Hanging basket cantilever casting type bridge pre-camber prediction method based on machine learning

Abstract

The invention provides a hanging basket cantilever casting type bridge pre-camber prediction method based on machine learning, and relates to the technical field of bridge pre-camber prediction.The method comprises the steps that bridge structure design parameters, construction process monitoring data and material characteristic data are collected; beam structure design parameters, construction process monitoring data and material characteristic data are cleaned and preprocessed; a Gaussian process regression model is constructed, a rational quadratic kernel function is selected, kernel function hyper-parameters are adjusted through a Bayesian optimization algorithm, and the model is trained to establish a nonlinear mapping relation between input variables and pre-camber; inputting real-time construction data, predicting a pre-camber value and a confidence interval through the trained model, and evaluating prediction precision; model hyper-parameters are dynamically updated based on actually measured data, a prediction-actually measured-correction closed-loop feedback mechanism is formed, and a subsequent prediction result is optimized; according to the method, an efficient and high-quality solution is provided for predicting the pre-camber of the hanging basket cantilever casting construction bridge.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge pre-camber prediction, and in particular to a method for predicting the pre-camber of a cantilever-cast bridge based on machine learning. Background Art

[0002] In recent years, with the rapid growth of my country's economy, the country's infrastructure has continued to develop. As a common bridge construction method, cantilever casting with a hanging basket is suitable for complex terrain, does not interrupt traffic under the bridge, and is low-cost. It is currently one of the commonly used construction methods for bridge construction in my country.

[0003] Bridges constructed using a cantilevered construction basket often interact with each other during construction, necessitating the design of pre-camber values ​​for each segment's main beam. This discrepancy between design parameters and actual construction site parameters can lead to discrepancies between design and actual construction. Cantilevered construction can lead to errors in pre-camber settings due to factors such as temperature fluctuations during construction, uncertainties in construction loads, prestressing errors, and concrete shrinkage and creep.

[0004] Improper setting of bridge pre-camber can lead to problems controlling beam deformation. Segmental dimensional errors and uneven contact surfaces between segments can cause longitudinal unevenness in the bridge deck, impacting smoothness. Deviations in the relative position and elevation of the beams at both ends of the closure section can make closing the bridge difficult. Furthermore, improper pre-camber setting can lead to excessive deflection or camber.

[0005] Bridge pre-camber is a crucial parameter in bridge construction. Traditional pre-camber prediction methods include: the finite element method (FEM), which uses finite element software to discretize structural elements and construct refined finite element models to predict bridge pre-camber; the empirical formula method, which uses empirical formulas based on extensive engineering practice and experimental research and is provided in bridge design codes worldwide to estimate bridge pre-camber; the field monitoring method, which uses sensors placed on beams to monitor strain and displacement changes at various construction stages in real time and predicts pre-camber based on measured data combined with structural analysis methods; and the model testing method, which uses scaled models of bridge structures with a certain similarity ratio to simulate actual construction processes and load conditions, and then measures the deformation of the models to predict the actual bridge pre-camber. However, these methods have limitations. The FE method is cumbersome to build models and requires significant computational resources and time; the empirical formula method has a limited scope of application and cannot account for specific project characteristics; the field monitoring method is limited by sensor accuracy and durability issues; and the model testing method suffers from uncertainty in the results. Summary of the Invention

[0006] In view of the above technical problems, the technical solution adopted by the present invention is:

[0007] According to a machine learning-based method for predicting the pre-camber of a cantilever cast-in-place bridge provided in this application, the method comprises the following steps:

[0008] S100, collecting bridge structure design parameters, construction process monitoring data, and material property data, selecting cantilever length L, beam height H, web thickness F, bottom plate thickness D, concrete bulk density γ, concrete elastic modulus E, hanging basket load G, strength P, and cable tension T as input variables, and pre-camber value as output variable;

[0009] S200: Cleaning and preprocessing of beam structure design parameters, construction process monitoring data, and material property data, including outlier correction, missing value interpolation, and standardization, and dividing into training and test sets;

[0010] S300, constructing a Gaussian process regression model, selecting a rational quadratic kernel function, adjusting the kernel function hyperparameters through the Bayesian optimization algorithm, and training the model to establish a nonlinear mapping relationship between input variables and pre-camber;

[0011] S400: Input real-time construction data, predict the pre-camber value and confidence interval through the trained model, and calculate the root mean square error, mean absolute error, and determination coefficient to evaluate the prediction accuracy;

[0012] S500 dynamically updates model hyperparameters based on measured data, forming a closed-loop feedback mechanism of prediction-measurement-correction to optimize subsequent prediction results.

[0013] Furthermore, the bridge structure design parameters include: cantilever length L, beam height H, web thickness F and bottom plate thickness D;

[0014] The construction process monitoring data includes: concrete bulk density γ, concrete elastic modulus E and strength P;

[0015] The material property data include: basket load G and cable force T.

[0016] Furthermore, the standardization process in S200 uses the Z-score normalization method, the formula is: X norm =(x-μ) / σ; where x is the original data, X norm is the corresponding normalized data, μ is the mean, and σ is the standard deviation.

[0017] Furthermore, the rational quadratic kernel function expression in S300 is:

[0018]

[0019] Among them, x i and x jis the input data point; ||x i -x j || is the Euclidean distance between two points; l is the length scale parameter, and α is the scale mixing parameter.

[0020] Furthermore, the Bayesian optimization in S300 is implemented by maximizing the marginal likelihood function.

[0021] Furthermore, the confidence interval prediction formula in S400 is:

[0022]

[0023]

[0024] Where K is the covariance matrix of the training data, σ 2 n Observation noise variance; x * For a given new input, μ(x * ) is the predicted value and σ 2 (x * ) is the variance, I is the identity matrix, and X is the input feature matrix of the training dataset.

[0025] Furthermore, the accuracy evaluation in S400 includes:

[0026] Draw a comparison chart of predicted values ​​and true values, and a residual chart;

[0027] Calculate RMSE and MAE to quantify the prediction error, R 2 To assess the goodness of model fit.

[0028] Furthermore, the dynamic updating of model hyperparameters based on measured data includes:

[0029] The model is calibrated using measured data from a preset number of segments in front of the bridge, and the model hyperparameters are updated.

[0030] The present invention has at least the following beneficial effects:

[0031] The machine learning-based method for predicting the pre-camber of a cantilever-cast bridge has the following beneficial effects:

[0032] 1. The present invention accurately captures the relationship between complex factors. By flexibly selecting kernel functions, it can automatically fit the mapping between input variables and pre-camber, avoiding the difficulty of empirical formulas or linear regression models in describing the coupling relationship between complex factors.

[0033] 2. The present invention has data-driven dynamic adaptability and can use data from real-time monitoring during the construction process to continuously optimize the model, forming a closed loop of "prediction-measurement-correction". It can adjust subsequent predictions based on the measured pre-camber of the previous beam sections, gradually reducing errors. It can automatically mine data features through hyperparameter optimization, reducing dependence on engineer experience.

[0034] 3. The present invention has the ability to quantify prediction uncertainty. It can not only provide predicted values, but also output confidence intervals to help construction parties judge the risk level. In the linear control of the joint section, combined with the confidence interval, a safety margin can be reserved to avoid cost increases caused by excessive conservatism or risk-taking.

[0035] 4. This invention provides an efficient and high-quality solution for predicting the pre-camber of bridges constructed using cantilevered cast-in-place construction. Its technical value lies not only in precise control during the construction phase but also in health monitoring and maintenance throughout the structure's lifecycle, ultimately achieving the safety, economics, and sustainability goals of bridge engineering. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 A flow chart of a method for predicting the pre-camber of a cantilever cast-in-place bridge using machine learning provided in an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of selecting influencing parameters provided by an embodiment of the present invention;

[0039] Figure 3 A schematic diagram of hyperparameter optimization provided by an embodiment of the present invention;

[0040] Figure 4 A schematic diagram of a pre-camber test after training provided by an embodiment of the present invention;

[0041] Figure 5 A schematic diagram of evaluation indicators provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0043] It should be noted that, based on this disclosure, those skilled in the art will appreciate that an aspect described herein can be implemented independently of any other aspect, and that two or more of these aspects can be combined in various ways. For example, any number of the aspects described herein can be used to implement an apparatus and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement such an apparatus and / or practice such a method.

[0044] The following will refer to Figure 1 The flowchart of the method for predicting the pre-camber of a cantilever cast-type bridge with a hanging basket based on machine learning is shown, which introduces a method for predicting the pre-camber of a cantilever cast-type bridge with a hanging basket based on machine learning.

[0045] The machine learning-based method for predicting the pre-camber of a cantilever cast-in-place bridge may include the following steps:

[0046] S100, collect bridge structure design parameters, construction process monitoring data and material property data, select cantilever length L, beam height H, web thickness F, bottom plate thickness D, concrete density γ, concrete elastic modulus E, hanging basket load G, strength P, cable tension T as input variables, and pre-camber value as output variable.

[0047] Furthermore, the bridge structure design parameters include: cantilever length L, beam height H, web thickness F and bottom plate thickness D;

[0048] The construction process monitoring data includes: concrete bulk density γ, concrete elastic modulus E and strength P;

[0049] The material property data include: basket load G and cable force T.

[0050] In this embodiment, the influencing factors are selected from five aspects: the influence of structural deadweight, the influence of structural deadweight, the influence of shrinkage and creep, the influence of hanging basket deformation, the influence of prestressing and the influence of cable tension. Nine parameters, namely, cantilever length L, beam height H, web thickness F, bottom plate thickness D, concrete density γ, concrete elastic modulus E, hanging basket load G, strength P, and cable tension T, are extracted as the influencing factors affecting pre-camber. These nine influencing parameters are used as input variables, and the pre-camber value is used as the output variable. Figure 2 As shown, the nine parameters are input variables and the pre-camber is the output variable.

[0051] Multi-source data fusion ensures the model comprehensively captures influencing factors, avoiding the limitations of traditional methods that rely on a single data source. Input parameters are selected based on the mechanical properties of bridge construction, improving the model's adaptability to engineering scenarios.

[0052] S200, cleans and preprocesses the beam structure design parameters, construction process monitoring data and material property data, including outlier correction, missing value interpolation, and standardization, and divides them into training sets and test sets.

[0053] Extract relevant data, including input variables and corresponding pre-camber output values. Select an appropriate kernel function to describe the correlation between input variables, which determines the model's complexity and generalization ability. The kernel function describes the nonlinear relationship between input and output variables, enabling prediction of unknown data points and quantification of uncertainty.

[0054] In this embodiment, the collected data is organized into a unified format to ensure that each data sample contains complete input variable information and the corresponding pre-curvature output value, and key features that have a greater impact on the pre-curvature are selected from the numerous input variables to reduce redundant information. The data is standardized to eliminate dimensional effects, improve the model convergence speed, and enhance the model generalization ability. The processed data is divided into a certain proportion for training and testing.

[0055] Furthermore, the standardization process in S200 uses the Z-score normalization method, the formula is: X norm =(x-μ) / σ; where x is the original data, X norm is the corresponding normalized data, μ is the mean, and σ is the standard deviation.

[0056] Outlier correction: Identify abnormal data through box plots and correct them with domain knowledge (such as outliers caused by sensor failure).

[0057] Missing value interpolation: Use the K-nearest neighbor (KNN) algorithm to fill missing values ​​and preserve data distribution characteristics.

[0058] Standardization processing: Z-score normalization is used to eliminate dimensional differences.

[0059] Dataset partitioning: Divide the training set and test set into a ratio (e.g. 7:3) to ensure the generalization ability of the model.

[0060] Beneficial effects:

[0061] Improved data quality: Cleaned data is more reliable and reduces the interference of noise on the model.

[0062] Accelerated model convergence: Standardization makes feature distribution consistent and improves training efficiency.

[0063] S300, construct a Gaussian process regression model, select a rational quadratic kernel function, adjust the kernel function hyperparameters through the Bayesian optimization algorithm, and train the model to establish a nonlinear mapping relationship between the input variables and the pre-camber.

[0064] To capture the multi-scale characteristics of the data, the rational quadratic (RQ) kernel function was selected. By maximizing the marginal likelihood function, the kernel function hyperparameters were automatically optimized to adapt the model to the data characteristics. The kernel function hyperparameters were optimized using Bayesian optimization to adjust the model to the data.

[0065] Furthermore, the rational quadratic kernel function expression in S300 is:

[0066]

[0067] Among them, x i and x j is the input data point; ||x i -x j || is the Euclidean distance between two points; l is the length scale parameter, which controls the degree of mixing of different length scales, allowing the kernel function to adapt to feature changes at different scales. α is the scale mixing parameter, which controls the degree of mixing of different length scales, allowing the kernel function to adapt to feature changes at different scales.

[0068] The RQ kernel captures local mutations and global trends, which is better than traditional linear models; Bayesian optimization avoids manual trial and error and improves model accuracy.

[0069] Furthermore, the Bayesian optimization in S300 is implemented by maximizing the marginal likelihood function.

[0070] Modify the machine learning model, use the Gaussian process regression (GPR) algorithm with good predictive performance, select the rational quadratic (RQ) kernel function, extract the data for standardization and partitioning, define the hyperparameters of mean, covariance and likelihood function required for Gaussian process regression, perform Bayesian optimization, and find the hyperparameter combination that minimizes the objective function in the given hyperparameter search space, such as Figure 3 Find the optimal hyperparameters.

[0071] S400, input real-time construction data, predict the pre-camber value and confidence interval through the trained model, and calculate the root mean square error, mean absolute error, and determination coefficient to evaluate the prediction accuracy.

[0072] Furthermore, the confidence interval prediction formula in S400 is:

[0073]

[0074] Where K is the covariance matrix of the training data, σ 2 n Observation noise variance; x * For a given new input, μ(x * ) is the predicted value and σ 2 (x * ) is the variance, I is the identity matrix, and X is the input feature matrix of the training dataset.

[0075] Retrain the model using the optimal hyperparameters, make predictions on the test set, and calculate the 95% confidence interval to obtain Figure 4 The results were used to calculate the root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R 2 ),get Figure 5 For each indicator value, draw a comparison chart between the predicted value and the true value, a prediction interval chart, a relative error histogram and a residual chart.

[0076] It can be seen from the prediction results that the pre-camber results predicted by the method of the present invention are in line with the rules of general hanging basket cantilever casting bridges, and have accurate prediction results, which can provide guidance for bridge construction control, enable timely corrections, identify potential risks in advance, and improve the accuracy of construction control of hanging basket cantilever casting bridges.

[0077] Beneficial effects:

[0078] Uncertainty quantification: Confidence intervals provide risk warnings (e.g., safety margins for joint closures).

[0079] Visual feedback: A comparison chart of predicted values ​​and true values ​​intuitively displays model performance, facilitating engineers’ decision-making.

[0080] S500 dynamically updates model hyperparameters based on measured data, forming a closed-loop feedback mechanism of prediction-measurement-correction to optimize subsequent prediction results.

[0081] Combined with the selected input parameters, construction monitoring is carried out during the bridge construction period, focusing on observing the changes in various parameters of the current construction section, collecting and processing data, and using the model to predict pre-camber, ensuring dynamic adjustment and risk control during bridge construction.

[0082] The pre-camber of each section during construction can be predicted. The model can be calibrated using measured data from the first three to five sections of the bridge, improving subsequent prediction accuracy. This reduces personnel costs and improves efficiency. The method presented in this paper can obtain accurate pre-camber data for bridge pre-camber prediction using a relatively simple method, reducing the workload for designers and constructors and improving efficiency.

[0083] Incremental learning: After completing every 3-5 construction sections, the measured data is added to the training set and the hyperparameters are re-optimized.

[0084] Closed-loop mechanism: Through the "prediction-measurement-correction" cycle iteration, the cumulative error is gradually reduced.

[0085] Beneficial effects:

[0086] Dynamic adaptability: The model is continuously optimized as construction progresses, reducing the cumulative error by more than 40%.

[0087] Reduce manual dependency: Automated updates reduce the workload of engineers for manual parameter adjustment.

[0088] In this embodiment, through data-driven and algorithmic innovation, the pain points of traditional pre-camber prediction methods such as low computational efficiency, poor accuracy, and insufficient dynamic adaptability are systematically solved, providing an efficient and intelligent solution for bridge construction.

[0089] It has at least the following beneficial effects:

[0090] 1. The present invention accurately captures the relationship between complex factors. By flexibly selecting kernel functions, it can automatically fit the mapping between input variables and pre-camber, avoiding the difficulty of empirical formulas or linear regression models in describing the coupling relationship between complex factors.

[0091] 2. The present invention has data-driven dynamic adaptability and can use data from real-time monitoring during the construction process to continuously optimize the model, forming a closed loop of "prediction-measurement-correction". It can adjust subsequent predictions based on the measured pre-camber of the previous beam sections, gradually reducing errors. It can automatically mine data features through hyperparameter optimization, reducing dependence on engineer experience.

[0092] 3. The present invention has the ability to quantify prediction uncertainty. It can not only provide predicted values, but also output confidence intervals to help construction parties judge the risk level. In the linear control of the joint section, combined with the confidence interval, a safety margin can be reserved to avoid cost increases caused by excessive conservatism or risk-taking.

[0093] 4. This invention provides an efficient and high-quality solution for predicting the pre-camber of bridges constructed using cantilevered cast-in-place construction. Its technical value lies not only in precise control during the construction phase but also in health monitoring and maintenance throughout the structure's lifecycle, ultimately achieving the safety, economics, and sustainability goals of bridge engineering.

[0094] The machine learning approach adopted replaces the traditional crude prediction methods that rely on empirical formulas or simple linear regression. It can comprehensively consider the coupling relationship between multiple complex factors such as structural deadweight, shrinkage and creep, hanging basket deformation, and prestressing. Through in-depth data mining and analysis, it obtains an accurate pre-camber prediction model. This facilitates construction teams to predict the pre-camber of bridges with various complex structures and different construction environments, effectively reducing construction risks and subsequent maintenance costs caused by inaccurate predictions, improving the accuracy of pre-camber setting during bridge construction, significantly improving construction efficiency, and providing strong guarantees for the smooth progress of bridge projects.

[0095] Furthermore, although the steps of the method of the present disclosure are described in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in this particular order, or that all steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0096] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0097] The program product may be implemented in any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0098] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0099] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0100] The program code for performing the operations of the present application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0101] An embodiment of the present invention further provides an electronic device including a processor and the aforementioned non-transitory computer-readable storage medium.

[0102] The electronic device is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0103] The electronic device is implemented as a general-purpose computing device. Components of the electronic device may include, but are not limited to, the aforementioned at least one processor, the aforementioned at least one memory, and a bus connecting different system components (including the memory and the processor).

[0104] The memory stores program codes, which can be executed by the processor, so that the processor performs the steps of various embodiments described in this specification.

[0105] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read-only memory (ROM).

[0106] The memory may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0107] The bus may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

[0108] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface. Furthermore, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the electronic device via a bus. It should be understood that, although not shown in the figures, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0109] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiments of the present disclosure.

[0110] An embodiment of the present invention further provides a computer program product comprising program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.

[0111] Although some specific embodiments of the present invention have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention.

Claims

1. A method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning, characterized in that: The method comprises the following steps: S100, collecting bridge structure design parameters, construction process monitoring data, and material property data, selecting cantilever length L, beam height H, web thickness F, bottom plate thickness D, concrete bulk density γ, concrete elastic modulus E, hanging basket load G, strength P, and cable tension T as input variables, and pre-camber value as output variable; S200: Cleaning and preprocessing of beam structure design parameters, construction process monitoring data, and material property data, including outlier correction, missing value interpolation, and standardization, and dividing into training and test sets; S300, constructing a Gaussian process regression model, selecting a rational quadratic kernel function, adjusting the kernel function hyperparameters through the Bayesian optimization algorithm, and training the model to establish a nonlinear mapping relationship between input variables and pre-camber; S400: Input real-time construction data, predict the pre-camber value and confidence interval through the trained model, and calculate the root mean square error, mean absolute error, and determination coefficient to evaluate the prediction accuracy; S500 dynamically updates model hyperparameters based on measured data, forming a closed-loop feedback mechanism of prediction-measurement-correction to optimize subsequent prediction results.

2. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: The bridge structure design parameters include: cantilever length L, beam height H, web thickness F and bottom plate thickness D; The construction process monitoring data includes: concrete bulk density γ, concrete elastic modulus E and strength P; The material property data include: basket load G and cable force T.

3. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: The standardization process in S200 uses the Z-score normalization method, the formula is: X norm =(x-μ) / σ; where x is the original data, X norm is the corresponding normalized data, μ is the mean, and σ is the standard deviation.

4. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: The rational quadratic kernel function expression in S300 is: Among them, x i and x j is the input data point; ||x i -x j || is the Euclidean distance between two points; l is the length scale parameter, and α is the scale mixing parameter.

5. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: Bayesian optimization in S300 is implemented by maximizing the marginal likelihood function.

6. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: The confidence interval prediction formula in S400 is: Where K is the covariance matrix of the training data, σ 2 n Observation noise variance; x * For a given new input, μ(x * ) is the predicted value and σ 2 (x * ) is the variance, I is the identity matrix, and X is the input feature matrix of the training dataset.

7. The method for predicting the pre-camber of a cantilever cast-in-place bridge based on machine learning according to claim 1 is characterized in that: Accuracy assessment in S400 includes: Draw a comparison chart of predicted values ​​and true values, and a residual chart; Calculate RMSE and MAE to quantify the prediction error, R 2 To assess the goodness of model fit.

8. The method for predicting the pre-camber of a cantilever cast-in-place bridge using machine learning according to claim 1 is characterized in that: The method of dynamically updating model hyperparameters based on measured data includes: The model is calibrated using measured data from a preset number of segments in front of the bridge, and the model hyperparameters are updated.

Images