Data-driven high-speed railway bridge builder template construction quality management method

Abstract

This invention discloses a data-driven method for quality management of formwork construction in high-speed railway bridge-building machines, belonging to the field of engineering construction management technology. It includes acquiring multi-source real-time status data, environmental monitoring data, and historical adjustment records; calculating based on an initial quality sensitivity model to generate a dynamic quality sensitivity graph and perform resource scheduling to obtain refined monitoring data; then performing digital twin simulation of the construction process to generate a simulation report and conduct multi-objective optimization decision analysis; generating adjustment instructions; controlling the execution mechanism to execute the adjustment instructions; collecting actual formwork response data and comparing it with the simulation report; generating model calibration data; and updating the initial quality sensitivity model. This invention employs a collaborative working mode of multi-source data perception, digital twin simulation, closed-loop feedback control, and model self-optimization, enabling predictive management of construction quality, scientific optimization of decisions, and continuous improvement of the process.

Description

Technical Field

[0001] This invention relates to the field of engineering construction management technology, and in particular to a data-driven method for quality management of formwork construction for high-speed railway bridge building machines. Background Technology

[0002] As a key component of modern transportation infrastructure, high-speed railways face extremely stringent requirements regarding structural safety and durability. Bridge-building machines are the core equipment for constructing the superstructure of high-speed railway bridges, and the formwork system they support serves as the direct mold for shaping the concrete beams. The geometric accuracy, stability, and stiffness of the formwork system directly determine the final alignment, dimensions, and internal quality of the bridge structure. Therefore, high-quality management of the formwork construction process using bridge-building machines is a prerequisite for ensuring the overall quality of high-speed railway bridge projects.

[0003] Currently, the construction quality management of formwork for high-speed railway bridge construction machines mainly relies on pre-set construction plans and on-site manual monitoring. Construction units typically dispatch surveyors to conduct static, discrete geometric measurements of key control points of the formwork using equipment such as total stations during critical construction phases, such as before and after concrete pouring. When deviations are found in the measurement data, on-site technicians adjust the formwork support system based on their engineering experience. Throughout the process, dynamic information such as the impact of environmental factors, changes in equipment operating status, and fluctuations in concrete material properties is primarily managed through on-site personnel's observation and qualitative judgment.

[0004] Existing technologies exhibit several shortcomings in addressing the extreme precision requirements of high-speed railway construction. First, measurement data suffers from significant lag; manual, periodic measurements cannot reflect the dynamic response of formwork under concrete pouring and environmental changes in real time, often making it difficult to completely correct quality deviations by the time they are discovered. Second, adjustment decisions heavily rely on personal experience, lacking quantitative prediction of the consequences of adjustments, leading to the risk of over- or under-adjustment, and potentially even triggering new quality problems. Finally, multi-source, heterogeneous data on-site, such as environmental monitoring data, equipment status data, and geometric measurement data, are typically fragmented and fail to be effectively integrated for comprehensive analysis and risk prediction. This results in a reactive management process lacking proactive prevention capabilities. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a data-driven method for managing the construction quality of formwork for high-speed railway bridge-building machines. This method employs a collaborative working mode that integrates multi-source data sensing, digital twin simulation, closed-loop feedback control, and model self-optimization. This enables predictive management of construction quality, scientific optimization of decision-making, and continuous improvement of the process.

[0006] The above objectives can be achieved through the following approach:

[0007] A data-driven method for quality management of formwork construction in high-speed railway bridge-building machines includes: acquiring multi-source real-time status data, environmental monitoring data, and historical adjustment records of the formwork system; calculating and generating a dynamic quality sensitivity map of the current construction stage based on a preset initial quality sensitivity model; dynamically sensing and scheduling resources according to the dynamic quality sensitivity map to obtain refined monitoring data; acquiring preset proposed adjustment instructions; performing a digital twin simulation of the construction process using the proposed adjustment instructions, the refined monitoring data, real-time measured parameters of concrete materials, and current environmental boundary conditions to generate a simulation report; performing multi-objective optimization decision analysis based on the simulation report to generate adjustment instructions, controlling the execution mechanism to execute the adjustment instructions, and collecting actual formwork response data; generating model calibration data based on the comparison results between the actual formwork response data and the simulation report, and updating the initial quality sensitivity model using the model calibration data.

[0008] Optionally, generating the quality sensitivity dynamic map for the current construction stage includes: collecting geometric deformation data, hydraulic jacking pressure data, and vibration frequency data of the formwork system as multi-source real-time status data; simultaneously acquiring ambient temperature, humidity, wind speed, and vibration data from adjacent construction sites as environmental monitoring data; and retrieving historical adjustment frequency and adjustment amplitude records for the same construction location; inputting the multi-source real-time status data, environmental monitoring data, and historical adjustment records into the initial quality sensitivity model to calculate the quality deviation risk coefficient for each monitoring point; classifying the risk level of each area of ​​the formwork according to the quality deviation risk coefficient, and performing dynamic color rendering to generate the quality sensitivity dynamic map.

[0009] Optionally, the method further includes: collecting historical geometric deformation data, hydraulic jacking response data, and ambient temperature and humidity fluctuation ranges of the formwork system at different construction stages, and constructing a deformation feature library of the formwork under different load and environmental coupling conditions; performing cluster analysis on the data in the deformation feature library to generate sensitivity weight coefficients for each monitoring point to the disturbance source; spatially mapping the sensitivity weight coefficients to the structural partitions of the formwork system, and normalizing the sensitivity distribution at different construction stages to generate an initial quality sensitivity model.

[0010] Optionally, obtaining refined monitoring data includes: analyzing the risk distribution based on the quality sensitivity dynamic map to determine risk areas and non-sensitive areas; mobilizing mobile sensors to detect risk areas to obtain local detection data, and simultaneously reducing the sampling frequency of fixed sensors in non-sensitive areas to obtain fixed sensor data; and performing feature-level fusion of the local detection data and the fixed sensor data to generate refined monitoring data.

[0011] Optionally, generating the simulation report includes: performing multiphysics coupling simulation based on the proposed adjustment instruction, the refined monitoring data, the measured parameters of the concrete material, and the current environmental boundary conditions to generate preliminary simulation results; performing data post-processing based on the preliminary simulation results to extract geometric shape prediction data and potential internal defect risk prediction data; and integrating the geometric shape prediction data and the internal defect risk prediction data to form a simulation report.

[0012] Optionally, the acquisition of actual template response data includes: coupling and comparing the geometric shape prediction data and the internal defect risk prediction data in the simulation report to identify adjustable variables and constraint boundaries during construction, and generating multiple sets of candidate adjustment parameters; performing multi-source feedback comparison based on the multiple sets of candidate adjustment parameters to select an adjustment scheme that simultaneously meets the requirements of morphological accuracy and structural density as an adjustment instruction; issuing the adjustment instruction and acquiring real-time response data of template strain, displacement, and concrete flow state as actual template response data.

[0013] Optionally, the generation of model calibration data includes: spatially registering and comparing the actual geometric data in the actual response data of the template with the geometric shape prediction data to generate geometric deviation data; performing correlation analysis on the actual mechanical response data in the actual response data of the template with the internal defect risk prediction data in the simulation report to generate risk prediction deviation data; and fusing the geometric deviation data with the risk prediction deviation data to form model calibration data.

[0014] Optionally, updating the initial quality sensitivity model using the model calibration data includes: adjusting the sensitivity weight coefficients in the initial quality sensitivity model based on the geometric deviation data to generate updated sensitivity weight coefficients; correcting the threshold parameters related to defect sensitivity in the initial quality sensitivity model based on the risk prediction deviation data to generate updated risk warning thresholds; and synchronously updating the updated sensitivity weight coefficients and the updated risk warning thresholds to generate a corrected quality sensitivity model.

[0015] Optionally, the method further includes: continuously monitoring the geometric deviation data of the preset monitoring points; and when it is identified that the geometric deviation data of the monitoring points continuously exceeds the preset deviation threshold in multiple consecutive construction cycles, increasing the sensitivity weight coefficient of the monitoring points in the modified quality sensitivity model.

[0016] Based on the same inventive concept, this invention also provides a data-driven high-speed railway bridge-building machine formwork construction quality management system. The system includes: a data modeling module, used to acquire multi-source real-time status data, environmental monitoring data, and historical adjustment records of the formwork system, and calculate based on a preset initial quality sensitivity model to generate a dynamic quality sensitivity map for the current construction stage; a perception and scheduling module, used to dynamically perceive and schedule resources according to the dynamic quality sensitivity map to obtain refined monitoring data; a twin simulation module, used to acquire preset proposed adjustment instructions, and perform digital twin simulation of the construction process using the proposed adjustment instructions, the refined monitoring data, real-time acquired measured parameters of concrete materials, and current environmental boundary conditions to generate a simulation report; an optimization and control module, used to perform multi-objective optimization decision analysis based on the simulation report, generate adjustment instructions, control the execution mechanism to execute the adjustment instructions, and collect actual formwork response data; and a calibration and update module, used to generate model calibration data based on the comparison results between the actual formwork response data and the simulation report, and use the model calibration data to update the initial quality sensitivity model.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] 1. This invention achieves predictability and proactivity in construction quality management by constructing a data-driven quality sensitivity model and generating dynamic risk maps using real-time multi-source data. It transforms traditional reactive quality inspection into proactive risk prediction, enabling managers to identify weaknesses and potential risk sources in the formwork system in advance. This shifts the focus of quality control from post-event remediation to pre-event prevention, thereby improving the level of construction quality assurance.

[0019] 2. This invention achieves refined and efficient management of monitoring resources through a sensing resource scheduling mechanism based on a dynamic quality sensitivity graph. It intelligently and dynamically allocates high-precision, high-frequency monitoring resources to the highest-risk areas while reducing monitoring intensity in non-sensitive areas. This not only optimizes the configuration of sensing devices and data processing resources but also effectively avoids data redundancy. While ensuring the comprehensiveness of quality information for key components, it improves the operational efficiency and economy of the entire quality monitoring system.

[0020] 3. This invention utilizes digital twin simulation and multi-objective optimization decision analysis of the construction process to significantly improve the scientific rigor and reliability of construction adjustment instructions. Before any proposed adjustment scheme is implemented, a high-fidelity simulation prediction is conducted in virtual space to comprehensively assess its combined impact on the final geometric shape and internal structural density. This avoids the risks associated with blindly trying and failing based on personal experience, ensuring that the final executed adjustment instruction is the optimal solution that satisfies multiple quality objectives.

[0021] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a data-driven method for managing the construction quality of formwork for high-speed railway bridge building machines, according to an embodiment of the present invention.

[0024] Figure 2 This is a dynamic graph showing the quality sensitivity of the template system in an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of a digital twin simulation report according to an embodiment of the present invention.

[0026] Figure 4 This is a schematic diagram of the structure of a data-driven high-speed railway bridge-building machine formwork construction quality management system according to an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Reference Figure 1 One embodiment of the present invention proposes a data-driven method for construction quality management of formwork for high-speed railway bridge building machines. It adopts a collaborative working mode of multi-source data perception, digital twin simulation, closed-loop feedback control and model self-optimization, which can realize predictive management of construction quality, scientific optimization of decision-making and continuous improvement of the process.

[0029] The method described in this embodiment specifically includes:

[0030] The system acquires multi-source real-time status data, environmental monitoring data, and historical adjustment records from the template system, and calculates based on a preset initial quality sensitivity model to generate a dynamic quality sensitivity graph for the current construction stage.

[0031] Optionally, generating the quality sensitivity dynamic map for the current construction stage includes:

[0032] Collect geometric deformation data, hydraulic jacking pressure data, and vibration frequency data of the template system as multi-source real-time status data. Simultaneously acquire ambient temperature, humidity, wind speed, and vibration data of nearby construction sites as environmental monitoring data. Also retrieve historical adjustment frequency and adjustment amplitude records for the same construction site.

[0033] The multi-source real-time status data, environmental monitoring data and historical adjustment records are input into the quality sensitivity initial model to calculate the quality deviation risk coefficient of each monitoring point.

[0034] Based on the quality deviation risk coefficient, the risk level of each area of ​​the template is divided, and dynamic color rendering is performed to generate a quality sensitivity dynamic map.

[0035] Specifically, the process begins with the synchronous acquisition and integration of multi-dimensional data. The system uses sensor arrays deployed at key nodes of the formwork to collect real-time multi-source status data, including geometric deformation data acquired using high-precision total stations or laser scanners (with sampling frequencies typically set between 0.2-1 Hz); hydraulic jacking pressure data collected by pressure sensors integrated into the jacking cylinders; and vibration frequency data collected by acceleration sensors attached to the main beam and web of the formwork. Simultaneously, an integrated environmental monitoring unit installed on the bridge-building machine acquires environmental monitoring data such as temperature, humidity, and wind speed, and combines this with external seismic monitoring instruments or vibration sensors placed near adjacent piers to obtain vibration data from nearby construction sites. Furthermore, historical adjustment records matching the current construction segment type and spatial location are retrieved from the project's historical database, with a focus on extracting the frequency and magnitude of historical adjustments. Subsequently, the integrated data is input into a preset quality sensitivity initial model to calculate the quality deviation risk coefficient. The quality deviation risk coefficient is a dimensionless comprehensive indicator used to quantify the probability of excessive deformation or concrete pouring defects at a specific monitoring point under current working conditions. For calculating the quality deviation risk coefficient of monitoring point i ,have:

[0036] ;

[0037] in, , , These are the measured values ​​of the geometric deformation, associated hydraulic pressure, and vibration frequency at that point, respectively. The comprehensive environmental impact parameters representing this point; Adjust the historical feature parameters for this point; , , , , These are the weighting coefficients for each influencing factor, which are pre-calibrated by the initial quality sensitivity model based on historical data analysis. , , , These are temperature-related factors such as deformation caused by thermal expansion, humidity-related factors such as material moisture absorption and expansion, and vibration-related factors from nearby construction such as vibration frequency. to These are the weighting coefficients for each environmental factor, which can be set by fitting historical data or through experience; , These represent the historical adjustment frequency and average adjustment magnitude for that point, respectively. , The fusion weights are frequency and amplitude, respectively, and can be determined based on engineering experience or cluster analysis; For example, the minimum-maximum normalization function (MPNU) is used in engineering to convert physical quantities of different dimensions into standardized risk contribution values ​​between 0 and 1, such as for geometric deformation data. Suppose that the measured value of the geometric deformation at a certain monitoring point i is 8 mm, while the minimum value of the geometric deformation recorded at this point in historical data is 2 mm and the maximum value is 12 mm, then its normalized value... Finally, based on the calculated quality deviation risk coefficients of each monitoring point, risk levels are classified and visualized. For example, three risk thresholds are set, dividing the risk into low, medium, and high levels. This risk-classified point information is mapped onto the 3D digital twin model of the bridge-building machine template, and smooth interpolation is performed to form a continuous color block distribution, thereby generating a real-time updated dynamic quality sensitivity map. Figure 2 As shown, this figure is an example of a dynamic graph of the quality sensitivity of a template section. The graph is dynamically colored and rendered by varying shades of gray. The darker the color, the higher the risk coefficient of the quality deviation, which is the "high-risk area"; the lighter the color, the "medium-risk" or "non-sensitive area".

[0038] For example, consider the construction of a 32-meter simply supported box girder in a high-speed railway project using the moving formwork method. Before pouring concrete for a certain segment of the box girder, a data acquisition program is initiated. First, real-time geometric deformation data for that area is acquired at a frequency of 0.5 Hz using a total station prism positioned in the middle of the left flange of the formwork; the current reading is 8 mm. Simultaneously, the hydraulic jacking pressure data transmitted by the pressure sensor of the jacking cylinder associated with that area is 20 MPa. The main vibration frequency measured by the acceleration sensor attached to the web of the main girder of the formwork is 30 Hz. These constitute multi-source real-time status data. At the same time, the integrated environmental monitoring unit on the bridge-building machine shows an ambient temperature of 35 degrees Celsius, humidity of 75%, wind speed of 3 m / s, and no significant construction vibration from nearby piers. Historical adjustment records for the same section of the same segment previously poured were retrieved from the historical database, revealing that this section had undergone two minor adjustments due to deformation, with an average adjustment range of 2.5 mm. The weight coefficients of each influencing factor are preset by the model, such as... It is 0.4. It is 0.3. It is 0.1. It is 0.1. The value is 0.1. First, the physical quantities are normalized to obtain... The values ​​were set to 0.6, 0.7, 0.4, 0.5, and 0.6 respectively. These data were then input into the initial quality sensitivity model to calculate the quality deviation risk coefficient of the monitoring points. If the set risk thresholds are low risk (less than 0.4), medium risk (0.4 to 0.7), and high risk (greater than 0.7), then the point is classified as medium risk. Ultimately, an intuitive dynamic quality sensitivity graph is generated. Through the fusion calculation and visualization of multi-source data, the level of proactive control over construction quality can be improved during the critical preparation stage of concrete pouring.

[0039] Optionally, the method further includes:

[0040] Collect historical geometric deformation data, hydraulic jacking response data, and ambient temperature and humidity fluctuation ranges of the formwork system at different construction stages, and construct a deformation feature library of the formwork under different load and environmental coupling conditions;

[0041] Cluster analysis is performed on the data in the deformation feature library to generate sensitivity weight coefficients for each monitoring point to the disturbance source;

[0042] The sensitivity weight coefficients are spatially mapped to the structural partitions of the template system, and the sensitivity distribution at different construction stages is normalized to generate an initial quality sensitivity model.

[0043] Specifically, the process begins with continuously collecting historical operational data of the formwork system at different construction stages. This data covers the formwork's response behavior in the actual engineering environment. Specifically, this includes historical geometric deformation data recorded by devices such as laser displacement sensors and inertial measurement units, typically at a frequency of 1 Hz; hydraulic jacking response data recorded by pressure and displacement sensors on hydraulic cylinders, including the hysteresis curve between jacking force and displacement; and environmental temperature and humidity fluctuations recorded by an array of environmental sensors installed at the construction site. This data is stored and integrated into a structured historical database, forming a comprehensive deformation feature library. Subsequently, cluster analysis is performed on the massive historical data in the deformation feature library. This analysis aims to identify typical deformation patterns at each monitoring point of the formwork under different load conditions, such as the concrete pouring stage and the jacking stage, and under different environmental conditions, such as high temperatures in summer and low temperatures in winter, and their correlation with load and environmental factors. For example, unsupervised learning algorithms such as K-Means clustering can be used to group the high-dimensional historical data. Within each cluster, principal component analysis (PCA) or correlation analysis is used to identify the main influencing factors causing deformation at the monitoring points. For example, for a given monitoring point, if its deformation under gravity load has a high correlation with other factors such as temperature, then that point is considered to be more sensitive to temperature changes. Through this analysis, a sensitivity weighting coefficient can be calculated for each monitoring point to its disturbance source, such as load change, temperature gradient, or wind load. This coefficient characterizes the relative magnitude of the contribution of a specific disturbance source to the deformation of the monitoring point under a given working condition. The sensitivity weighting coefficient for monitoring point i under the action of disturbance source k is calculated as follows: ,have:

[0044] ;

[0045] in, The deformation of monitoring point i under the action of disturbance source k; This represents the total number of disturbance sources. The values ​​range from 0 to 1, and for any monitoring point i, the sum of the sensitivity weight coefficients of all disturbance sources is 1. Further, the sensitivity weight coefficients are spatially mapped to the actual structural partitions of the formwork system. The formwork system is typically divided into different areas based on its function and structural characteristics, such as the main beam area and the support leg area. This spatial mapping extends the sensitivity of discrete monitoring points to the entire structural partition, ensuring the spatial continuity of the entire model. Next, the sensitivity distribution at different construction stages is subjected to max-min normalization. Finally, these spatially mapped and normalized sensitivity weight coefficients, along with preset structural limits and material properties, are used to construct the initial quality sensitivity model. First, a three-dimensional structural digital model of the template system is established based on finite element analysis or equivalent numerical simulation methods. Sensitivity weight coefficients for each structural zone, such as the main beam zone and supporting leg zone, are assigned as field variables to each node of the model, forming a spatially distributed sensitivity field. Second, based on material mechanical properties such as elastic modulus and structural design specifications, allowable deformation thresholds and ultimate bearing capacities for each zone under disturbances such as load and temperature are set as the model's quality constraint boundaries. Finally, the sensitivity field and quality constraint boundaries are coupled and integrated with a real-time data interface to construct an initial quality sensitivity model capable of responding to multi-source disturbances in real time and dynamically assessing the quality deviation risk of each region. This initial quality sensitivity model is trained offline using multi-source historical data covering at least 20 consecutive construction cycles. The gradient descent method is used to iteratively optimize the weight coefficients. Model convergence is determined when the loss function value on the validation set decreases by less than 0.01 for five consecutive iterations. The training data sources cover different seasonal temperature ranges, concrete pouring volumes, and template structural zones to ensure the model's generalization ability.

[0046] For example, before the official commencement of the aforementioned high-speed railway project, technicians collected construction data from another similar project, such as a bridge, over the past year. First, they retrieved historical geometric deformation data for key monitoring points of the formwork under different seasons and pouring stages from historical archives. This data was recorded by laser displacement sensors at a frequency of 1 Hz. Simultaneously, they collected pressure and displacement response data of the hydraulic system when pushing the formwork and bearing concrete loads. Furthermore, they compiled data on temperature fluctuations, such as 30 to 42 degrees Celsius during the summer high-temperature period and -5 to 10 degrees Celsius during winter. These data collectively constituted a deformation feature library. Subsequently, based on the historical data of the bottom support point A of the main beam of the formwork, two main types of typical working conditions were identified: one was the summer high-temperature, fully loaded pouring condition, and the other was the winter normal-temperature, unloaded pushing condition. Under the first type of working condition, principal component analysis revealed that the main influencing factors causing deformation at this point were the concrete self-weight load and thermal expansion caused by high temperature. If there are two disturbance sources, disturbance source 1 is the concrete self-weight, and disturbance source 2 is thermal stress. Analysis shows that under this working condition, the deformation caused by the concrete's self-weight is 5 mm, while the deformation caused by thermal stress is 3 mm. If monitoring point i is support point A, and the total number of disturbance sources n is 2, how do we calculate the sensitivity weighting coefficient of monitoring point i under disturbance source 1? Similarly, the sensitivity weighting coefficient corresponding to thermal stress is calculated to be 0.375. These two coefficients indicate that, under this working condition, the deformation of support point A is more sensitive to its own weight load than to thermal stress. Furthermore, the calculated sensitivity weighting coefficients of all monitoring points are spatially mapped to the corresponding nodes of the three-dimensional digital model of the template structure, forming a continuous sensitivity field. Finally, combined with the mass constraint boundary, which allows for a deformation threshold of ±8 mm in this area according to the design specifications, the sensitivity field is coupled with the constraint boundary to form an initial mass sensitivity model capable of dynamically assessing quality risks.

[0047] Dynamic resource scheduling is performed based on the aforementioned quality sensitivity dynamic graph to obtain refined monitoring data;

[0048] Optionally, obtaining refined monitoring data includes:

[0049] Based on the risk distribution in the aforementioned quality sensitivity dynamic diagram, risk areas and non-sensitive areas are determined.

[0050] Movable sensors are used to detect risk areas to obtain local detection data, while the sampling frequency of fixed sensors in non-sensitive areas is reduced simultaneously to obtain fixed sensor data.

[0051] The local detection data and the fixed sensor data are fused at the feature level to generate refined monitoring data.

[0052] Specifically, the process begins with an in-depth analysis of the risk distribution information in the generated quality sensitivity dynamic map. This map visually indicates the risk level of each region within the template. Using image processing algorithms or threshold-based region segmentation methods, "risk areas" requiring focused attention and "non-sensitive areas" where monitoring intensity can be appropriately reduced are identified. Subsequently, intelligent scheduling of dynamic sensing resources is implemented. For identified risk areas, mobile monitoring robots or drones equipped with high-precision sensors are deployed for close-range, traversal detection. This detection method acquires high-resolution local detection data, such as using a laser line scanner to obtain millimeter-level deformation data or using acoustic imaging equipment to detect internal microcracks. Mobile sensors are typically equipped with high-frequency data acquisition modules, with sampling frequencies set to 10-50 Hz to capture instantaneous dynamic changes. Simultaneously, in areas designated as non-sensitive, the sampling frequency of fixed sensors is intelligently reduced, for example, from the conventional 1 Hz to 0.1 Hz or even lower, to reduce data redundancy and processing load. At this point, fixed sensors continue to collect data, such as those from general-purpose displacement sensors and temperature sensors; this data is collectively referred to as fixed sensor data. This strategy reduces the enormous data volume and computational burden associated with high-precision, high-frequency detection of the entire area. Finally, feature-level fusion is performed on the acquired local detection data and fixed sensor data to generate the final refined monitoring data. Feature-level fusion means not simply splicing the original data, but extracting meaningful features from both types of data. The mean of the fused refined monitoring data at a certain time or spatial point is then considered. ,have:

[0053] ;

[0054] in, Let u be the u-th feature value in the local detection data. If a drone uses a laser scanner to acquire point cloud data of a region, This may represent the maximum deformation or average curvature change at the center point of the region; For the v-th feature value in the fixed sensor data, such as for a fixed displacement sensor, It could be its displacement reading at the current moment; fusion weighting coefficient and The two satisfy the condition that their sum equals 1. In high-risk areas such as crack tips, mobile sensors that can acquire high-precision data are more trusted. It will be set very high, such as 0.9. The value is 0.1.

[0055] For example, the central control system issues commands to a tracked monitoring robot equipped with a high-precision laser line scanner. The robot automatically navigates to the area below the left wing flange and begins a comprehensive, close-proximity probe of the risk area at a frequency of 10 Hz, acquiring millimeter-resolution 3D point cloud data of the area. This data constitutes the local probe data. Simultaneously, commands are sent to fixed displacement sensors located on the right side of the web to reduce their data sampling frequency from the usual 1 Hz to 0.1 Hz. The data collected by these sensors at this point is referred to as fixed sensor data. Finally, at the center of the risk area, the maximum deformation feature value extracted from the robot's scanned local probe data is +7.8 mm. The last data collected by the fixed sensor at this location before the frequency reduction, i.e., the feature value in the fixed sensor data, is +7.5 mm. Because this area is a risk area, and the mobile robot has higher detection accuracy, and Let's set them to 0.8 and 0.2 respectively. Then calculate... Millimeters. This result replaces the original coarse reading of 8 millimeters. Feature-level fusion ensures the integrity and systematic nature of the data, improving monitoring efficiency.

[0056] Obtain the preset adjustment instructions, and perform a digital twin simulation of the construction process by combining the proposed adjustment instructions, the refined monitoring data, the real-time measured parameters of concrete materials, and the current environmental boundary conditions, and generate a simulation report.

[0057] Optionally, the generation of the simulation report includes:

[0058] Based on the proposed adjustment command, the refined monitoring data, the measured parameters of the concrete material, and the current environmental boundary conditions, a multiphysics coupling simulation is performed to generate preliminary simulation results.

[0059] Based on the preliminary simulation results, data post-processing is performed to extract geometric shape prediction data and potential internal defect risk prediction data;

[0060] The geometric shape prediction data is integrated with the internal defect risk prediction data to form a simulation report.

[0061] Specifically, multiphysics coupling simulation is performed first. In this phase, the core objective is to simulate how the formwork system responds and deforms after executing the proposed adjustment commands, and under current environmental and material conditions. Input proposed adjustment commands include, for example, a specific adjustment amount for the hydraulic outriggers, such as raising or lowering it by 5 mm, or the activation scheme of the vibrator. Detailed monitoring data provides the precise geometric state and stress distribution of the current formwork system, providing initial conditions and boundary constraints for the simulation model. Measured parameters of the concrete material, such as the initial slump (180-220 mm), setting time, and early strength development curve, are crucial for accurately simulating the fluid-to-solid transition of concrete and its interaction with the formwork. Current environmental boundary conditions include real-time temperature, humidity, and wind speed, which affect the thermal deformation of the formwork and the curing process of the concrete. During the simulation, for example, the fluid-structure interaction module simulates the flow and filling of the concrete slurry, as well as its impact and static load on the formwork; the thermodynamic coupling module simulates the expansion and contraction of the formwork material and the heat of hydration of the concrete caused by changes in ambient temperature. The output of this stage is the raw simulation data, covering time-series displacement, stress, and temperature fields. Next, post-processing is performed on these preliminary simulation results to extract the most critical information for construction quality management. The raw simulation data is typically massive amounts of 3D mesh data and time-series data, requiring processing to transform it into meaningful indicators. Data post-processing includes, but is not limited to, mesh data interpolation, downsampling, and extraction of time history data for specific points of interest. For example, from the displacement field data at the time step, the final geometric position and maximum deformation of key formwork points, such as the bottom of beams and the top of side forms, can be extracted throughout the construction process; this is the geometric morphology prediction data. Simultaneously, by analyzing the stress distribution and temperature gradient within the concrete in the simulation model, combined with the early strength and cracking criteria of the concrete, potential internal defect risks can be predicted. For example, excessive tensile stress areas may indicate cracking risk, while locally excessive heat of hydration may lead to thermal stress cracking. These risk predictions are presented as specific numerical values, such as stress values ​​or crack indices, forming potential internal defect risk prediction data. Finally, the obtained geometric morphology prediction data and potential internal defect risk prediction data are integrated to ultimately form a simulation report. The simulation report not only includes a digitized list of data, such as the predicted geometric deviation values ​​for key points (e.g., a vertical deviation of 1.5 mm relative to the design position), but also an intuitive 3D visualization. Figure 3As shown, the curved surface in the figure represents the geometric shape prediction data. It displays the final predicted deformation shape of the template surface after the proposed adjustment instructions are executed, presented in high fidelity as a 3D cloud map. The points marked with 'x' represent potential internal defect risk prediction data, revealing the locations of possible defects such as honeycombing and voids within the concrete structure. The locations marked by the black spheres in the figure indicate the locations of the maximum prediction deviation. These images clearly show the final geometry achieved by the template after adjustment, as well as the location and severity of potential defects within the concrete, such as high-stress areas or potential crack locations marked with different symbols. This simulation report is a key input for subsequent decision analysis, providing comprehensive digital evidence for evaluating the effectiveness and potential risks of the proposed adjustment instructions.

[0062] For example, based on refined monitoring data, the deformation of the left flange of the formwork was 7.74 mm, approaching the warning threshold of 8 mm. The on-site engineer input an adjustment command: "Lift the No. 2 hydraulic outrigger under the left flange of the formwork upwards by 2 mm." Upon receiving this command, a multiphysics coupling simulation was performed. The simulation model used the 7.74 mm deformation state described by the refined monitoring data as the initial geometric conditions. Simultaneously, measured parameters of the concrete material were input, such as a slump of 200 mm and an initial setting time of 6 hours. The current environmental boundary conditions were a temperature of 35 degrees Celsius and no wind. After the simulation started, the fluid-structure interaction module simulated the impact and static pressure on the formwork during concrete pouring, while the thermodynamic coupling module calculated the thermal deformation of the formwork caused by the ambient temperature and the heat of hydration of the concrete. After approximately 15 minutes of calculation, preliminary simulation results were generated, including the displacement, stress, and temperature changes of each node over the next few hours. Next, these massive amounts of raw data were post-processed. By extracting the final displacement field from the simulation time series, geometric shape prediction data was obtained. The results showed that after a 2 mm jacking, the final residual deformation in this area would be controlled to +1.5 mm. Simultaneously, by analyzing the stress field and temperature gradient within the concrete, and combining this with an early-stage concrete strength development model, a slight risk of surface microcracks was identified at a corner of the formwork-concrete interface due to localized stress concentration. This risk probability was quantified as 15%, constituting the potential internal defect risk prediction data. Finally, the geometric shape prediction data (final deformation +1.5 mm) and the internal defect risk prediction data (15% microcrack risk) were integrated and accompanied by a 3D visualization cloud map to form a complete simulation report. This decision-making model reduces the risk of construction rework and quality accidents caused by blind adjustments or erroneous operations.

[0063] Based on the simulation report, multi-objective optimization decision analysis is performed to generate adjustment instructions, and the execution mechanism is controlled to execute the adjustment instructions to collect actual response data of the template.

[0064] Optionally, the acquisition of the actual response data of the template includes:

[0065] The geometric shape prediction data in the simulation report is coupled and compared with the internal defect risk prediction data in the simulation report to identify adjustable variables and constraint boundaries in the construction process and generate multiple sets of candidate adjustment parameters.

[0066] Based on the multiple sets of candidate adjustment parameters, a multi-source feedback comparison is performed to select an adjustment scheme that simultaneously meets the requirements of morphological accuracy and structural density as the adjustment instruction.

[0067] The adjustment command is issued, and response data of formwork strain, displacement, and concrete flow state are collected in real time as the actual response data of the formwork.

[0068] Specifically, the geometric shape prediction data and internal defect risk prediction data in the simulation report are first coupled and compared. Geometric shape prediction data is typically presented as a 3D model or deviation field, indicating the shape of the formwork surface under the proposed adjustment command. Internal defect risk prediction data displays potential defect areas within the concrete and their likelihood of occurrence in the form of a heat map or probability distribution. By overlaying and analyzing both, key parameters that significantly affect both morphological accuracy and internal quality can be identified. For example, if the simulation report shows a large geometric deviation in a local area, accompanied by a high risk of internal honeycombing, it indicates that the support point adjustment parameters for that area are key adjustable variables. These adjustable variables include, but are not limited to, the displacement of hydraulic outriggers, fine-tuning of the formwork angle, and the frequency or duration of local vibrators. Simultaneously, considering factors such as the maximum stroke of the equipment, load-bearing capacity, material physical properties, and safety regulations, upper and lower limits for these adjustable variables are defined, forming multiple sets of candidate adjustment parameters. Subsequently, multi-source feedback comparison is performed based on these multiple sets of candidate adjustment parameters. Here, "multi-source feedback" refers to the prediction data on geometric shape accuracy and structural density provided in the simulation report, which are two main optimization objectives. For each candidate combination of adjustment parameters, the generated geometric shape prediction data is evaluated to ensure it meets preset geometric accuracy requirements, such as template surface deviation within ±3 mm. Simultaneously, the internal defect risk prediction data is also evaluated to ensure the internal concrete density meets design requirements, such as internal porosity less than 1% or no significant crack risk. The adjustment schemes selected through this comparison process... ,have:

[0069] ;

[0070] in, Candidate adjustment parameter vectors, such as hydraulic outrigger displacement and vibrator frequency; For feasible parameter space; This refers to the geometric deviation under parameter x, such as the deviation between the template surface and the design position; This represents the probability of internal defect risk under parameter x, such as honeycomb, voids, cracks, etc. For geometric parameters such as displacement and angle under parameter x; , These are the maximum and minimum allowable values ​​for the geometric parameters in engineering, respectively. For internal defect parameters such as stress and temperature gradient under parameter x; , These are the maximum and minimum allowable values ​​for the internal defect parameters in engineering, respectively. The values ​​of the independent variables that minimize the objective function. From all candidate parameters that satisfy the constraints, choose the one that minimizes the objective function. and Simultaneously, the minimum adjustment scheme is selected. Ultimately, the adjustment schemes that simultaneously meet the following requirements—such as critical point accuracy of ±2mm, surface flatness conforming to industry Class A standards, structural density without major honeycombing or voids, and concrete compressive strength exceeding 95% of the design requirements—are determined as adjustment instructions. For example, an instruction might include: "Raise the hydraulic cylinder of the right support point of the formwork by 1.5 mm, and simultaneously activate the central vibrator for 30 seconds, setting the vibration frequency to 80Hz." Then, through the industrial control interface or on-site operation management system, the adjustment instructions are issued to control the actuators, such as hydraulic cylinders, electric actuators, or vibrators, to execute these specific adjustment actions. During the adjustment process, formwork strain and displacement data are collected in real time. This data is typically acquired using strain gauges and laser displacement sensors, with a sampling frequency of 5-10 Hz. Simultaneously, to monitor the internal quality and flow state of the concrete, embedded wireless sensors such as temperature, humidity, and vibration sensors, or microwave detectors, are used to monitor the concrete's fluidity, setting and hardening rate, and internal density in real time. These real-time feedback data constitute the actual response data of the template.

[0071] For example, upon receiving the simulation report, a multi-objective optimization decision analysis is initiated. First, the simulation report is coupled and compared. The report shows that while a jacking amount of 2 mm can effectively control the geometry, it introduces a 15% risk of microcracks. The jacking amount of the hydraulic outrigger in this area is identified as a key adjustable variable, with a constraint boundary of 0 to 5 mm. Subsequently, within the range of 0 to 3 mm, 11 sets of candidate adjustment parameters are generated in 0.1 mm increments, including "jacking 1.5 mm", "jacking 1.6 mm", ... "jacking 2.5 mm". Then, rapid digital twin simulations are performed on each of these 11 sets of candidate adjustment parameters, followed by multi-source feedback comparison. The goal of the comparison is to simultaneously meet the requirements of morphological accuracy (e.g., the absolute value of the final deviation is less than 2 mm) and structural density (e.g., the probability of internal defect risk is less than 5%). After comparison, it was found that when the jacking amount is 1.8 mm, the predicted final geometric deviation is +1.9 mm, and the probability of internal defect risk is 4.5%. This solution satisfies both objectives and achieves optimal overall performance. Therefore, the final adjustment command was determined to be "lifting the No. 2 hydraulic outrigger below the left flange of the formwork upwards by 1.8 mm." This command was automatically transmitted to the PLC controller via the industrial bus, controlling the servo valve of the No. 2 hydraulic outrigger to precisely execute the 1.8 mm lifting action. Simultaneously, a wireless temperature sensor embedded in the concrete transmitted the hydration heat rise curve. This real-time collected data was integrated and used as the actual response data storage for the formwork. This closed-loop control based on multi-objective optimization improves the scientific nature of construction decisions and the final project quality.

[0072] Based on the comparison between the actual response data of the template and the simulation report, model calibration data is generated, and the initial quality sensitivity model is updated using the model calibration data.

[0073] Optionally, the generated model calibration data includes:

[0074] Spatial registration and comparison are performed between the actual geometric data in the actual response data of the template and the geometric shape prediction data to generate geometric deviation data;

[0075] The actual mechanical response data in the actual response data of the template is correlated with the internal defect risk prediction data in the simulation report to generate risk prediction deviation data.

[0076] The geometric deviation data and the risk prediction deviation data are fused together to form model calibration data.

[0077] Specifically, the first step is to spatially register and compare the actual geometric data in the actual response data of the template with the geometric shape prediction data in the simulation report to generate geometric deviation data. The actual geometric data is typically obtained through high-precision on-site measuring equipment, such as a 3D laser scanner or total station, to acquire the true 3D shape data of the template or poured concrete. This data may be in point cloud form or a mesh model. Before comparison, these two sets of data must be accurately spatially registered to ensure they are aligned in the same coordinate system. The registration algorithm often uses the Iterative Closest Point (ICP) method, which converts the geometric shape prediction data in the simulation report into a 3D spatial reference point set. Then, using the measured point cloud data as the target surface, the optimal spatial transformation matrix between the two is found through least-squares iterative calculation, minimizing the sum of squared Euclidean distances between the predicted point set and the measured point cloud. After registration, geometric deviation data is generated by calculating the deviation between the predicted and actual geometric shapes at each corresponding point or region. The geometric deviation data quantifies the degree of discrepancy between the prediction and the actual shape, size, and location; for example, a deviation of +3.5 mm or -2.1 mm between the predicted and actual values ​​at a certain point. For calculating this geometric deviation data ,have:

[0078] ;

[0079] in, These are the coordinates of a point in three-dimensional space. The actual geometric position vector is obtained from field measurement data; To predict the geometric position vector, data is obtained from the simulation report. Next, a correlation analysis is performed between the actual mechanical response data from the actual formwork response data and the internal defect risk prediction data from the simulation report to generate risk prediction deviation data. The actual mechanical response data includes real-time acquired formwork strain and displacement data, as well as concrete internal condition monitoring data such as temperature and vibration. This data reflects the actual curing process and internal quality of the concrete. The internal defect risk prediction data consists of the predicted locations and probability distributions of potential defects such as honeycomb, pores, and cracks within the concrete, as reported in the simulation report. For example, if the simulation report predicts a high risk of honeycomb in a certain area, and the actual acquired ultrasonic pulse velocity in that area is significantly lower than normal, indicating poor compaction, then the prediction is relatively accurate. Conversely, if the prediction is significantly inconsistent with reality, a large risk prediction deviation will occur. Correlation analysis typically quantifies the model's accuracy in predicting internal quality by comparing the predicted risk indicators with actually observed quality indicators such as the presence and severity of defects. The risk prediction deviation data characterizes the inconsistency between the predicted probability of internal defect occurrence and the actual observed severity of defects. Finally, the generated geometric deviation data and risk prediction deviation data are fused to form the final model calibration data. For example, geometric deviation data may be stored as a spatial deviation field, while risk prediction deviation data may be stored as risk factor adjustments for a specific region. The fusion process typically involves aligning and integrating these two types of deviation data in both temporal and spatial dimensions to form a structured dataset. This dataset will clearly indicate in which spatial locations, for which physical quantities, there are significant differences between the simulation report and the actual results, as well as the magnitude and nature of these differences. For example, the calibration data might contain entries such as "Geometric upward deviation of region A is 2.5 mm, and the predicted cellular risk is 10% lower than the actual risk."

[0080] For example, after executing the 1.8 mm jacking command and completing the concrete pouring and curing of that segment, a 3D laser scanner was used to scan the left flange area of ​​the hardened concrete box girder to obtain the actual geometric point cloud data, which is the actual geometric data. This point cloud data was then spatially registered using the Iterative Closest Point (ICP) algorithm with the geometric data predicted in the simulation report, i.e., a prediction deviation of +1.9 mm. After registration, at the center point q of the flange, the data was compared using the geometric deviation calculation formula. The calculation showed that the actual geometric deviation vector at this point had a vertical component of +2.6 mm, which differed from the predicted +1.9 mm by 0.7 mm. This difference is the geometric deviation data for that point. Secondly, ultrasonic pulse detection of the hardened concrete revealed that the wave velocity in this area was normal, and the wave velocity decrease corresponding to the 4.5% microcrack risk predicted in the simulation report did not occur, meaning that no defects actually occurred. This result differed from the internal defect risk prediction data in the simulation report. Through correlation analysis, risk prediction deviation data for this area was generated and quantified as "risk overestimation 100%". Finally, the geometric deviation data (deviations of 0.7 mm at various points, etc.) was spatially aligned and merged with the risk prediction deviation data (risk overestimation 100%) to form a structured dataset, such as "left flange center point, geometric prediction underestimation 0.7 mm, microcrack risk prediction overestimation." This serves as the model calibration data for the next model update. By quantitatively comparing the precise measured results with the model's predictions, it is possible to accurately identify in which aspects and to what extent the current model deviates from reality.

[0081] Optionally, updating the initial quality sensitivity model using the model calibration data includes:

[0082] Based on the geometric deviation data, the sensitivity weight coefficients in the initial quality sensitivity model are adjusted to generate updated sensitivity weight coefficients.

[0083] Based on the risk prediction deviation data, the threshold parameters related to defect sensitivity in the initial quality sensitivity model are corrected to generate updated risk warning thresholds;

[0084] The updated sensitivity weight coefficients and the updated risk warning thresholds are updated synchronously to generate a corrected quality sensitivity model.

[0085] Specifically, geometric deviation data reflects the difference between the simulated predicted geometry and the actual geometry. When persistent, systematic geometric deviations are observed, it indicates that certain parameters in the model, particularly the sensitivity weight coefficients that determine the template deformation behavior, may not reflect reality. For example, if the model underestimates the impact of a particular support point on the overall deformation, causing the predicted displacement to consistently be less than the actual displacement, then the sensitivity weight coefficient for that support point in the mass sensitivity model needs to be increased accordingly. This requires calculating updated sensitivity weight coefficients. ,have:

[0086] ;

[0087] in, These are the sensitivity weight coefficients of the quality sensitivity model before the update; The learning rate typically ranges from 0.01 to 0.1. Number of key points; Let be the Euclidean norm of the geometric deviation data at the i-th point; The absolute value of the allowable deviation at that point is used for normalization. In this way, the model gradually "learns" and adapts to actual deformation patterns, making predictions of future geometric shapes more accurate. Secondly, based on the risk prediction deviation data, the threshold parameters related to defect sensitivity in the initial quality sensitivity model are corrected to generate updated risk warning thresholds. Risk prediction deviation data quantifies the gap between the internal defect risk predicted in the simulation report and the actual observed defect severity. If the model frequently "misses" actual internal defects or "falsely reports" defects that have not occurred, its internal defect risk warning threshold needs to be adjusted. For example, when actual ultrasonic pulse velocity monitoring results show that the density of a certain area does not meet the standard, but the internal defect risk prediction in the simulation report is very low, it indicates that the model's sensitivity to this type of risk is insufficient, and the corresponding warning threshold needs to be lowered. This allows the model to issue risk warnings earlier or more accurately under the same input conditions. The updated risk warning threshold is then calculated. ,have:

[0088] ;

[0089] in, The risk warning threshold was the one before the update. This is a correction factor, and its value typically ranges from 0.1 to 0.5. To quantify the risk prediction deviation, a quantitative value is obtained by comparing actual monitoring indicators such as ultrasonic pulse velocity in key areas inside the concrete with the risk probabilities predicted by the simulation model, using statistical root mean square error. Finally, the updated sensitivity weighting coefficients, such as the adjusted set of weighting coefficients, and the updated risk warning thresholds, such as more precise threshold ranges for different defect types, are updated synchronously to generate a corrected quality sensitivity model. This model includes more accurate weighting coefficients and risk warning thresholds that better reflect reality.

[0090] For example, calibration data shows that in the left flange region, the model's geometric predictions are too low, and the risk prediction for microcracks is overly conservative. First, the sensitivity weighting coefficients are adjusted based on the geometric deviation data. In this region, the norm of the geometric deviation data is 0.7 mm, the allowable deviation for this point is 5 mm, the original load-related sensitivity weighting coefficient for this point is 0.625, the learning rate is 0.1, and the number of keypoints is 1. Therefore, the updated sensitivity weighting coefficients are calculated. Secondly, the risk warning threshold is adjusted based on the risk prediction deviation data. Since the model overestimates the risk of microcracks, the corresponding warning threshold needs to be increased to make it less likely to trigger an alarm. If the original stress warning threshold related to microcracks was 5 MPa, the risk prediction deviation ∆R is quantified as a correction factor, here a positive value such as +0.2 MPa. Let the correction coefficient be 0.5. According to the update formula... (MPa). This means that the model will only issue a microcrack risk warning when the predicted stress exceeds 5.1 MPa. Finally, based on the updated data, synchronous updates are performed to generate a more realistic corrected quality sensitivity model. By utilizing feedback data from actual construction, the long-term stability and reliability of the entire intelligent control system are ensured, enabling it to continuously provide high-quality decision support for the project.

[0091] Optionally, the method further includes:

[0092] The geometric deviation data of the preset monitoring points are continuously monitored. When it is found that the geometric deviation data of the monitoring points continuously exceeds the preset deviation threshold in multiple consecutive construction cycles, the sensitivity weight coefficient of the monitoring points in the modified quality sensitivity model is increased.

[0093] Specifically, the corrected quality sensitivity model continuously monitors the geometric deviation data generated in the previous steps. This data comes from various preset key monitoring points, such as geometric key dimension measurement points of the formwork, support points, and corner points of areas prone to deformation. The geometric deviation data of each monitoring point is compared in real time with a preset deviation threshold. This deviation threshold is usually set based on engineering design standards, construction specifications, or historical experience. For example, for bridge deck thickness, the deviation threshold might be ±3 mm. When the geometric deviation data of a specific monitoring point is found to continuously exceed the preset deviation threshold in multiple consecutive construction cycles, the sensitivity weighting coefficient enhancement mechanism is triggered. For example, if a point's vertical displacement deviation continuously exceeds 5 mm in the last three pouring cycles, rather than being an occasional single exceedance, this indicates that the monitoring point may have inherent structural weaknesses, installation accuracy issues, or be subjected to unexpected loads. The specific number of "multiple consecutive construction cycles" is a configurable parameter, such as two, three, or more consecutive cycles, designed to avoid frequent adjustments due to occasional errors. Once the above conditions are met, the sensitivity weight coefficient of the monitoring point in the revised quality sensitivity model will be increased. The engineering purpose of this increase is to make any minor anomalies at the monitoring point easier to identify and mark as higher risk during the subsequent generation of the quality sensitivity dynamic map. The increase in the sensitivity weight coefficient is usually gradual, rather than a one-time large adjustment. Regarding the increased sensitivity weight coefficient... ,have:

[0094] ;

[0095] in, Set the preset weight adjustment step size, such as 0.05; This refers to the number of consecutive construction cycles. Let be the Euclidean norm of the geometric deviation data of the monitoring point in the a-th cycle; This is a preset deviation threshold; This is an indicator function that takes the value 1 when the condition is true and 0 otherwise.

[0096] For example, when continuing construction of subsequent segments using the modified quality sensitivity model, monitoring points preset at various locations on the formwork are continuously monitored. At the monitoring point in the middle of the left flange, despite the model being modified, the average final geometric deviation measured over three consecutive construction cycles was 2.5 mm, consistently exceeding the preset deviation threshold of 2 mm. This phenomenon triggered the enhancement mechanism of the sensitivity weight coefficient. Currently, the sensitivity weight coefficient for this point in the modified quality sensitivity model is 0.627. Assume the preset weight adjustment step size is 0.05. According to the enhancement formula, in each cycle a, since the deviation of 2.5 mm is greater than the threshold of 2 mm, its value is always 1; therefore, the enhanced sensitivity weight coefficient... .

[0097] By increasing the sensitivity weight of these points, it is ensured that potential, systemic structural defects or safety hazards are not overlooked, thereby enhancing the risk identification capability and long-term safety of the construction process.

[0098] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a data-driven high-speed railway bridge-building machine formwork construction quality management system, the system comprising:

[0099] The data modeling module is used to acquire multi-source real-time status data, environmental monitoring data and historical adjustment records of the template system, and to calculate based on the preset quality sensitivity initial model to generate a dynamic quality sensitivity map of the current construction stage.

[0100] The perception and scheduling module is used to dynamically schedule perception resources based on the quality sensitivity dynamic graph to obtain refined monitoring data.

[0101] The twin simulation module is used to obtain preset adjustment instructions, and to perform digital twin simulation of the construction process with the proposed adjustment instructions, the refined monitoring data, the real-time measured parameters of concrete materials, and the current environmental boundary conditions to generate a simulation report;

[0102] The optimization control module is used to perform multi-objective optimization decision analysis based on the simulation report, generate adjustment instructions, control the execution mechanism to execute the adjustment instructions, and collect actual response data of the template.

[0103] The calibration update module is used to generate model calibration data based on the comparison results between the actual response data of the template and the simulation report, and to update the initial quality sensitivity model using the model calibration data.

[0104] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0105] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A data-driven method for quality management of formwork construction in high-speed railway bridge building machines, characterized in that, The method includes: The process involves acquiring multi-source real-time status data, environmental monitoring data, and historical adjustment records of the formwork system to obtain an initial quality sensitivity model. This initial model acquisition includes: collecting historical geometric deformation data, hydraulic jacking response data, and corresponding environmental temperature and humidity fluctuation ranges of the formwork system at different construction stages; constructing a deformation feature library of the formwork under different load and environmental coupling conditions; performing cluster analysis on the data in the deformation feature library to generate sensitivity weight coefficients for each monitoring point to the disturbance source; spatially mapping the sensitivity weight coefficients to the structural partitions of the formwork system and normalizing the sensitivity distribution at different construction stages to generate the initial quality sensitivity model; and performing further analysis based on the initial quality sensitivity model. The process involves calculating and generating a dynamic quality sensitivity map for the current construction stage. This generation includes: collecting geometric deformation data, hydraulic jacking pressure data, and vibration frequency data of the formwork system as multi-source real-time status data; simultaneously acquiring ambient temperature, humidity, wind speed, and vibration data from nearby construction sites as environmental monitoring data; and retrieving historical adjustment frequency and amplitude records for the same construction location. The multi-source real-time status data, environmental monitoring data, and historical adjustment records are then input into the initial quality sensitivity model to calculate the quality deviation risk coefficient for each monitoring point. Based on the quality deviation risk coefficient, the risk level of each area of ​​the formwork is classified, and dynamic color rendering is performed to generate the dynamic quality sensitivity map. Dynamic sensing resource scheduling is performed based on the quality sensitivity dynamic map to obtain refined monitoring data; wherein obtaining refined monitoring data includes: analyzing the risk distribution based on the quality sensitivity dynamic map to determine risk areas and non-sensitive areas; deploying mobile sensors to detect risk areas to obtain local detection data, and simultaneously reducing the sampling frequency of fixed sensors in non-sensitive areas to obtain fixed sensor data; and performing feature-level fusion of the local detection data and the fixed sensor data to generate refined monitoring data; Obtain the preset adjustment instructions, and perform a digital twin simulation of the construction process by combining the proposed adjustment instructions, the refined monitoring data, the real-time measured parameters of concrete materials, and the current environmental boundary conditions, and generate a simulation report. Based on the simulation report, multi-objective optimization decision analysis is performed to generate adjustment instructions, and the execution mechanism is controlled to execute the adjustment instructions to collect actual response data of the template. Based on the comparison between the actual response data of the template and the simulation report, model calibration data is generated, and the initial quality sensitivity model is updated using the model calibration data.

2. The data-driven high-speed railway bridge-building machine formwork construction quality management method according to claim 1, characterized in that, The generated simulation report includes: Based on the proposed adjustment command, the refined monitoring data, the measured parameters of the concrete material, and the current environmental boundary conditions, a multiphysics coupling simulation is performed to generate preliminary simulation results. Based on the preliminary simulation results, data post-processing is performed to extract geometric shape prediction data and potential internal defect risk prediction data; The geometric shape prediction data is integrated with the internal defect risk prediction data to form a simulation report.

3. The data-driven high-speed railway bridge-building machine formwork construction quality management method according to claim 2, characterized in that, The actual response data of the template obtained by the collection includes: The geometric shape prediction data in the simulation report is coupled and compared with the internal defect risk prediction data in the simulation report to identify adjustable variables and constraint boundaries in the construction process and generate multiple sets of candidate adjustment parameters. Based on the multiple sets of candidate adjustment parameters, a multi-source feedback comparison is performed to select an adjustment scheme that simultaneously meets the requirements of morphological accuracy and structural density as the adjustment instruction. The adjustment command is issued, and response data of formwork strain, displacement, and concrete flow state are collected in real time as the actual response data of the formwork.

4. The data-driven high-speed railway bridge-building machine formwork construction quality management method according to claim 3, characterized in that, The generated model calibration data includes: Spatial registration and comparison are performed between the actual geometric data in the actual response data of the template and the geometric shape prediction data to generate geometric deviation data; The actual mechanical response data in the actual response data of the template is correlated with the internal defect risk prediction data in the simulation report to generate risk prediction deviation data. The geometric deviation data and the risk prediction deviation data are fused together to form model calibration data.

5. The data-driven high-speed railway bridge-building machine formwork construction quality management method according to claim 4, characterized in that, The step of updating the initial quality sensitivity model using the model calibration data includes: Based on the geometric deviation data, the sensitivity weight coefficients in the initial quality sensitivity model are adjusted to generate updated sensitivity weight coefficients. Based on the risk prediction deviation data, the threshold parameters related to defect sensitivity in the initial quality sensitivity model are corrected to generate updated risk warning thresholds; The updated sensitivity weight coefficients and the updated risk warning thresholds are updated synchronously to generate a corrected quality sensitivity model.

6. The data-driven high-speed railway bridge-building machine formwork construction quality management method according to claim 5, characterized in that, The method further includes: The geometric deviation data of the preset monitoring points are continuously monitored. When it is found that the geometric deviation data of the monitoring points continuously exceeds the preset deviation threshold in multiple consecutive construction cycles, the sensitivity weight coefficient of the monitoring points in the modified quality sensitivity model is increased.

7. A data-driven high-speed railway bridge-building machine formwork construction quality management system, applied to the data-driven high-speed railway bridge-building machine formwork construction quality management method as described in any one of claims 1-6, characterized in that, The system includes: The data modeling module is used to acquire multi-source real-time status data, environmental monitoring data, and historical adjustment records of the template system to obtain an initial quality sensitivity model. This initial quality sensitivity model acquisition includes: collecting historical geometric deformation data, hydraulic jacking response data, and corresponding environmental temperature and humidity fluctuation ranges of the template system at different construction stages; constructing a deformation feature library of the template under different load and environmental coupling conditions; performing cluster analysis on the data in the deformation feature library to generate sensitivity weight coefficients for each monitoring point to the disturbance source; spatially mapping the sensitivity weight coefficients to the structural partitions of the template system and normalizing the sensitivity distribution at different construction stages to generate an initial quality sensitivity model; and based on the initial quality sensitivity model... The initial model is used to calculate and generate a dynamic quality sensitivity map for the current construction stage. This generation includes: collecting geometric deformation data, hydraulic jacking pressure data, and vibration frequency data of the formwork system as multi-source real-time status data; simultaneously acquiring ambient temperature, humidity, wind speed, and vibration data from nearby construction sites as environmental monitoring data; and retrieving historical adjustment frequency and amplitude records for the same construction location. The multi-source real-time status data, environmental monitoring data, and historical adjustment records are input into the initial quality sensitivity model to calculate the quality deviation risk coefficient for each monitoring point. Based on the quality deviation risk coefficient, the risk level of each area of ​​the formwork is classified, and dynamic color rendering is performed to generate the dynamic quality sensitivity map. The perception scheduling module is used to dynamically schedule perception resources based on the quality sensitivity dynamic map to obtain refined monitoring data. The process of obtaining refined monitoring data includes: analyzing the risk distribution in the quality sensitivity dynamic map to determine risk areas and non-sensitive areas; deploying mobile sensors to detect risk areas to obtain local detection data, and simultaneously reducing the sampling frequency of fixed sensors in non-sensitive areas to obtain fixed sensor data; and performing feature-level fusion of the local detection data and the fixed sensor data to generate refined monitoring data. The twin simulation module is used to obtain preset adjustment instructions, and to perform digital twin simulation of the construction process with the proposed adjustment instructions, the refined monitoring data, the real-time measured parameters of concrete materials, and the current environmental boundary conditions to generate a simulation report; The optimization control module is used to perform multi-objective optimization decision analysis based on the simulation report, generate adjustment instructions, control the execution mechanism to execute the adjustment instructions, and collect actual response data of the template. The calibration update module is used to generate model calibration data based on the comparison results between the actual response data of the template and the simulation report, and to update the initial quality sensitivity model using the model calibration data.

Images