A method for predicting water inflow in sustainable tunnel construction based on machine learning technology

Abstract

The present application belongs to the field of engineering technology and environmental protection, and discloses a sustainable tunnel construction water inflow prediction method based on machine learning technology, which comprises the following steps: data collection and preprocessing; using multivariate statistics and machine learning methods to identify the main control factors of water inflow; using Python programming language, according to the key influencing factors of water inflow, based on random forest, XGBoost and deep neural network model algorithm, respectively establishing efficient water inflow prediction model, and according to the evaluation parameters of the output performance of the initial water inflow prediction model, evaluating and comparing the performance of different efficient water inflow prediction models. The efficient water inflow prediction model with the closest output performance of the training set and test set of each efficient water inflow prediction model is determined as the best water inflow prediction model; and a sustainable tunnel construction water inflow intelligent prediction system is constructed. The present application optimizes the traditional water inflow prediction method by combining various machine learning technologies.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of engineering technology and environmental protection, and relates to a sustainable tunnel construction water inflow prediction method based on machine learning technology. BACKGROUND

[0002] In recent years, China has attached great importance to ecological civilization construction, and the development concept has gradually shifted from economic development to environmental protection and ecological balance. Under this background, various large-scale infrastructure construction projects, especially railway construction projects, have put forward higher requirements for environmental protection. In the construction of railway tunnels under complex geological conditions, accurate prediction of water inflow is crucial to ensuring construction safety and progress and reducing environmental impact.

[0003] In the field of water inflow prediction, the commonly used methods include theoretical calculation, numerical simulation and non-deterministic methods. Theoretical calculation relies on simplified groundwater flow models and engineering empirical formulas, and is suitable for simple geological and hydrological conditions. However, in complex geological and hydrological environments, theoretical calculation may have large errors and cannot accurately reflect the changes in water inflow during tunnel construction. Numerical simulation models the process of groundwater flow and solute transport through numerical solution, and can handle more complex geological and hydrological conditions. Although it has high precision, the calculation process is complex and depends on a large amount of measured data, resulting in large calculation volume and long time consumption, which leads to low efficiency in practical application. Non-deterministic methods (such as grey system model, stochastic process model, etc.) simulate the influence of uncertain factors on water inflow through probability analysis, have the advantages of simple concept and do not require detailed description of geological and hydrological characteristics in the modeling process, so they have gradually become a research hotspot in the prediction of water inflow during tunnel construction. Traditional prediction methods generally have the problems of insufficient accuracy, complex calculation and low efficiency. In recent years, machine learning technology has made significant progress in data processing and pattern recognition, and has shown strong prediction ability. Machine learning can automatically extract rules from historical data and model complex nonlinear relationships, thus having potential advantages in water inflow prediction. Compared with theoretical calculation, machine learning methods can significantly improve prediction accuracy; compared with numerical simulation, machine learning reduces the dependence on complex geological and hydrological conditions, reduces calculation cost, and improves the universality and accuracy of the model. In addition, machine learning technology can identify key factors affecting water inflow, further improving prediction efficiency. Therefore, combining modern data analysis technology to optimize water inflow prediction methods has become a difficult problem to be solved.

[0004] Studies have shown that traditional water inflow prediction models focus on physical models or statistical methods, and fail to fully utilize the advantages of machine learning technology for in-depth analysis and model optimization. Therefore, exploring machine learning-based tunnel construction water inflow prediction methods not only helps to improve prediction accuracy, but also significantly improves prediction efficiency, and has wide application prospects.

[0005] The present application aims to provide a sustainable tunnel construction water inflow prediction method based on machine learning technology, which comprehensively considers natural geographical conditions, geological conditions, hydrogeological conditions and tunnel characteristic conditions, accurately predicts the tunnel construction water inflow, provides scientific decision support for tunnel construction, optimizes the construction scheme and reduces the engineering risk. SUMMARY

[0006] The present application provides a tunnel construction water inflow prediction method based on machine learning technology, which aims to solve the problems of large error and complex calculation in traditional prediction methods. By constructing an efficient and intelligent prediction system, data automatic processing and result output are realized, the practicability is improved, and the method is widely applied to tunnel construction and environmental management.

[0007] The technical scheme of the present application is as follows:

[0008] A sustainable tunnel construction water inflow prediction method based on machine learning technology, the steps are as follows:

[0009] Step 1: Data collection and preprocessing

[0010] Collect the pre-construction data of the tunnel, including the geological exploration report of the tunnel site area, drilling data, and actual water inflow data and corresponding construction progress data within the construction period; organize the data to form the original water inflow influencing factor data set; use interpolation method and IQR method to check the missing values and abnormal values of the data, and perform descriptive statistical analysis to determine the distribution and fluctuation characteristics of each parameter;

[0011] The original water inflow influencing factors include two categories, namely engineering conditions and hydrogeological conditions; the engineering conditions include construction method (numerical value 1 for drill and blast method, numerical value 2 for TBM method), number of holes (the number of holes in the main tunnel section and auxiliary tunnel section is 1; the number of holes in the main tunnel and flat guide, TBM left and right line scene is 2; the number of holes when the auxiliary tunnel enters the main tunnel and flat guide at the same time is 4), number of working faces (the specific number needs to be determined according to the pre-construction preparation and actual site conditions), construction cross-sectional area (i.e. working face area, represented by the maximum contour length x height, m 2The tunneling parameters include the tunnel length (m), the tunnel cross-sectional area (m2), the cumulative footage (m, the distance from the tunnel face to the tunnel portal), the daily footage (m, the tunneling distance on the same day), the tunnel depth (m, the distance from the ground surface to the tunnel), the static water height (m, the distance from the stable underground water level to the tunnel roof), the hydrogeological conditions, and the rainfall (mm / d).

[0012] Further, the descriptive statistical analysis refers to obtaining the minimum value, the maximum value, the average value, and the standard deviation of each influencing factor.

[0013] Step 2: identifying the main control factors of the water inflow using multivariate statistical and machine learning methods;

[0014] First, the Python programming language is used to perform Granger causality test on the rainfall and the water inflow based on the vector autoregression (VAR) model in the statsmodels library; according to the lag time of the significant causal relationship, the time interval of data resampling is determined to balance the time-varying and non-time-varying factors.

[0015] Further, the general form of the VAR model of the two time series X and Y (i.e., the rainfall and the water inflow) is as follows:

[0016]

[0017] wherein X t and Y t represent the observation values of the time series X and Y at time t; p is the maximum lag period set; i is the lag period, and the value range is 1-p; and and are the coefficients in the model, respectively representing the influence degree of X and Y at time t-i on X and Y at time t; and are error terms;

[0018] Further, Grange causality test is conducted by F test, the null hypothesis is that there is no significant causality between time series X and Y, if the F value is significant, the null hypothesis is rejected, and it is considered that there is a significant Granger causality between rainfall and water inflow.

[0019] Secondly, the correlation analysis tool in SPSS 26.0 software is used to identify the influencing factors highly correlated with water inflow by using Spearman correlation coefficient.

[0020] Finally, the Python programming language is used to build an initial water inflow prediction model based on the Random Forest model algorithm using the water inflow and water inflow influencing factor data obtained in step 1, and the feature permutation importance tool (Permutation Importance) and SHAP (SHapley Additive exPlanations) tool are introduced to analyze the importance ranking of each influencing factor and its marginal contribution to the water inflow prediction result, and the importance of each influencing factor is ranked according to the feature importance score and SHAP value.

[0021] Further, in order to ensure the training effect of the data and ensure the repeatability of the results, 80% of the data is used as the training set and 20% of the data is used as the test set, and the random seed number is set to 42 when building the initial water inflow prediction model.

[0022] Further, the performance evaluation parameters of the initial water inflow prediction model include mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), determination coefficient (R 2 ), according to the formula:

[0023]

[0024] Where n is the number of samples, y i is the measured value of each sample, is the average value of the measured value of the index, is the predicted value obtained by using the model.

[0025] Further, the feature permutation importance tool quantifies the influence of each influencing factor on the prediction ability of the initial water inflow prediction model by evaluating the difference between the performance of the initial water inflow prediction model on the original water inflow influencing factor data set and the performance after the influencing factors are randomly permuted. Based on the initial water inflow prediction model, the contribution of each influencing factor to the change of water inflow is evaluated, and the feature value permutation importance ranking is performed according to the feature permutation importance score.

[0026] Further, the SHAP tool is based on the Shapley value in game theory, which quantifies the contribution of each influencing factor to the change in water inflow by calculating the marginal contribution of each influencing factor. According to the swarm chart obtained by SHAP tool analysis, the contribution and importance of each influencing factor to the change in water inflow are ranked. In the swarm chart, blue indicates a positive impact on the predicted water inflow, red indicates a negative impact on the predicted water inflow, and the deeper the color, the higher the degree. The SHAP value calculation is based on the formula:

[0027]

[0028] wherein: is the SHAP value of the kth influencing factor; N is the set of all influencing factors; S is a subset of features including the kth influencing factor; f(S) is the model prediction value of the subset S; f(S∪{k}) is the model prediction of the features in S plus the feature

[0029] Further, factors with a correlation coefficient > 0.3, feature permutation importance > 0.03, and SHAP value > 500 are determined as key influencing factors of water inflow, providing a basis for subsequent establishment of an efficient water inflow prediction model.

[0030] Step 3: Using Python programming language, based on the key influencing factors of water inflow selected in step 2, efficient water inflow prediction models are established based on random forest, XGBoost, and deep neural network (DNN) model algorithms. The performance of different efficient water inflow prediction models is evaluated and compared based on the evaluation parameters of the initial water inflow prediction model in step 2. In this step, 85% of the key influencing factors and water inflow data are used as the training set, and the remaining 15% are used as the test set. The training set and test set data for all efficient water inflow prediction models are the same, and the grid search method is used to adjust the parameters of the efficient water inflow prediction models.

[0031] Step 4: The efficient water inflow prediction model with the closest output performance of the training set and test set in step 3 is determined as the best water inflow prediction model. A series of different hydrogeological conditions and engineering conditions are set to verify the applicability of the best water inflow prediction model to different engineering construction environments, ensuring its effectiveness and accuracy in actual application.

[0032] Further, the output performance evaluation parameters of the initial water inflow prediction model in step 2 are used to evaluate the applicability of the best water inflow prediction model to different engineering construction environments. When the mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and determination coefficient (R 2 ) are within the acceptable range, it is considered that the best water inflow prediction model has high applicability to actual working conditions.

[0033] Step 5: Construct a sustainable tunnel construction water inflow intelligent prediction system; use the mysql-connector-python library to realize efficient connection between Python and MySQL, and ensure fast interaction of the best water inflow prediction model database. The system inputs the key influencing factor data into the best water inflow prediction model through database management, and performs prediction; the prediction results are returned to the database and stored, facilitating subsequent query and analysis.

[0034] Further, in the sustainable tunnel construction water inflow intelligent prediction system, MySQL is mainly used to store and manage various parameter data, historical water inflow data and prediction results of the best water inflow prediction model of tunnel construction.

[0035] Further, the sustainable tunnel construction water inflow intelligent prediction embedded dataset will be continuously and dynamically updated, and the universality and prediction accuracy of the model will be improved by continuously increasing new training samples, to ensure that the system continuously optimizes the prediction performance when the data volume increases.

[0036] The beneficial effects of the present application are:

[0037] 1. The present application optimizes the traditional water inflow prediction method by combining multiple machine learning techniques (such as XGBoost regression, SHAP interpreter, deep neural network, etc.). Through automatic data processing and model training, the interference of human factors is reduced, and the accuracy and efficiency of water inflow prediction are significantly improved, effectively reducing the calculation cost and time consumption.

[0038] 2. The present application can accurately predict the possible water inflow changes during tunnel construction, and provide scientific basis to help the construction team to adjust the construction scheme in real time. This method effectively reduces the risk of safety accidents and construction delay caused by water inflow, and provides reliable data support for water resource management and groundwater protection, reducing the negative impact on the environment.

[0039] 3. The present application not only improves the precision and efficiency of tunnel construction water inflow prediction, but also promotes the application of machine learning technology in the field of groundwater prediction. Through optimization of model training and data analysis process, the present application promotes the technological innovation in the field of tunnel construction in terms of water inflow prediction, and provides new technical means for the engineering practice in this field. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 is a flowchart of a tunnel construction water inflow prediction method based on machine learning technology of the present application.

[0041] Figure 2are the Granger causality test results of rainfall and water inflow, where (a) is the drilling and blasting method and flat guide construction scenario, (b) is the drilling and blasting method and inclined shaft construction scenario, and (c) is the TBM method and left-right line construction scenario.

[0042] Figure 3 are the Spearman correlation graphs of various influencing factors and water inflow.

[0043] Figure 4 are the importance ranking graphs of the influence of various parameters on water inflow based on feature permutation importance (PI) calculation.

[0044] Figure 5 is the SHAP value bee swarm graph.

[0045] Figure 6 is the deep neural network (DNN) prediction model for the C tunnel No. 4 cross-hole water inflow prediction verification graph.

[0046] Figure 7 is the deep neural network (DNN) prediction model for the F tunnel No. 2 inclined shaft water inflow prediction verification graph. DETAILED DESCRIPTION

[0047] The specific embodiments of the present application are further described below in combination with the drawings and technical solutions.

[0048] EMBODIMENT

[0049] The flowchart of a sustainable tunnel construction water inflow prediction method based on machine learning technology is shown in Figure 1 , which includes the following steps:

[0050] Step 1: Collect the pre-tunnel construction data such as geological exploration reports and drilling data, and collect the actual water inflow data and corresponding construction progress data during the construction period, and complete the establishment of the water inflow data and its influencing factor data set. In order to establish a sample sufficient database, a typical tunnel construction site of a certain high-speed railway construction is taken as the research object, the site covers drilling and blasting method and TBM construction method, different hydrogeological conditions and combination mode of main tunnel and auxiliary tunnel, and the site characteristics and data period are shown in Table 1:

[0051] Table 1 Site characteristics and data period

[0052]

[0053] Further, the collected data is checked for missing values and outliers by interpolation method and IQR method, and then descriptive statistical analysis is performed, and the descriptive statistical data are shown in Table 2.

[0054] Table 2 Statistical table of each index parameter

[0055] Indicator Unit Minimum value Maximum value Average value Standard deviation Inflow m 3 / d]]> 368.6 96775.6 15199.9 15626.1 Rainfall mm / d 0.0 205.9 10.9 23.9 Daily footage m 0.0 15.9 3.1 3.0 Cumulative footage m 88.0 7822.6 2175.1 1309.3 Tunnel depth m 59.0 1804.0 605.7 302.9 Static water height m 0.0 1053.0 362.9 197.8 Infiltration coefficient - 0.1 1.0 0.3 0.2 Permeability coefficient m / d 0.0 0.9 0.1 0.1 Number of tunnels - 1.0 4.0 1.5 0.7 Number of working faces - 1.0 4.0 1.7 0.9 Construction method - 1.0 2.0 1.1 0.4 Construction cross-sectional area m 2 ]]> 54.7 98.5 60.0 12.7 Rock mass grade - 2.0 5.0 4.2 0.6 Water enrichment degree L / s 0.2 7.0 1.5 1.6 Type of groundwater - 1.0 3.0 1.9 0.3 Length of karst and fault zone m 0.0 136.0 12.8 30.2

[0056] Step 2: Identify the main controlling factors of water inflow using multivariate statistical and machine learning methods.

[0057] Firstly, the Granger causality test between rainfall and water inflow was conducted, and the test results are shown in Figure 2 . The Granger causality analysis of rainfall and water inflow was conducted at the representative work points of B tunnel horizontal guide exit, F tunnel No. 2 inclined shaft and K tunnel exit. The lag time of water inflow at B tunnel horizontal guide exit on rainfall was 2-3 days, and there was a significant causal relationship between them (p<0.05). The rainfall and water inflow at F tunnel No. 2 inclined shaft had a significant causal relationship (p<0.05) when the lag time was 1-5 days, and the significance gradually weakened with the increase of lag time. The rainfall and water inflow at K tunnel exit had a significant causal relationship (p<0.05) when the lag time was 1-3 days, but the causal relationship was no longer significant when the lag time exceeded 4 days. The Granger causality analysis results showed that rainfall had a significant impact on water inflow only in a short period. According to the lag time of significant causal relationship, the time interval of data resampling was determined to be 5 days.

[0058] Secondly, the correlation analysis tool in SPSS 26.0 software was used to identify the influencing factors with high correlation with water inflow based on Spearman correlation coefficient and significance test results, and the analysis results are shown in Figure 3 . The factors with correlation coefficient greater than 0.3 with water inflow included permeability coefficient, construction method, water enrichment degree, groundwater type and length of crossing karst and fault.

[0059] Finally, the Python programming language was used to build the initial water inflow prediction model of Random Forest based on the water inflow and 15 influencing factors data obtained in step 1, and the Permutation Importance and SHAP (SHapley Additive exPlanations) tools were introduced to analyze the importance ranking of each influencing factor and its marginal contribution to the prediction results of water inflow, and the analysis results are shown in Figure 4 and Figure 5 . According to the Permutation Importance score, Figure 4 ), the factors with score greater than 0.03 included fault and fracture zone length, groundwater type, static water height, construction method, rainfall and permeability coefficient. According to the SHAP value bee swarm chart, Figure 5)It can be seen that the factors with marginal contribution greater than 500 are water-rich degree (10315), karst, fault and fracture zone length (4819), hole number (3243), groundwater type (1652), permeability coefficient (1176), tunnel depth (650), and static water height (502).

[0060] Based on the results of Granger causality analysis, correlation analysis and importance evaluation, the construction method, hole number, groundwater-rich degree, groundwater type, fault, fracture zone length, static water height, permeability coefficient, and rainfall were finally selected as the key influencing factors of tunnel construction water inflow. A water inflow prediction model was established based on these eight factors.

[0061] Step 3: According to the key influencing factors of water inflow selected in step 2, different efficient prediction models were established using the Python programming language pycharm library. Before establishing the model, 15% of the sample data was randomly selected as the test set and the remaining 85% as the training set.

[0062] Firstly, in the sklearn library, RandomForestRegressor was configured and trained to establish a random forest (RF) prediction model. The grid search method was used to adjust the model parameters (the number of trees in the forest, the maximum depth of the tree, the minimum sample size of each node, and the minimum sample size of each leaf) to find the optimal parameter combination, achieving higher training set fitting effect and test set prediction effect. When the model parameter combination is the number of trees in the forest is 300, the maximum depth is 10, the minimum sample size of each node is 2, and the minimum sample size of each leaf is 1, the evaluation index is optimal.

[0063] Table 3 Influence of different parameters on evaluation index of random forest model

[0064]

[0065] Secondly, in the sklearn library, XGBRegressor was configured and trained to establish an XGBoost prediction model. The grid search method was used to adjust the model parameters (the number of decision trees, the maximum depth of the tree, the learning rate, and the sample sampling ratio) to find the optimal parameter combination, achieving higher training set fitting effect and test set prediction effect. When the model parameter combination is the number of decision trees is 100, the maximum depth is 9, the learning rate is 0.1, and the sample sampling ratio is 1, the evaluation index is optimal.

[0066] Table 4 Influence of different parameters on evaluation index of XGBoost model

[0067]

[0068] Finally, in the sklearn and Keras library, the MLPRegressor is configured and trained to establish a deep neural network (DNN) prediction model. The grid search method is used to adjust the model parameters (selection of optimizer, number of hidden layers, number of neurons in each layer, learning rate, activation function, batch size, maximum number of iterations) to find the optimal parameter combination, so as to achieve higher fitting effect of the training set and prediction effect of the test set. When using different parameter combinations, the evaluation indicators of the deep neural network (DNN) model are shown in Table 5. When the model parameter combination is: iteration number is 1000, number of hidden layers is 5, number of neurons is 128, learning rate is 0.01 and batch size is 32, the evaluation result is optimal.

[0069] Table 5 Influence of different parameters on neural network model evaluation indicators

[0070]

[0071] Compared with the random forest model and the XGBoost model, although the fitting effect of the training set of the deep neural network model is slightly worse, the prediction effect of the test set is better, and the evaluation indicators of the test set and the training set in the deep neural network model are close, which indicates that the model does not exist overfitting. In view of the randomness and difficulty of accurate prediction of tunnel construction water inrush, the deep neural network water inrush prediction model has better applicability and generalization than the random forest prediction model and the XGBoost prediction model.

[0072] Step 4: Use the deep neural network (DNN) prediction model under the optimal parameter combination to predict and verify the water inrush of C tunnel No. 4 cross tunnel (drilling and blasting method, cross tunnel into the main tunnel and flat guide, karst development construction scene) and F tunnel No. 2 inclined shaft (drilling and blasting method, inclined shaft not into the main tunnel and flat guide, fault zone development construction scene). The verification period is the water monitoring period of the work site. The minimum water inrush is obtained when the rainfall is the minimum (rainfall is zero), and the maximum water inrush is obtained when the average of the maximum daily rainfall in the past five years is the maximum rainfall. The verification results are shown in Figure 6 and Figure 7 .

[0073] The verification results show that the deep neural network prediction performance is good, the actual water inrush of C tunnel No. 4 cross tunnel and F tunnel No. 2 inclined shaft demonstration work site is generally within the prediction range, and the water inrush in karst and fault area also has good prediction effect, which can meet the actual demand.

[0074] Step 5: Based on the optimal deep neural network (DNN) prediction model, an intelligent tunnel construction water inflow prediction system is constructed through Python and MySQL database technology. The MySQL database is responsible for efficiently storing and managing various types of data, providing the foundation data for water inflow prediction. Python interacts with the MySQL database through the pymysql library, extracts historical data, and performs cleaning and preprocessing to ensure data quality and consistency. The processed data is input into the deep neural network (DNN) model based on Keras for training. Through this model, the system can identify factors related to water inflow and make accurate predictions based on input parameters.

[0075] This intelligent prediction system can automatically process various types of input data and generate real-time water inflow prediction results, assisting in construction site and drainage design decisions, optimizing construction plans, reducing risks, and improving construction efficiency and safety.

Claims

1. A method for predicting water inflow in sustainable tunnel construction based on machine learning techniques, characterized by, The steps are as follows: Step 1: Data collection and preprocessing; Collect pre-tunnel construction data, including geological exploration reports, drilling data, and actual water inflow data and corresponding construction progress data during the construction period; organize the original water inflow influencing factor dataset; use interpolation and IQR methods to check for missing values and outliers in the original water inflow influencing factor dataset, and perform descriptive statistical analysis to determine the distribution and fluctuation characteristics of each parameter; Step 2: Identify the main control factors of water inflow using multivariate statistical and machine learning methods; First, use Python programming language to perform Granger causality test on rainfall and water inflow based on the vector autoregressive model in the statsmodels library; determine the time interval for data resampling based on the lag time of significant causal relationships to balance time-varying and non-time-varying factors; Second, use the correlation analysis tool in SPSS 26.0 software to identify highly correlated factors with water inflow using the Spearman correlation coefficient; Finally, use Python programming language to build an initial water inflow prediction model based on the random forest model algorithm using the water inflow and water inflow influencing factor data obtained in step 1, and introduce feature permutation importance tools and SHAP tools to analyze the importance ranking of water inflow influencing factors and their marginal contributions to water inflow prediction results, and sort the importance of water inflow influencing factors based on feature importance scores and SHAP values; Step 3: Use Python programming language to build efficient water inflow prediction models based on random forest, XGBoost, and deep neural network model algorithms using the key influencing factors of water inflow selected in step 2, and evaluate and compare the performance of different efficient water inflow prediction models based on the performance evaluation parameters of the initial water inflow prediction model in step 2; Step 4: Determine the best water inflow prediction model by comparing the output performance of the training set and test set of each efficient water inflow prediction model in step 3 with the initial water inflow prediction model; set different hydrogeological conditions and engineering condition scenarios to verify the applicability of the best water inflow prediction model in different engineering construction environments and ensure its effectiveness and accuracy in practical applications; Step 5: Build a sustainable tunnel construction water inflow intelligent prediction system; use the mysql-connector-python library to achieve efficient connection between Python and MySQL, ensuring fast interaction with the best water inflow prediction model database; the sustainable tunnel construction water inflow intelligent prediction system inputs key influencing factor data into the best water inflow prediction model through database management for prediction; the prediction results are returned to the database and stored for subsequent query and analysis. 2.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, The specific implementation process of step one is as follows: The original water inflow influencing factors include two categories, namely engineering conditions and hydrogeological conditions. The engineering conditions include construction method, number of tunnel bodies, number of tunnel faces, construction section area, cumulative footage, daily footage, tunnel depth, and static water height. The hydrogeological conditions include water-rich degree, groundwater type, length of karst and fault crossing, surrounding rock grade, permeability coefficient, and infiltration coefficient. 3.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, The vector autoregressive model of the two time series X and Y (rainfall and water inflow) is as follows: where X t and Y t represent the observed values of time series X and Y at time t; p is the maximum number of lag periods set; i is the lag period, with a value range of 1-p; and and are coefficients in the model, representing the influence degree of X and Y at time t-i on X and Y at time t, respectively; and are error terms. 4.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, The Grange causality test is performed using F test. The null hypothesis is that there is no significant causality between the time series X and Y. If the F value is significant, the null hypothesis is rejected, and it is considered that there is a significant Granger causality between rainfall and water inflow. 5.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, In constructing the initial water inflow prediction model, to ensure that the data training effect does not appear overfitting phenomenon and to ensure the repeatability of the results, 80% of the data is used as the training set, and 20% of the data is used as the test set. The random seed number is set to 42. 6.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, The output performance evaluation parameters of the initial water inflow prediction model include mean absolute error, mean square error, root mean square error, and determination coefficient, which are calculated according to the following formulas: where n is the number of samples, y i is the measured value for each sample, is the average of the measured values of the indicator, is the predicted value obtained using the model. 7.The machine learning technique based sustainable tunneling inflow prediction method of claim 1, wherein, The feature permutation importance tool quantifies the influence of each influencing factor on the prediction ability of the initial water inflow prediction model by evaluating the difference between the performance of the initial water inflow prediction model on the original water inflow influencing factor data set and the performance after the influencing factors are randomly permuted. Using the feature permutation importance tool, the contribution of each influencing factor to the change of water inflow is evaluated based on the initial water inflow prediction model. The feature value permutation importance is sorted according to the feature permutation importance score. The SHAP tool is based on the Shapley value in game theory. It quantifies the contribution of each influencing factor to the change of water inflow by calculating the marginal contribution of each influencing factor. The contribution and importance ranking of each influencing factor to the change of water inflow are obtained from the swarm plot analyzed by the SHAP tool. In the swarm plot, blue represents a positive influence on the predicted water inflow, and red represents a negative influence on the predicted water inflow. The deeper the color, the higher the degree. The SHAP value is calculated according to the following formula: where: is the SHAP value for the kth influence factor; N is the set of all influence factors; S is the subset of features for influence factor k; f(S) is the model prediction for the subset S; f(S U {k}) is the model prediction for the subset of features in S plus feature The factors with a correlation coefficient greater than 0.3, a feature permutation importance greater than 0.03, and a SHAP value greater than 500 are determined as the key influencing factors of water inflow, providing a basis for establishing an efficient water inflow prediction model. 8.The machine learning technology based sustainable tunneling inflow prediction method of claim 1, wherein, The 85% of the key influencing factors and water inflow data are used as the training set, and the remaining 15% are used as the test set. The same data is used for training and testing all efficient water inflow prediction models. The grid search method is used to adjust the parameters of the efficient water inflow prediction model.

9. The machine learning technology-based sustainable tunnel construction water inflow prediction method according to claim 1, characterized in that, In the intelligent prediction system for sustainable tunnel construction water inflow, MySQL is used to store and manage various parameter data of tunnel construction, historical water inflow data, and prediction results of the best water inflow prediction model. The sustainable tunnel construction water inflow intelligent prediction embedded dataset will be continuously and dynamically updated, the universality and prediction accuracy of the model are improved by continuously increasing new training samples, and the prediction performance of the system is continuously optimized when the data volume increases.

Images