Shield construction earthwork volume prediction method based on Kriging interpolation and classification regression tree
The integration of Kriging interpolation and classification regression trees with dynamic variance adjustments and Bayesian updates addresses the limitations of traditional methods in predicting soil volume during tunneling, enhancing accuracy and adaptability in complex geological conditions.
Patent Information
- Application Number
- CN202510807519.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-17
- Publication Date
- 2025-07-15
- Estimated Expiration
- 2045-06-17
AI Technical Summary
Traditional earth-to-log volume prediction methods are difficult to adapt to stratigraphic changes when facing complex geological conditions and construction environments, resulting in insufficient prediction accuracy. Especially in local geological mutation areas, the existing technology Kriging interpolation method has limited ability to adjust the variogram function.
Combining the methods of Kriging interpolation and classification regression tree, the variograms are dynamically adjusted by collecting construction, geological and environmental parameters in real time, multi-scale interpolation fusion strategy is adopted, and an adaptive classification regression tree model is constructed, and the Bayesian update method is used for online optimization and weight adjustment to generate the final earth-to-log quantity prediction result.
The spatial accuracy and adaptability of soil flow prediction are improved, overfitting problems are avoided, the model's adaptability to new data is enhanced, and the accuracy and stability of the prediction results are ensured.
Smart Images

Figure CN120317460A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of earthwork volume prediction in shield tunneling construction, specifically a method for predicting earthwork volume in shield tunneling construction based on Kriging interpolation and classification and regression tree. Background Technique
[0002] During the shield tunneling construction process, the prediction of earthwork volume is of great significance for construction progress and cost control. Traditional earthwork volume prediction methods usually rely on historical data and simple mathematical models, but these methods often cannot fully consider the dynamic changes of construction parameters, geological conditions, and environmental factors. Therefore, many researchers have tried to combine machine learning algorithms with traditional numerical methods to improve the accuracy of earthwork volume prediction. For example, some methods use multiple linear regression models to predict earthwork volume and conduct statistical analysis based on construction parameters and geological data. However, traditional methods often show low prediction accuracy when dealing with complex geological conditions and construction environments, especially in scenarios with drastic formation changes and complex construction conditions.
[0003] Most of the existing earthwork volume prediction technologies rely on traditional regression analysis methods or simple interpolation algorithms. These methods are often difficult to adapt to complex situations when facing the variability of construction parameters and geological environments. Although the traditional Kriging interpolation method can perform a certain degree of interpolation on geological data in space, for local mutation regions of the formation, its ability to adjust the variogram is limited, resulting in poor prediction accuracy of the interpolation results in these regions. Therefore, there are still significant deficiencies in the accuracy and adaptability of the existing technology in multi-parameter and multi-stage earthwork volume prediction. Summary of the Invention
[0004] The present invention proposes a method for predicting earthwork volume in shield tunneling construction based on Kriging interpolation and classification and regression tree. By collecting construction parameters, geological parameters, and environmental parameters in real time, and combining a dynamically adjusted variogram and a multi-scale interpolation fusion strategy, the problem of prediction accuracy of traditional Kriging interpolation in local geological change regions is solved.
[0005] Among them, the method for predicting earthwork volume in shield tunneling construction based on Kriging interpolation and classification and regression tree includes the following steps: S1. Collect construction parameters, geological parameters, and environmental parameters generated during the shield tunneling construction process in real time, and perform outlier detection, missing value imputation, and smoothing processing on the collected data; S2. Establish a spatial data model of the shield tunneling construction area through geological parameters, and perform spatial prediction on the formation data by combining local and global interpolation, adopting a dynamically adjusted variogram and a multi-scale interpolation fusion strategy, and dynamically correct the interpolation results using the Bayesian update method during the shield propulsion process; S3. Perform feature selection and interaction analysis based on construction parameters, geological parameters, and environmental parameters, construct an initial classification and regression tree model, and perform online update and optimization of the model by combining the dynamic pruning algorithm and the incremental learning strategy; S4. According to the trained adaptive classification and regression tree model, predict the changes in earthwork volume in the short term, medium term, and long term respectively, input the short-term, medium-term, and long-term prediction results into the Bayesian fusion model, calculate the credibility weights of different models, and adjust the weights of each prediction model according to the actual errors in the construction process through the dynamic weight adjustment strategy to generate the final earthwork volume prediction result.
[0006] Further, in the step S1, the construction parameters at least include the propulsion speed of the shield machine, the cutter head rotation speed, the soil chamber pressure, and the screw conveyor rotation speed: the geological parameters at least include the formation distribution, water content, sand content, and porosity within the shield construction area; the environmental parameters at least include the groundwater level and surface settlement within the shield construction area.
[0007] Further, the step S2 specifically includes the following steps: S201. Establish an initial spatial data grid based on the preprocessed geological parameters, and calculate the variogram of the initial spatial data grid through the traditional Kriging interpolation method; S202. For the formation mutation area, dynamically adjust the local variogram parameters, and re-model the local geological characteristics to establish a local interpolation model; S203. Calculate the formation trend of the entire shield construction area based on historical data, and perform fitting using global Kriging interpolation to establish a global interpolation model; S204. Use the multi-scale interpolation fusion strategy to perform weight allocation and fusion of the local interpolation model and the global interpolation model.
[0008] Further, in the step S201, the variogram is specifically expressed as: ; wherein, the represents the variogram, the represents the spatial distance, the represents the number of sample point pairs with a distance of , the represents the sample index, the represents the formation parameter value at the position , and the represents the formation parameter value at the position .
[0009] Further, the step S3 specifically includes the following steps: S301. Construct a candidate feature set based on the preprocessed construction parameters, geological parameters, and environmental parameters; S302. Calculate the information entropy and perform feature interaction analysis on the constructed candidate feature set, quantitatively evaluate the impact of each feature on the change in earthwork volume, screen the key influencing factors that make significant contributions to the prediction results, and eliminate redundant or low - contribution features; S303. Construct an initial classification and regression tree model based on the screened key influencing factors, determine the optimal splitting point using the Gini coefficient or information gain criterion at each splitting node, and capture the non - linear relationship between the construction parameters and the earthwork volume; S304. Optimize the initial classification and regression tree through an adaptive pruning algorithm, and automatically delete redundant branches by evaluating the error contribution of each branch; S305. Update the classification and regression tree model online according to new data during the shield tunneling construction process through an incremental learning strategy.
[0010] Further, the step S302 specifically includes the following sub - steps: S3021. Calculate the mutual information values between the construction parameters, geological parameters, and environmental parameters, measure the correlation degree between each variable and the earthwork volume, and evaluate the contribution of the variables to the final prediction accuracy using the information entropy analysis method; S3022. Eliminate variables with high correlation but low information gain with other features through the maximum correlation and minimum redundancy algorithm; S3023. After feature selection, classify different construction scenarios and classify the key influencing factors under each scenario.
[0011] Further, the step S303 specifically includes the following sub - steps: S3031. Combine the CART algorithm, use the screened key influencing factors as input variables and the earthwork volume per unit time as the target variable to construct an initial tree structure; S3032. Calculate the Gini coefficient or information gain of all candidate features at each splitting node, and select the optimal feature as the splitting basis; S3033. Determine the optimal splitting point according to the data characteristics, and capture the non - linear relationship between the construction parameters and the earthwork volume at each split through the information gain ratio of the minimum variance reduction criterion.
[0012] Further, the step S4 specifically includes the following sub - steps: S401. Generate short - term prediction results, medium - term prediction results, and long - term prediction results respectively based on the trained adaptive classification and regression tree model; S402. Input the short-term, medium-term, and long-term prediction results into the constructed Bayesian fusion model, and calculate the credibility weights through the Bayesian fusion model; S404. Adopt a dynamic weight adjustment strategy to online correct the weights of each prediction model according to the real-time feedback data and error analysis results during the construction process, enabling the Bayesian fusion model to adaptively update; S405. Generate the final comprehensive prediction result.
[0013] Further, the step S401 specifically includes the following sub-steps: S4011. In the short-term prediction stage, with a sliding window of 5 minutes as the time step, summarize the real-time collected construction data into a short-term data set, and input this data set into the adaptive classification regression tree model to output the earthwork volume prediction result within the next 5 minutes; S4012. In the medium-term prediction stage, with a sliding window of 30 minutes as the time step, summarize the construction data collected within the last 30 minutes into a medium-term data set, and input this medium-term data set into the adaptive classification regression tree model to output the earthwork volume prediction result within the next 30 minutes; S4013. In the long-term prediction stage, use the historical data covering the entire construction period as the long-term data set, and input this long-term data set into the adaptive classification regression tree model to output the prediction of the earthwork volume change trend throughout the construction period.
[0014] Further, in the step S402, the specific process of calculating the credibility weights through the Bayesian fusion model is expressed as: ; ; ; where, the represents the credibility weight, the represents the posterior probability of model under the current construction data , the represents the posterior probability of model under the current construction data , the represents the result type, including short-term, medium-term, and long-term prediction results, the represents the likelihood probability that model generates the current data , the represents the likelihood probability that model generates the current data , and the represents the prediction variance. Indicates the prediction error.
[0015] The beneficial effects of the invention are as follows: Through the Bayesian update method, the present invention dynamically corrects the formation interpolation results, improving the accuracy of spatial prediction. In the construction of the classification and regression tree model, the present invention introduces an adaptive pruning algorithm and an incremental learning strategy, which can update and optimize the model online according to real-time construction data, avoid the overfitting problem in traditional methods, and enhance the adaptability of the model to new data. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is a flowchart of the shield construction earthwork volume prediction method based on Kriging interpolation and classification and regression tree provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0017] The technical solution of the present invention will be further described in detail below with reference to the drawings, but the protection scope of the present invention is not limited to the following description.
[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention, that is, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.
[0019] Therefore, the detailed description of the embodiments of the present invention provided in the drawings below is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts belong to the protection scope of the present invention. It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.
[0020] Moreover, the term "comprising", "including" or any other variant thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "including a..." does not exclude the presence of additional identical elements in the process, method, article or device including the element.
[0021] The features and performance of the present invention will be further described in detail below in conjunction with embodiments.
[0022] Embodiment 1 Among them, as Figure 1 , a method for predicting the earthwork volume of shield tunneling construction based on Kriging interpolation and classification and regression tree includes the following steps: S1. During the construction process of the shield machine, collect construction parameters, geological parameters, and environmental parameters in real time, and perform outlier detection, missing value imputation, and smoothing processing on the collected data; S2. Establish a spatial data model for the shield tunneling construction area through geological parameters, and adopt a combination of local and global interpolation methods, and use a dynamic adjustment variogram and a multi-scale interpolation fusion strategy to perform spatial prediction on the formation data, and use the Bayesian update method to dynamically correct the interpolation results during the shield propulsion process; S3. Perform feature selection and interaction analysis based on construction parameters, geological parameters, and environmental parameters, construct an initial classification and regression tree model, and combine a dynamic pruning algorithm and an incremental learning strategy to perform online update and optimization on the model; S4. According to the trained adaptive classification and regression tree model, predict the earthwork volume changes in the short term, medium term, and long term respectively, input the short-term, medium-term, and long-term prediction results into the Bayesian fusion model, calculate the credibility weights of different models, and adjust the weights of each prediction model according to the actual error during the construction process through a dynamic weight adjustment strategy to generate the final earthwork volume prediction result.
[0023] Furthermore, in step S1, the construction parameters at least include the propulsion speed, cutter head rotation speed, soil chamber pressure, and screw conveyor rotation speed of the shield machine: the geological parameters at least include the formation distribution, water content, sand content, and porosity in the shield tunneling construction area; the environmental parameters at least include the groundwater level and surface settlement in the shield tunneling construction area. Specifically, the specific implementation process of step S1 is as follows: In the real-time collection and preprocessing stage, collect the construction parameters, geological parameters, and environmental parameters of the shield machine through installed sensors to ensure the real-time and stability of the data; perform outlier detection on the collected data, and use statistical methods to identify and eliminate outliers; at the same time, for missing values, use interpolation methods or machine learning methods to fill them, and then through smoothing processing, use the moving average method and weighted average method to reduce the noise in the data and ensure the data is stable.
[0024] Furthermore, step S2 specifically includes the following steps: S201. Establish an initial spatial data grid based on the preprocessed geological parameters, and calculate the variogram of the initial spatial data grid through the traditional Kriging interpolation method; S202. For the formation mutation region, dynamically adjust the local variogram parameters, re-model the local geological features, and establish a local interpolation model; S203. Based on historical data, calculate the formation trend of the entire shield construction area, and use global Kriging interpolation for fitting to establish a global interpolation model; S204. Use the multi-scale interpolation fusion strategy to perform weight assignment and fusion on the local interpolation model and the global interpolation model.
[0025] Further, in step S201, the variogram is specifically expressed as: ; wherein, the represents the variogram, the represents the spatial distance, the represents the number of sample point pairs with a distance of , the represents the sample index, the represents the formation parameter value at the position, the represents the position and the
[0026] Specifically, in the spatial modeling process of geological data, it is first necessary to establish a preliminary spatial data grid based on the processed geological data. The construction of the spatial data grid takes into account the geological characteristics of the region and ensures that it can reflect various geological changes within the region. At this time, the Kriging interpolation method is used, and its basic principle is to calculate the variogram through the distance relationship between geological data points to quantify the degree of geological change at different positions in space. When performing spatial interpolation, considering the existence of the formation mutation area, a strategy of dynamically adjusting the variogram is adopted, which means that in the mutation area, the parameters of the interpolation method will be automatically adjusted according to the local geological characteristics to enhance the interpolation accuracy. For non-mutation areas, the global Kriging interpolation method is used to fit the overall formation trend within the region. The advantage of Kriging interpolation is that it can use spatial autocorrelation to infer geological parameters at unknown positions and has a certain robustness to data noise. Through the fusion of local and global interpolation methods, it is ensured that the model can not only accurately describe the characteristics of the mutation area but also reasonably reflect the changes of the overall formation.
[0027] Further, step S3 specifically includes the following steps: S301. Construct a candidate feature set according to the preprocessed construction parameters, geological parameters, and environmental parameters; S302. Calculate the information entropy and conduct feature interaction analysis on the constructed candidate feature set, quantitatively evaluate the impact of each feature on the change in earthwork volume, screen out the key influencing factors that make significant contributions to the prediction results, and eliminate redundant or low-contribution features; S303. Construct an initial classification and regression tree model based on the screened key influencing factors, and determine the optimal splitting point at each splitting node using the Gini coefficient or information gain criterion to capture the non-linear relationship between construction parameters and earthwork volume; S304. Optimize the initial classification and regression tree through an adaptive pruning algorithm, and automatically delete redundant branches by evaluating the error contribution of each branch; S305. Update the classification and regression tree model online according to new data during the shield tunneling construction process through an incremental learning strategy.
[0028] Specifically, the classification and regression tree (CART) model is the core for predicting the construction earthwork volume. Through the recursive splitting of the tree structure, the CART model can capture the complex non-linear relationships between different construction parameters and earthwork volume. When constructing the CART model, it is first necessary to conduct feature selection and interaction analysis on construction parameters, geological parameters, and environmental parameters. The core purpose of this step is to eliminate redundant information and find the features most relevant to the change in earthwork volume. During the feature selection process, by calculating the information entropy value, the contribution degree of each feature to the final prediction result can be evaluated, and mutual information analysis is further used to quantify the mutual relationship between different parameters and earthwork volume. After feature screening, a decision tree is constructed based on the CART algorithm, and the optimal splitting point is determined by calculating the Gini coefficient or information gain of each node, thereby optimizing the structure of the tree. The advantage of the CART model is that it can handle large-scale high-dimensional data, and through the division of splitting nodes, the model can capture the impact of key parameters during the construction process on earthwork volume. To improve the generalization ability of the model, an adaptive pruning method is adopted to dynamically prune redundant branches during the growth of the tree to avoid overfitting. In addition, to further improve the accuracy of the prediction model, an incremental learning strategy is used to dynamically update the classification and regression tree. The incremental learning method can continuously update the model according to new construction data during the shield tunneling construction process, making the prediction more suitable for the gradually changing environment and conditions during the construction process. Each time it is updated, the model will adjust the structure and weights of the tree according to the newly added data to make it more adaptable to the new construction environment.
[0029] Further, the step S302 specifically includes the following sub-steps: S3021. Calculate the mutual information values between construction parameters, geological parameters, and environmental parameters, measure the correlation degree between each variable and earthwork volume, and evaluate the contribution of variables to the final prediction accuracy using the information entropy analysis method; S3022. Eliminate variables that have a high correlation with other features but a low information gain through the maximum correlation and minimum redundancy algorithm; S3023. After feature selection is completed, classify different construction scenarios and classify the key influencing factors under each type of scenario.
[0030] Further, the step S303 specifically includes the following sub-steps: S3031. Combine the CART algorithm, use the selected key influencing factors as input variables, and the earthwork volume per unit time as the target variable to construct an initial tree structure; S3032. Calculate the Gini coefficient or information gain of all candidate features at each splitting node, and select the optimal feature as the splitting basis; S3033. Determine the optimal splitting point according to the data characteristics, and capture the non-linear relationship between the construction parameters and the earthwork volume at each split through the minimum variance reduction criterion information gain ratio.
[0031] Further, the step S4 specifically includes the following sub-steps: S401. Generate short-term prediction results, medium-term prediction results, and long-term prediction results respectively based on the trained adaptive classification and regression tree model; S402. Input the short-term, medium-term, and long-term prediction results into the constructed Bayesian fusion model, and calculate the credibility weights through the Bayesian fusion model; S404. Adopt a dynamic weight adjustment strategy, and online correct the weights of each prediction model according to the real-time feedback data and error analysis results during the construction process, so that the Bayesian fusion model can be adaptively updated; S405. Generate the final comprehensive prediction result.
[0032] Specifically, in the fusion stage of the prediction results, the Bayesian fusion model combines the short-term, medium-term, and long-term prediction results to calculate the credibility weights of each model. The Bayesian inference principle plays an important role here. It can calculate the posterior probability of each prediction result based on the current construction data and historical data, and allocate weights according to the credibility of different prediction models. Further, the Bayesian model dynamically adjusts the weights of each model by calculating the likelihood and posterior probability of each model's prediction result, enabling the Bayesian fusion model to automatically adapt to the changes in different construction stages and prediction time periods. As the real-time data during the construction process flows in, the Bayesian model will correct the prediction results according to the actual errors to ensure the continuous accuracy of the model.
[0033] In this process, the dynamic weight adjustment strategy plays a key role. When new construction data appears during the construction process, the Bayesian model adjusts the weights of each prediction model according to this real-time data, ensuring that in different prediction stages of short-term, medium-term, and long-term, the contribution degree of each model matches the current construction environment and construction progress. The adaptive adjustment strategy effectively copes with the uncertainties and complexities during the construction process, guaranteeing the accuracy of the final prediction result.
[0034] Further, the step S401 specifically includes the following sub-steps: S4011. In the short-term prediction stage, with a sliding window with a time step of 5 minutes, the real-time collected construction data is summarized into a short-term data set, and this data set is input into the adaptive classification and regression tree model to output the earthwork volume prediction result within the next 5 minutes; S4012. In the medium-term prediction stage, with a sliding window with a time step of 30 minutes, the construction data collected within the last 30 minutes is summarized into a medium-term data set, and this medium-term data set is input into the adaptive classification and regression tree model to output the earthwork volume prediction result within the next 30 minutes; S4013. In the long-term prediction stage, the historical data covering the entire construction period is used as the long-term data set, and this long-term data set is input into the adaptive classification and regression tree model to output the prediction of the earthwork volume change trend during the entire construction period.
[0035] Further, in the step S402, the specific process of the credibility weight calculated by the Bayesian fusion model is expressed as: ; ; ; Among them, the represents the credibility weight, the represents the posterior probability of model under the current construction data , the represents the posterior probability of model under the current construction data , the represents the result type, including short-term, medium-term, and long-term prediction results, the represents the likelihood probability that model generates the current data , the represents the likelihood probability that model generates the current data , the represents the prediction variance, and the represents the prediction error.
[0036] Further, the Bayesian fusion model is constructed before step S401, that is, it further includes step S400: establishing an initialized Bayesian model and assigning initial weights to each prediction model. The initial weights are determined based on the performance of the cross-validation model during the training process in S3. Exemplarily, if the short-term model has a smaller error on recent data, a higher weight is assigned. Further, the short-term, medium-term, and long-term prediction models respectively calculate the predicted earthwork volume, and the Bayesian model calculates the overall predicted value based on these prediction results. As the shield construction progresses, new data is continuously input, and the Bayesian model continuously updates the weights based on the actual error. Each time a new prediction is calculated, the system analyzes the error and reassigns the weights of each model, so that the weight of the model with a larger error is reduced, and the weight of the model with a smaller error is increased. The above dynamic adjustment mechanism ensures that the prediction model can adapt to the changes in the construction environment and improves the prediction accuracy.
[0037] Embodiment 2 Further, as a preferred implementation of the above embodiment, it is proposed to construct a classification and regression tree by calculating the Gini index of each feature. For the sample set in node t, the Gini index is: ; where the represents the base number, the represents the probability that the sample belongs to class , and the represents the number of classes; the feature and the corresponding threshold with the smallest Gini index are selected for splitting each node: ; where the represents the feature with the smallest Gini index, the represents the threshold corresponding to feature , the represents the subset split based on feature and threshold , and the represents the Gini index of subset .
[0038] Embodiment 3 Further, as an exemplary technical solution of the above embodiment, an application scenario is proposed. This scenario applies the Kriging interpolation method. Assuming that the geological location is a two-dimensional coordinate system (x, y), the soil type at unknown locations is inferred from the data points at known locations. Specifically: Given the soil type and depth data, calculate the variogram (such as the semi-variogram) between every two locations, and calculate the spatial variability based on the geological characteristics (soil type, depth) at different locations; set the earthwork volume at the location to be predicted (6.5, 12.5), and calculate it through the Kriging interpolation method based on the samples at known locations, that is: ; wherein, the represents the earthwork volume of the predicted target spatial position, the represents the weight coefficient calculated by the variogram, and the represents the data value at the known position Fuse the interpolated geological prediction results with the real-time collected construction and environmental parameters, use the Classification and Regression Tree (CART) model to mine the non-linear relationship affecting the change of earthwork volume, and realize the adaptive optimization of the model through the online update mechanism. According to the features in the data (such as humidity, thrust, etc.), the Classification and Regression Tree will select the optimal features and splitting points based on the minimized Gini index. Select the shield machine thrust as the splitting feature and use the Gini index to calculate the splitting effect; each time new data (such as the soil humidity of (10, 18, 21) is 40%) arrives, incremental learning adapts to the new data by updating the model weights. Use three models: model M1 based on soil humidity, model M2 based on shield machine thrust, and model M3 based on depth; calculate the posterior probability of each model and perform weighted fusion according to Bayes' theorem; each time new data arrives, update the Bayesian posterior probability according to the real-time data. Assume that the new data point increases the data of the shield machine thrust and its posterior probability is updated, and the posterior probabilities of other models are adjusted accordingly. Finally, the prediction result is output.
[0039] The above are only the preferred embodiments of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. And the changes and modifications made by those skilled in the art that do not depart from the spirit and scope of the present invention shall fall within the protection scope of the appended claims of the present invention.
Claims
1. A method for predicting the earthwork volume of shield construction based on Kriging interpolation and classification and regression tree, characterized in that, It includes the following steps: S1. During the construction of the shield machine, the construction parameters, geological parameters, and environmental parameters are collected in real time, and the collected data is subjected to outlier detection, missing value imputation, and smoothing processing; S2. A spatial data model of the shield construction area is established through geological parameters, and through a combination of local and global interpolation, a dynamic adjustment variogram and a multi-scale interpolation fusion strategy are used to perform spatial prediction on the formation data, and the Bayesian update method is used to dynamically correct the interpolation results during the shield propulsion process; S3. Feature selection and interaction analysis are performed based on construction parameters, geological parameters, and environmental parameters to construct an initial classification and regression tree model, and the model is updated and optimized online by combining a dynamic pruning algorithm and an incremental learning strategy; S4. According to the trained adaptive classification and regression tree model, the changes in earthwork volume in the short term, medium term, and long term are predicted respectively, and the short-term, medium-term, and long-term prediction results are input into the Bayesian fusion model to calculate the credibility weights of different models. Through a dynamic weight adjustment strategy, according to the actual error during the construction process, the weights of each prediction model are adjusted to generate the final earthwork volume prediction result.
2. The shield tunneling earthwork volume prediction method based on Kriging interpolation and classification and regression tree according to claim 1, wherein In step S1, the construction parameters at least include the propulsion speed, cutter head rotation speed, chamber pressure, and screw conveyor rotation speed of the shield machine: the geological parameters at least include the formation distribution, water content, sand content, and porosity within the shield construction area; the environmental parameters at least include the groundwater level and surface settlement within the shield construction area.
3. The shield tunneling earthwork volume prediction method based on Kriging interpolation and classification and regression tree according to claim 1, characterized in that Step S2 specifically includes the following steps: S201. An initial spatial data grid is established based on the preprocessed geological parameters, and the variogram of the initial spatial data grid is calculated by the traditional Kriging interpolation method; S202. For the formation mutation area, the local variogram parameters are dynamically adjusted, and the local geological features are re-modeled to establish a local interpolation model; S203. Based on historical data, the formation trend of the entire shield construction area is calculated, and global Kriging interpolation is used for fitting to establish a global interpolation model; S204. Using a multi-scale interpolation fusion strategy, the local interpolation model and the global interpolation model are weighted and fused.
4. The shield construction earthwork volume prediction method based on Kriging interpolation and classification and regression tree according to claim 3, characterized in that In step S201, the variogram is specifically expressed as: ; Among them, the represents the variogram, the represents the spatial distance, the represents the number of sample point pairs with a distance of , the represents the sample index, the represents the position at which the formation parameter value is represents the position at which the formation parameter value is 5. The method for predicting the earthwork volume of shield tunneling construction based on Kriging interpolation and classification and regression tree according to claim 1, wherein, Step S3 specifically includes the following steps: S301. A candidate feature set is constructed based on the preprocessed construction parameters, geological parameters, and environmental parameters; S302. Information entropy calculation and feature interaction analysis are performed on the constructed candidate feature set to quantitatively evaluate the influence of each feature on the change in earthwork volume, screen the key influencing factors that have a significant contribution to the prediction result, and eliminate redundant or low-contribution features; S303. An initial classification and regression tree model is constructed based on the screened key influencing factors, and the Gini coefficient or information gain criterion is used to determine the optimal splitting point at each splitting node to capture the non-linear relationship between construction parameters and earthwork volume; S304. The initial classification and regression tree is optimized by an adaptive pruning algorithm, and redundant branches are automatically deleted by evaluating the error contribution of each branch; S305. Update the classification and regression tree model online according to new data during the shield tunneling construction process through an incremental learning strategy.
6. The method for predicting the earthwork volume of shield construction based on Kriging interpolation and classification and regression tree according to claim 5, characterized in that, The specific steps of step S302 include the following sub-steps: S3021. Calculate the mutual information values among construction parameters, geological parameters, and environmental parameters to measure the correlation degree between each variable and the earthwork volume, and use the information entropy analysis method to evaluate the contribution of variables to the final prediction accuracy; S3022. Eliminate variables with high correlation with other features but low information gain through the maximum correlation and minimum redundancy algorithm; S3023. After feature selection, classify different construction scenarios and classify the key influencing factors under each scenario.
7. The method for predicting the earthwork volume of shield tunneling construction based on Kriging interpolation and classification and regression tree according to claim 5, characterized in that, The specific steps of step S303 include the following sub-steps: S3031. Combine the CART algorithm, use the selected key influencing factors as input variables, and the earthwork volume per unit time as the target variable to construct an initial tree structure; S3032. Calculate the Gini coefficient or information gain of all candidate features at each splitting node, and select the optimal feature as the splitting basis; S3033. Determine the optimal splitting point according to the data characteristics, and capture the non-linear relationship between the construction parameters and the earthwork volume at each split through the information gain ratio of the minimum variance decreasing criterion.
8. The method for predicting the earthwork volume of shield tunneling construction based on Kriging interpolation and classification and regression tree according to claim 1, wherein, The specific steps of step S4 include the following sub-steps: S401. Generate short-term prediction results, medium-term prediction results, and long-term prediction results respectively based on the trained adaptive classification and regression tree model; S402. Input the short-term, medium-term, and long-term prediction results into the constructed Bayesian fusion model, and calculate the credibility weights through the Bayesian fusion model; S404. Adopt a dynamic weight adjustment strategy to online correct the weights of each prediction model according to the real-time feedback data and error analysis results during the construction process, so that the Bayesian fusion model can be adaptively updated; S405. Generate the final comprehensive prediction result.
9. The method for predicting the earthwork volume of shield tunneling construction based on Kriging interpolation and classification and regression tree according to claim 8, characterized in that, The specific steps of step S401 include the following sub-steps: S4011. In the short-term prediction stage, use a sliding window with a time step of 5 minutes to summarize the real-time collected construction data into a short-term data set, and input this data set into the adaptive classification and regression tree model to output the earthwork volume prediction result within the next 5 minutes; S4012. In the medium-term prediction stage, use a sliding window with a time step of 30 minutes to summarize the construction data collected in the recent 30 minutes into a medium-term data set, and input this medium-term data set into the adaptive classification and regression tree model to output the earthwork volume prediction result within the next 30 minutes; S4013. In the long-term prediction stage, use the historical data covering the entire construction period as the long-term data set, and input this long-term data set into the adaptive classification and regression tree model to output the prediction of the earthwork volume change trend during the entire construction period.
10. The method for predicting the earthwork volume of shield construction based on Kriging interpolation and classification and regression tree according to claim 8, characterized in that In step S402, the specific process of calculating the credibility weights through the Bayesian fusion model is expressed as: ; ; ; Among them, the represents the credibility weight, and the represents the posterior probability of the model under the current construction data . The represents the posterior probability of the model under the current construction data . The represents the result type, including short-term, medium-term, and long-term prediction results. The represents the likelihood probability that the model generates the current data . The represents the likelihood probability that the model generates the current data . The represents the prediction variance, and the represents the prediction error.
Citation Information
Patent Citations
Method for predicting geological conditions in front of shield tunneling machine in real time
CN114970768A
Shield tunneling load advanced prediction method and system based on Bayesian optimization LSTM
CN115081749A
Shield settlement deformation monitoring method and system based on gray neural network
CN119005435A
Spectrum map construction method based on dual-scale fusion
CN120090741A
Shield tunnel muck improvement parameter prediction method and system
CN121347784A
Cited By
Foreign matter invasion monitoring system and method for railway business line construction
CN120510668A
Double-line self-updating method and system for asphalt pavement performance estimation
CN121188747A
Diameter-variable shield disturbance mixing test method and system
CN121384508A
Shield tunneling machine real-time stratum characteristic prediction method with uncertainty quantification
CN121456380A
Construction deformation prediction method and device, electronic equipment and storage medium
CN121615879A
