous candidate control reaches the reaction phase transition position are used as the state when the next candidate control begins to take effect. The reaction progress and deformation state continue to advance from the reaction phase transition position in the geopolymer reaction-deformation twin benchmark, switching to the maintenance state corresponding to the next candidate control, continuously extrapolating the control process. The initial risk time and risk frontier changes after the switch are analyzed. The initial risk time obtained from the subsequent simulation is compared with the delayed initial risk time maintained before the end of the candidate control of the previous reaction phase. At the same time, it is checked whether the risk frontier has shifted forward or whether the expansion range has increased. When the control switch causes the initial risk time to be advanced or the risk frontier continues to expand into the unrisk area, it is determined that the candidate control of the next reaction phase will generate new deformation risks. The next candidate control is selected from the same expected effective reaction phases according to the cross-phase risk suppression fidelity value from high to low to replace it. The subsequent simulation of adjacent reaction phases is carried out again until no initial risk time advances or risk frontier expansion occurs after the control switch of each adjacent reaction phase. The finally retained candidate controls are arranged according to the material reaction phase advancement order to generate a geopolymer robust control sequence.

[0072] Based on the arrival time sequence of geopolymer regulation, the reaction progress is advanced hourly in the geopolymer reaction deformation twin datum, starting from the current reaction progress and material reaction phase, according to the arrival time of the regulation effect corresponding to each regulation segment. The material reaction phase at the time of arrival of the regulation effect is determined according to the reaction progress range corresponding to each material reaction phase in the geopolymer reaction phase state data. When the regulation effect is reached, the corresponding reaction progress and material reaction phase are retained, and the segments are associated one by one according to the arrangement relationship of each regulation segment in the geopolymer robust regulation sequence, so that each regulation segment has a corresponding arrival time of regulation effect, reaction progress and material reaction phase, forming the geopolymer regulation arrival state.

[0073] Based on the arrival status of the geopolymer regulation, the reaction progress and material reaction phase at the arrival of each regulation segment are compared with the reaction progress range of the corresponding target reaction phase. When the reaction progress at the arrival of the regulation has not yet entered the target reaction phase, the geopolymer reaction deformation twin benchmark is used to continue to extrapolate until the reaction progress enters the target reaction phase, and the command generation time is back-calculated by combining the actual execution delay of the corresponding regulation segment. When the reaction progress at the arrival of the regulation has already entered the target reaction phase, the command generation time is directly corrected forward based on the arrival time of the regulation and the actual execution delay. When the reaction progress at the arrival of the regulation has exceeded the reaction progress range corresponding to the target reaction phase, it indicates that the target reaction phase has been missed when the corresponding regulation segment actually takes effect. The corresponding regulation segment is eliminated, and the regulation segments and the corrected command generation times are arranged in the order of material reaction phase advancement to generate the geopolymer time delay compensation regulation command.

[0074] S4. Implement maintenance control according to the geopolymer time delay compensation control command and correct the digital twin state based on the actual deformation feedback to obtain the geopolymer deformation closed-loop control record.

[0075] Based on the corrected command generation time and maintenance control content of each control segment in the geopolymer time delay compensation control command, the maintenance environment temperature, humidity and maintenance equipment operating parameters are adjusted at the corresponding time, and the actual start and end times of each control segment are recorded. During the maintenance control period, the temperature and humidity state of the geopolymer component to be controlled is continuously collected by temperature and humidity sensors, and the strain and displacement of each entity calculation area are continuously collected by strain gauges and displacement sensors. The measured deformation is determined based on the characteristic length of each finite element. According to the actual execution time, the maintenance environment temperature, humidity and maintenance equipment operating parameters used in each control segment are associated with the temperature and humidity state and measured deformation collected at the same calculation time, so that each control segment corresponds to continuous temperature and humidity state and measured deformation from the start to the end of execution, forming geopolymer control execution feedback.

[0076] Based on the feedback from the geopolymer regulation execution, the reaction progress, material reaction phase, deformation state, and temperature and humidity state at the calculation time before the actual start of each regulation segment are taken as the common starting state. The actual curing environment temperature, humidity, and curing equipment operating parameters are written into the geopolymer reaction-deformation twin datum according to the actual execution time. The deformation evolution under actual curing regulation is replayed hourly according to the reaction progress calculation process, temperature and humidity deformation determination process, and finite element method. Starting from the same common starting state, the curing environment temperature, humidity, and curing equipment operating parameters before the actual start of the regulation segment are kept unchanged, and the deformation evolution under the original curing state is replayed using the same calculation time interval. According to the same calculation time and the same finite element, the deformation evolution under actual curing regulation and the deformation evolution under the original curing state are continuously correlated with the measured deformation in the geopolymer regulation execution feedback, forming a geopolymer double mirror feedback trajectory.

[0077] For example, when applying the geopolymer time delay compensation control command to the curing process of geopolymer precast wall panels, the curing environment temperature is adjusted from 35℃ to 32℃ and the environmental humidity is increased from 60% to 75% at the 14th hour. The curing equipment actually starts executing at 14 hours and 10 minutes and continuously records the temperature and humidity status and the measured deformation. For example, the measured deformation at 14 hours and 10 minutes is 1.20mm, the measured deformation at 15 hours is 1.31mm, and the measured deformation at 16 hours is 1.45mm. By linking the curing environment temperature, environmental humidity, curing equipment operating parameters, and measured deformation at the actual execution time, a geopolymer control execution feedback is formed.

[0078] For example, taking the reaction progress, material reaction phase, deformation state, and temperature and humidity state before 14 hours and 10 minutes as a common starting state, the deformation evolution under actual curing control and under the original curing state is replayed in the geopolymer reaction deformation twin benchmark. By the 16th hour, the predicted deformation under actual curing control is 1.42 mm, the predicted deformation under the original curing state is 1.76 mm, and the measured deformation is 1.45 mm. Since the measured deformation is closer to the predicted deformation under actual curing control, the two deformation evolutions and the measured deformation are continuously correlated according to each calculation time to form a geopolymer double mirror feedback trajectory.

[0079] Based on the geopolymer dual-mirror feedback trajectory, the deformation difference between the measured deformation and the actual curing-controlled mirror deformation, and the deformation difference between the measured deformation and the mirror deformation maintaining the original curing state, are calculated separately for the same calculation time and the same finite element. The absolute values ​​of the two deformation differences are compared. When the absolute value of the deformation difference between the measured deformation and the actual curing-controlled mirror deformation is small, the actual deformation change at the corresponding calculation time is determined to be the change caused by the actual curing control. When the absolute value of the deformation difference between the measured deformation and the mirror deformation maintaining the original curing state is small, the actual deformation change at the corresponding calculation time is determined to be the change caused by the continuous reaction of the material maintaining the original curing state. The two proximity relationships are compared segment by segment along the continuous calculation time to determine the control effect attribution of the actual deformation change. Based on the control effect attribution, the temperature and humidity state and the measured deformation in the geopolymer control execution feedback are written into the same spatial position. The reaction progress of each finite element is rechecked in combination with the reaction progress range corresponding to the current material reaction phase. The material reaction phase is corrected according to the unidirectional locking state of the material reaction phase. The deformation state of each finite element is corrected according to the measured deformation using the finite element inversion method. Based on the corrected deformation shape... The spatial boundary between finite elements that have reached the allowable deformation boundary and adjacent finite elements that have not yet reached the allowable deformation boundary is redefined, and the risk frontier position is corrected, thereby revising the digital twin state. Based on the revised digital twin state, the reaction progress of each finite element is advanced according to the reaction progress calculation process from the current calculation time. Temperature and humidity deformation is determined according to the temperature and humidity changes, and the subsequent deformation state is deduced using the finite element method in combination with mechanical constraint relationships. The deformation state is compared with the allowable deformation boundary hourly to redetermine the first risk time of each finite element. The earliest risk position and risk expansion direction are redetermined according to the connection relationship of finite elements in the digital twin space, and the updated deformation risk is deduced. According to the order of the generation time, actual execution time, geopolymer control execution feedback formation time, and digital twin state revision time of the geopolymer time delay compensation control command, the curing environment temperature, environmental humidity, curing equipment operating parameters, material reaction phase, measured deformation, control action attribution, digital twin state revision content, and the updated first risk time and risk frontier position are continuously recorded for each control segment to obtain the geopolymer deformation closed-loop control record.

[0080] In summary, this invention achieves continuous comparison and cross-phase tracking of the risk suppression effect of candidate controls at different reaction phases by constructing a three-track counterfactual twin branch and comparing the differences in risk frontier migration, thus improving the accuracy of geopolymer deformation risk prediction and the adaptability of cross-phase control. Furthermore, by using a geopolymer phase discrimination model to perform falsification analysis on the geopolymer cross-phase control trajectory and screen candidate controls, it achieves joint verification of the continued effectiveness and anti-disturbance capability of candidate controls after crossing the reaction phase. This eliminates candidate controls that fail to suppress risk or generate new deformation risks, thereby improving the robustness of the geopolymer control scheme and the reliability of deformation risk suppression.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for regulating geopolymer deformation based on digital twins, characterized in that, include: Collect basic data of the geopolymer components to be regulated, establish a digital twin space and associate the reaction state with the deformation state to generate a geopolymer reaction-deformation twin benchmark; The reaction phase of the material is identified using the geopolymer reaction deformation twin benchmark, the geopolymer reaction phase state data is obtained, the full-field deformation is reconstructed and the deformation risk is deduced to form a geopolymer deformation risk map. Based on the geopolymer deformation risk map, the risk frontier is identified. Under the same digital twin state, a three-track counterfactual twin branch is deduced, which includes maintaining the current maintenance state, current reaction phase candidate regulation, and expected effective reaction phase candidate regulation. Then, the cross-phase regulation trajectory of geopolymer is formed by combining the risk frontier migration difference. The cross-phase regulation trajectory of geopolymer is analyzed using the geopolymer phase discrimination model to verify the risk suppression effect of candidate regulation after crossing the reaction phase. Candidate regulation with failure and increased deformation risk is eliminated to generate a robust geopolymer regulation sequence. Finally, geopolymer time delay compensation regulation instructions are generated by combining the actual execution time delay. Maintenance control is implemented based on the time delay compensation control command for geopolymers, and the digital twin state is corrected based on the actual deformation feedback to obtain a closed-loop control record of geopolymer deformation.

2. The geopolymer deformation control method based on digital twins as described in claim 1, characterized in that, The basic data of the geopolymer component to be regulated includes the component's geometric parameters, geopolymer mix proportions, initial temperature and humidity conditions, initial mechanical conditions, curing boundary conditions, and allowable deformation boundaries.

3. The geopolymer deformation control method based on digital twins as described in claim 1, characterized in that, The specific steps for establishing a digital twin space and associating reaction and deformation states to generate a geopolymer reaction-deformation twin reference are as follows: The entity calculation area is divided according to the basic data of the geopolymer component to be controlled and spatial discretized. A digital twin space is established. The initial temperature and humidity state and the initial mechanical state are written into the digital twin space and the temperature and humidity transfer boundary and mechanical constraint relationship are established to obtain the initial spatial state of the geopolymer. The reaction progress is calculated based on the initial spatial state of the geopolymer and the geopolymer ratio parameters. Chemical shrinkage and temperature and humidity deformation are determined according to the reaction progress, and the deformation state is obtained by combining the mechanical constraint relationship. Then, the reaction state and the deformation state are associated and the allowable deformation boundary is bound to generate a geopolymer reaction deformation twin reference.

4. The geopolymer deformation control method based on digital twins as described in claim 3, characterized in that, The specific steps for obtaining the geopolymer reaction phase state data are as follows: Real-time temperature and humidity conditions and strain conditions are written into the geopolymer reaction deformation twin reference. The reference deformation value is calculated with the reaction progress fixed. The reaction-sensitive deformation is extracted based on the difference between the measured deformation value and the reference deformation value. The phase space is reconstructed in combination with the reaction progress change to form the geopolymer reaction deformation phase trajectory. The geopolymer reaction deformation phase trajectory is continuously segmented and the structural change positions of the relationship between the reaction progress state and the deformation response are identified. Combined with the state correction of adjacent regions, abnormal turning points are corrected. Then, the material reaction phase is determined according to the reaction progress and unidirectional locking is performed to obtain the geopolymer reaction phase state data.

5. The geopolymer deformation control method based on digital twins as described in claim 4, characterized in that, The specific steps for forming the geopolymer deformation risk map are as follows: Phase space constraint relationship is established based on geopolymer reaction phase state data, full-field deformation inversion is performed in combination with measured strain, and the deformation evolution of each entity's calculation region is deduced according to the material reaction phase to form the geopolymer predicted deformation trajectory. By using geopolymer-predicted deformation trajectories to determine the initial risk time and form a geopolymer risk arrival time field, the earliest risk location and expansion direction are identified, and a geopolymer deformation risk map is generated.

6. The geopolymer deformation control method based on digital twins as described in claim 5, characterized in that, The risk frontier is locked based on the geopolymer deformation risk map. Under the same digital twin state, a three-track counterfactual twin branch is deduced to maintain the current maintenance state, the current response phase candidate regulation, and the expected effective response phase candidate regulation. Then, the cross-phase regulation trajectory of the geopolymer is formed by combining the differences in risk frontier migration. The specific steps are as follows: Based on the geopolymer deformation risk map, the risk frontier is locked, the digital twin state at the locking time is extracted, and the reaction phase and deformation state are fixed to form the geopolymer risk frontier mother state; Based on the parent state of the geopolymer risk frontier, candidate regulation is generated, and a homologous inference path is established for maintaining the current maintenance state, the current response phase candidate regulation, and the expected effective response phase candidate regulation, generating a three-track counterfactual twin branch; Based on the three-track counterfactual twin branch, the material reaction phase and deformation state are advanced according to the unified prediction time sequence, and the prediction state of each branch is synchronized at the reaction phase transition position. The risk frontier position and expansion process of each branch are continuously tracked to form a three-track risk frontier sequence. Based on the three-track risk frontier sequence, and using the risk frontier migration process under the current maintenance state as a control, the changes in the risk frontier position, the order of risk arrival, and the direction of expansion under the current response phase candidate regulation and the expected effective response phase candidate regulation were compared to obtain the differences in risk frontier migration. By using the differences in risk frontier migration to correlate the response phase transition position and the timing of regulatory intervention for candidate regulation, and connecting the risk frontier changes before and after the candidate regulation crosses the response phase according to a unified prediction time series, a geopolymer cross-phase regulation trajectory is formed.

7. The geopolymer deformation control method based on digital twins as described in claim 6, characterized in that, The process involves analyzing the cross-phase regulation trajectory of geopolymers using a geopolymer phase discrimination model, verifying the risk suppression effect of candidate regulation after crossing the reaction phase, eliminating candidate regulation with the risk of failure and increased deformation, and generating a robust geopolymer regulation sequence. The specific steps are as follows: Geopolymer phase discrimination model is used to analyze geopolymer cross-phase regulation trajectory. Based on the reaction phase transition position and regulation intervention time, the trajectory state of candidate regulation before and after crossing the reaction phase is aligned to generate geopolymer phase-aligned regulation trajectory. Based on the phase alignment control trajectory of geopolymers, cross-phase risk suppression constraints are established, and physically feasible disturbance scenarios are constructed by combining the reaction phase transition deviation and maintenance execution deviation. Then, counterfactual falsification inferences are performed on candidate controls to form the falsification trajectory of geopolymer candidate controls. The risk changes of each candidate regulation after crossing the reaction phase are verified by using the falsification trajectory of the candidate regulation of geopolymer. Candidate regulations that show risk suppression failure and risk frontier expansion are eliminated. Then, the cross-phase risk suppression fidelity value of the remaining candidate regulations is calculated to obtain the fidelity sequence of the candidate regulation of geopolymer. Based on the geopolymer candidate regulation fidelity sequence, candidate regulations are screened according to the expected effective response phase and the regulation order is determined by combining the cross-phase risk suppression fidelity value. Then, the candidate regulations of adjacent response phases are continuously simulated to verify the first risk time and risk frontier changes after regulation switching. Candidate regulations that generate new deformation risks are replaced to generate a geopolymer robust regulation sequence.

8. The geopolymer deformation control method based on digital twins as described in claim 7, characterized in that, The specific steps for generating the geopolymer delay compensation control command are as follows: Based on the robust geopolymer regulation sequence, the actual execution delay of each regulation segment is read, and the arrival time of the regulation effect to the risk frontier is determined by combining the risk frontier location, thus obtaining the geopolymer regulation arrival time series; Based on the arrival time of geopolymer regulation, the reaction progress and material reaction phase when the regulation effect arrives are deduced using the geopolymer reaction deformation twin benchmark, forming the geopolymer regulation arrival state. Then, the command generation time of each regulation segment is corrected according to the geopolymer regulation arrival state, and regulation segments that exceed the target reaction phase are eliminated, generating geopolymer time delay compensation regulation commands.

9. The geopolymer deformation control method based on digital twins as described in claim 8, characterized in that, The specific steps for obtaining the closed-loop regulation record of geopolymer deformation are as follows: Based on the geopolymer time delay compensation control command, maintenance control is implemented and temperature and humidity status and measured deformation are collected. The actual execution time is associated to form geopolymer control execution feedback. The actual maintenance control and deformation evolution under the original maintenance state are then replayed and associated with the measured deformation to form a geopolymer dual mirror feedback trajectory. Based on the comparison of the measured deformation with the actual curing-controlled mirror deformation and the mirror deformation maintaining the original curing state according to the dual mirror feedback trajectory of the geopolymer, the control effect of the actual deformation change is determined. The digital twin state is corrected by combining the material reaction phase and the risk frontier position, and the updated deformation risk is deduced. The curing control and digital twin correction process is recorded according to the control time sequence to obtain the closed-loop control record of geopolymer deformation.

10. The method for regulating geopolymer deformation based on digital twins as described in claim 7, characterized in that, The geopolymer phase discrimination model includes: The phase event coding layer receives the cross-phase regulation trajectory of the geopolymer, extracts the risk frontier state before and after the candidate regulation intervention according to the reaction phase transition position, and converts the reaction phase transition position, regulation intervention time and risk frontier migration state into continuous phase event codes; The cross-phase state mapping layer establishes a state mapping relationship between adjacent response phases based on the encoding of continuous phase events, maps the risk suppression state in the current response phase to the expected effective response phase, and obtains the same-scale state representation of the candidate regulation before and after crossing the response phase. The falsification constraint discrimination layer maps the differences between the same-scale state representation and the risk evolution under the current maintenance state, and uses the reaction phase transition bias and maintenance execution bias to limit the acceptable state range, and identifies the failure state that causes the first risk time to advance and the risk frontier to expand. The cross-phase fidelity analysis layer analyzes the degree of retention of the risk suppression effect of the current response phase after it is transmitted to the expected effective response phase, based on the location and duration of the failure state in the cross-phase trajectory of the candidate regulation, and obtains the cross-phase risk suppression fidelity value of each candidate regulation. The regulation succession discrimination layer establishes the candidate regulation succession relationship between adjacent response phases based on the cross-phase risk suppression fidelity value, determines whether the end state of regulation of the previous response phase can enter the effective starting state of candidate regulation of the next response phase, and outputs the robust discrimination result of candidate regulation.

Images