Hard rock surrounding rock deformation risk early warning method
By using the LSTM neural network model to conduct risk warning of surrounding rock deformation in hard rock tunnel construction, the problem of difficulty in accurately predicting tunnel deformation in the existing technology is solved, and higher prediction accuracy and construction safety are achieved.
Patent Information
- Application Number
- CN202510172633.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-17
- Publication Date
- 2025-06-06
AI Technical Summary
When tunnel construction is carried out in hard rock bodies, it is difficult for the prior art to accurately predict deformation, resulting in potential safety hazards.
A hard rock surrounding rock deformation risk warning method is adopted based on LSTM neural network. A model for prediction is built through preprocessing and timing prediction of monitoring data, and deformation warning standards are introduced for safety evaluation.
It improves the accuracy and timeliness of hard rock deformation prediction, effectively guides tunnel construction, optimizes engineering design and construction management, and improves project safety and economy.
Smart Images
Figure CN120106560A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of tunnel construction safety, and in particular relates to a hard rock mass surrounding rock deformation risk early warning method. Background Art
[0002] Compared with other types of engineering construction, tunnel engineering has its own unique complexity and challenges. The closed and changeable construction environment increases the difficulty of construction and puts higher requirements on engineering safety. Especially when constructing tunnels in hard rock, due to the hard characteristics of the rock itself, unexpected deformations are prone to occur during the excavation process. These deformations may cause damage to the tunnel structure and even cause serious safety accidents. The uncertainty of engineering geological conditions is one of the main risk factors. Rock type, fault distribution, groundwater activity, etc. will affect the stability of the tunnel. In addition, additional risks such as seismic activity and extreme climate brought by the geographical location and surrounding environment of the tunnel should not be ignored.
[0003] In order to effectively control such risks, researchers are committed to identifying and optimizing key factors that can significantly affect tunnel deformation. By adjusting support system parameters, optimizing blasting design, and improving construction methods, the degree of tunnel deformation can be effectively reduced. At the same time, using advanced monitoring technology and data analysis methods to predict tunnel deformation and introduce early warning indicators is one of the important means to improve the safety of tunnel construction. However, although there are many theoretical models and technical methods for evaluating tunnel stability at this stage, it is still difficult to fully and accurately predict the deformation of the tunnel after excavation in practical applications. This limitation brings potential safety hazards to the construction site. Summary of the invention
[0004] The purpose of the embodiment of the present invention is to provide a method for early warning of deformation risk of hard rock mass surrounding rock, so as to achieve accurate prediction of deformation of hard rock mass in advance, prevent engineering disasters such as sudden block falling and collapse of hard rock mass, and effectively ensure rapid and safe construction of hard rock mass.
[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is a method for early warning of deformation risk of hard rock mass surrounding rock, which is specifically carried out in the following steps:
[0006] S1. Determine the influencing parameters and perform data preprocessing based on the deformation monitoring data of the hard rock tunnel;
[0007] S2. Build and train an LSTM neural network model for time series prediction and make predictions;
[0008] S3. Based on the prediction results obtained in S3, a deformation warning standard is introduced to evaluate the safety of the predicted deformation, so as to predict the construction safety of the unconstructed section of the tunnel.
[0009] Furthermore, the specific process of S1 is as follows:
[0010] S101, determining an input sample set affecting tunnel settlement and deformation based on hard rock tunnel construction influencing parameters;
[0011] S102, normalizing the data of the input sample set that affects the tunnel settlement deformation, the normalization method is:
[0012]
[0013] In the formula, y is the normalized sample standard value, y max and min are the normalized maximum and minimum values of the influencing parameter samples that affect tunnel settlement deformation, usually 1 and 0, x is the sample value, x max and x min are the maximum and minimum values of the sample values of the influencing parameters that affect the tunnel settlement and deformation.
[0014] Furthermore, the influencing parameter determines that the input sample set affecting the tunnel settlement deformation includes the tunnel deformation x 1 , surrounding rock strength x 2 , cross-sectional size x 3 , Excavation footage x 4 , Construction step x 5 .
[0015] Furthermore, the specific process of S2 is as follows:
[0016] S201. Use Pandas to load the input sample set that affects tunnel settlement deformation collected on site, including tunnel deformation x 1 , surrounding rock strength x 2 , cross-sectional size x 3 , Excavation footage x 4 , Construction step x 5 Detailed data including the time series data are arranged in chronological order;
[0017] S202. During machine learning, the data of the set time step will be used as input to predict and output subsequent data.
[0018] S203, dividing the data set into a training set and a test set in a ratio of 8:2;
[0019] S204, converting the data format into the input format of the LSTM layer;
[0020] S205. Create an LSTM model using the pytorch deep learning framework;
[0021] S206, import the previously reshaped training and data into the constructed model for iterative training;
[0022] S207, using the trained model to predict the test set;
[0023] S208, performing denormalization on the prediction result;
[0024] S209: Evaluate the performance of the model.
[0025] Furthermore, the number of iterations in S205 is 200 to 400.
[0026] Furthermore, the denormalization formula in S208 is specifically:
[0027] y′=y×(x max -x min )+x min
[0028] In the formula, y' is the denormalized standard value, y is the normalized sample standard value, and x max and x min are the maximum and minimum values of the sample values of the influencing parameters that affect the tunnel settlement and deformation.
[0029] Furthermore, the performance evaluation includes the root mean square error RMSE and the goodness of fit R 2 ; Specifically:
[0030]
[0031]
[0032] Where n is the number of samples, y i The table shows the actual observed values; is the average value of the actual observations; f i is the predicted value, and i is the index variable of the sample.
[0033] Furthermore, the specific process of S3 is as follows:
[0034] S301, based on the existing deformation value U and the limit displacement value U 0 Constructing early warning classification index F based on cumulative deformation r :
[0035] F r =U / U 0
[0036] Among them, F r The larger the value, the greater the risk; conversely, the smaller the risk;
[0037] Among them, U 0 The fitting formula is:
[0038] k=U 0 e -B / t
[0039] Among them, k is the tunnel deformation value; B is the parameter to be fitted; t is the time parameter;
[0040] S302, early warning standard based on deformation rate;
[0041] S303. Establish comprehensive early warning standards for tunnel deformation.
[0042] Furthermore, the specific process of S302 is as follows:
[0043] The tunnel large deformation rate series is expressed as {X 1 , X 2 , …, X n}, and reorder them from large to small, and the new sequence after sorting is {Y 1 , Y 2 , …, Y n}, through the calculation of the two, the development trend evaluation index can be obtained, that is, the rank coefficient r s :
[0044]
[0045] R i For X i The rank of X i In the new sequence {Y 1 , Y 2 , …, Y n} is the sorted position; n is the number of samples; r s is the linear relationship between the order sequence and the ascending order of deformation rate; for repeated values in the original data, in the rank R i The average value should be taken as the standard;
[0046] Using the rank coefficient r s and the corresponding critical value W p The comparative analysis between the two can realize the evaluation of the development trend of the deformation rate. Specifically, when |r s |<W p When r is , it indicates that the development trend of the large deformation rate of the tunnel is not obvious; on the contrary, it indicates that its change trend is more significant, and when r is s When it is greater than 0, it indicates that its development has an increasing trend; otherwise, it has a decreasing trend;
[0047] The critical value W p The specific determination method is:
[0048] When n≤50, check the Spearman rank correlation coefficient cutoff table; when n>50, calculate r s The t value and look up the t value table:
[0049]
[0050] The beneficial effects of the present invention are:
[0051] The current prediction of surrounding rock deformation mainly focuses on factors such as tunnel construction methods and surrounding rock grades, without fully considering the size effect unique to hard rock and the coupling relationship between its strength parameters and construction dynamic parameters. This neglect leads to limitations in the selection of existing deformation prediction models, which in turn causes problems of prediction lag and insufficient fitting accuracy. The present invention targets the characteristics of hard rock, comprehensively considers its deformation properties, and combines the dynamic change characteristics during tunnel construction to construct a highly matched prediction model. The present invention not only improves the accuracy and timeliness of deformation prediction, but also can effectively guide tunnel construction practices under hard rock conditions, thereby optimizing engineering design and construction management, and improving engineering safety and economy. By introducing the coupling analysis between rock size effects and strength parameters and construction dynamic parameters, this model provides new ideas and technical means for solving the problems existing in traditional prediction methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 paying creative work.
[0053] Figure 1 It is a flow chart of the hard rock surrounding rock deformation risk early warning method based on LSTM neural network of the present invention. DETAILED DESCRIPTION
[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0055] like Figure 1 The embodiment of the present invention provides a hard rock mass surrounding rock deformation risk early warning method; in particular, it relates to a hard rock mass surrounding rock deformation risk early warning method based on LSTM neural network; it comprises the following steps:
[0056] S1: Based on the deformation monitoring data of hard rock tunnels, determine the influencing parameters and perform data preprocessing;
[0057] (1) Determine the input sample set that affects tunnel settlement and deformation based on the influencing parameters of hard rock tunnel construction;
[0058] There are four main influencing parameters in the construction process of hard rock tunnels, namely deformation parameters, surrounding rock parameters, geometric parameters and excavation parameters. This implementation method determines the input sample set that affects the tunnel settlement deformation based on the above influencing parameters: tunnel deformation (x 1 ), surrounding rock strength (x 2 ), cross-sectional size (x 3 )、Excavation footage(x 4 ), construction step (x 5 ).
[0059] (2) Data preprocessing
[0060] Before training the model, it is necessary to normalize the data of the input sample set of the influencing parameters that affect the tunnel settlement deformation to eliminate the influence of the eigenvalue dimension of different samples on the prediction efficiency and accuracy. This implementation method normalizes the data samples of different targets to the interval [0, 1] so that the input variables can be directly compared to avoid the influence of different magnitudes of different targets on subsequent optimization. The normalization method is as follows:
[0061]
[0062] In the formula, y is the normalized sample standard value, y max and min are the normalized maximum and minimum values of the influencing parameter samples that affect tunnel settlement deformation, usually 1 and 0, x is the sample value, x max and x min are the maximum and minimum values of the sample values of the influencing parameters that affect the tunnel settlement and deformation.
[0063] S2: Build an LSTM neural network model for time series prediction to train the model on input data;
[0064] (1) Import necessary tool libraries
[0065] Make sure all libraries required for data training are installed, such as NumPy, Pandas, Matplotlib, and TensorFlow or PyTorch;
[0066] (2) Loading time series data
[0067] Use Pandas to load the input sample set that affects tunnel settlement deformation collected on site, including tunnel deformation (x 1 ), surrounding rock strength (x 2 ), cross-sectional size (x 3 )、Excavation footage(x 4 ), construction step (x 5 ) are organized into time series data in chronological order.
[0068] (3) Create an observation window
[0069] By setting the time step, the time series data is converted into a data-driven supervised learning problem. During machine learning, the data of the set time step will be used as input to predict the output of the following data.
[0070] (4) The data set is divided into a training set and a test set in a ratio of 8:2;
[0071] The data is divided into two parts, one for training and the other for testing to verify the ability of the model.
[0072] (5) Reshape the format of input data
[0073] Convert the data format to the input format of the LSTM layer; in this embodiment, the input data format is a 2D array. For the LSTM neural network, the network requires the input data shape to be a 3D array of (input, time step, output). Therefore, in order to meet the requirements of the LSTM neural network, the input data format needs to be converted from a 2D array (number of samples, number of features) to a 3D array.
[0074] (6) Build LSTM model:
[0075] Create an LSTM model using the pytorch deep learning framework.
[0076] In this implementation, the deep learning framework is the basis for building the LSTM model. The LSTM model trains the input data to obtain predicted deformation, thereby realizing the function of tunnel deformation warning. The input of the model is the surrounding rock strength (x2), the cross-section size (x3), the excavation progress (x4), and the construction step (x5), and the output is the tunnel deformation (x1).
[0077] (7) Training model
[0078] Import the previously reshaped training and data into the constructed model for training.
[0079] In some specific implementations, the training iterations are 200 to 400 times, which can be set according to design requirements.
[0080] (8) Make predictions
[0081] Use the trained model to make predictions on the test set.
[0082] (9) Denormalization or Destandardization
[0083] If the data is normalized or standardized, the prediction results need to be reversed to the original range. The denormalization formula is as follows:
[0084] y′=y×(x max -x min )+x min
[0085] In the formula, y' is the denormalized standard value, y is the normalized standard value, and x max and x min are the maximum and minimum sample values.
[0086] (10) Model performance evaluation
[0087] The prediction model obtained by the LSTM algorithm also needs to evaluate the prediction accuracy of the model, so the root mean square error RMSE and goodness of fit R are introduced 2 To test the accuracy of the model. Among them, RMSE is a measure of the deviation between the true value and the predicted value. The better the model fitting effect, the closer its value is to 0; R 2 It is used to measure the discreteness of the sample. Its value range is 0 to 1. The better the model fitting effect, the closer its value is to 1. Each indicator is calculated by the following formula:
[0088]
[0089] Among them, n is the number of samples, and yi is the actual observed value; is the average value of the actual observations; fi is the predicted value, and i is the index variable of the sample.
[0090] S3. Based on the surrounding rock deformation prediction results of the prediction model, the deformation warning standard is introduced to evaluate the safety of the predicted deformation, so as to predict the construction safety of the unconstructed section of the tunnel.
[0091] In this embodiment, the early warning standards include an early warning standard based on cumulative deformation, an early warning standard based on deformation rate, and a comprehensive early warning standard for tunnel deformation, which are as follows:
[0092] (1) Early warning standard based on cumulative deformation
[0093] This standard uses the limit displacement criterion as the basis for judgment. Using displacement to judge the stability of the tunnel is to start from the various limit states that appear in the tunnel and find out the displacement value of each control point under a certain limit state, that is, the so-called limit displacement value U0 When the tunnel stability is higher, it proves that the tunnel has a relatively larger deformation margin. In terms of the performance of the monitored displacement value, the existing deformation value U will be higher than the limit displacement value U. 0 Therefore, based on the existing deformation value U and the limit displacement value U 0 Constructing early warning classification index F based on cumulative deformation r :
[0094] F r =U / U 0
[0095] The classification index F of the above formula r It can be seen that F r The larger the value, the greater the risk; conversely, the smaller the risk.
[0096] Since the existing deformation value U can be obtained through the statistics of the monitoring results of the surrounding rock deformation on site, it is desirable to obtain the early warning classification index F based on the cumulative deformation. r , then we need to find the limit displacement value U 0 ; Limit displacement value U 0 Exponential regression is often used to solve the problem, and the fitting formula is:
[0097] k=Ae -B / t
[0098] k is the tunnel deformation value; A and B are the parameters to be fitted; t is the time parameter. When the time parameter t approaches infinity, the tunnel deformation value approaches the constant value A, and this value is the maximum value, so it can be used as the ultimate displacement value U of the tunnel deformation. 0 .
[0099] Right now:
[0100] k=U 0 e -B / t
[0101] Therefore, the early warning classification standard based on cumulative deformation for the Mabaishan hard rock tunnel is shown in Table 1:
[0102] Table 1 Cumulative deformation warning classification standards
[0103]
[0104] (2) Early warning criteria based on deformation rate
[0105] This standard is mainly based on the deformation rate, and its focus is on evaluating the development trend of the tunnel deformation rate. First, the tunnel large deformation rate sequence is expressed as {X 1 , X 2 , …, X n}, and reorder them from large to small, and the new sequence after sorting is {Y 1 , Y 2 , …, Y n}, through the calculation of the two, the development trend evaluation index can be obtained, that is, the rank coefficient r s :
[0106]
[0107] R i For X i The rank of X i In the new sequence {Y 1 , Y 2 , …, Y n} is the sorted position; n is the number of samples; r s is the linear relationship between the order sequence and the ascending order of deformation rate. For the repeated values in the original data, in the rank R i The average value should be taken as the standard.
[0108] Using the rank coefficient r s and the corresponding critical value W p The comparative analysis between the two can realize the evaluation of the development trend of the deformation rate. The criterion is: when |r s |<W p When r is , it indicates that the development trend of the large deformation rate of the tunnel is not obvious; on the contrary, it indicates that its change trend is more significant, and when r is s When it is greater than 0, it indicates that its development has an increasing trend; otherwise, it has a decreasing trend.
[0109] Since the critical value W p It is related to the test level. That is, when the test level α is inconsistent, the trend strength of the judgment result is also different. When the number of sequence samples n is less than or equal to 50, the critical value W under different test levels α is determined by consulting the Spearman rank correlation coefficient boundary value table. p , as shown in Table 2, the test level α can be used to divide the development trend of tunnel deformation rate; when n is greater than 50, the rank coefficient r is calculated s The t-value can be determined by using the calculation method of the t-value in the Pearson correlation coefficient hypothesis test, as shown in the following formula, and by referring to the t-value table according to the test level.
[0110]
[0111] At this time, we need to use the rank coefficient r s The critical value W is determined by the t value table p The development trend of the deformation rate is evaluated by comparison. The evaluation method is the same as the above method.
[0112] Since the monitoring frequency of hard rock tunnels is not too frequent, the critical value W determined by the table lookup method when n=50 is selected. p The early warning standards are divided mainly based on deformation rate. The specific classification standards are shown in Table 3.
[0113] Table 2 Spearman rank correlation coefficient cutoff table
[0114]
[0115]
[0116] Table 3 Deformation rate development trend classification standard
[0117]
[0118] Among them, f 1 is the critical value corresponding to the significance level of 0.05, which is 0.297; f 2 It is the critical value corresponding to the significance level of 0.01, which is 0.363.
[0119] (3) Comprehensive early warning standard for tunnel deformation
[0120] The specific implementation methods of the two types of warning criteria constructed above are integrated and analyzed, and the warning levels are comprehensively divided. At the same time, based on the warning experience in previous engineering practices, the warning levels are generally divided into four levels. Referring to the engineering experience, the tunnel surrounding rock deformation warning is designed as a four-level warning division. The specific standards are shown in Table 4.
[0121] Table 4 Classification standards for comprehensive warning levels of tunnel deformation
[0122]
[0123] In the process of determining the warning level, the final warning level is determined according to the most unfavorable principle of different criteria.
[0124] The model established by the present invention takes into account the rock mass size effect and strength effect, and can evaluate the secondary crack development characteristics of hard rock mass under the action of blasting power and the residual strength degradation behavior of the disturbed weakened area, and then quantify its impact on the stability of the surrounding rock. By coupling the dynamic construction process of tunnel excavation, the present invention constructs a mechanical response characterization framework for the full deformation stage of hard rock mass, and realizes multi-dimensional dynamic monitoring and early warning of surrounding rock deformation risks.
[0125] On this basis, this implementation method establishes a comprehensive early warning level classification standard based on deformation criteria, converting surrounding rock deformation characteristics (such as displacement rate, strain gradient, and crack extension index, etc.) into quantifiable evaluation indicators. By integrating real-time monitoring data and prediction models, the system can dynamically output deformation criterion evaluation indicators and provide graded response thresholds for on-site technicians. This mechanism supports accurate decision-making on construction measures, significantly improves the timeliness and intelligence level of surrounding rock deformation management, and optimizes the safety control process of tunnel engineering under complex geological conditions.
[0126] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0127] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims
1. A method for early warning of deformation risk of hard rock mass surrounding rock, characterized in that: Follow these steps: S1. Determine the influencing parameters and perform data preprocessing based on the deformation monitoring data of the hard rock tunnel; S2. Build and train an LSTM neural network model for time series prediction and make predictions; S3. Based on the prediction results obtained in S3, a deformation warning standard is introduced to evaluate the safety of the predicted deformation, so as to predict the construction safety of the unconstructed section of the tunnel.
2. The method for early warning of deformation risk of hard rock mass surrounding rock according to claim 1, characterized in that: The specific process of S1 is as follows: S101, determining an input sample set affecting tunnel settlement and deformation based on hard rock tunnel construction influencing parameters; S102, normalizing the data of the input sample set that affects the tunnel settlement deformation, the normalization method is: In the formula, y is the normalized sample standard value, y max and min are the normalized maximum and minimum values of the influencing parameter samples that affect tunnel settlement deformation, usually 1 and 0, x is the sample value, x max and x min are the maximum and minimum values of the sample values of the influencing parameters that affect the tunnel settlement and deformation.
3. A hard rock mass surrounding rock deformation risk early warning method according to claim 2, characterized in that: The influencing parameters determine that the input sample set affecting the tunnel settlement deformation includes tunnel deformation x1, surrounding rock strength x2, section size x3, excavation footage x4, and construction step distance x5.
4. The method for early warning of deformation risk of hard rock mass surrounding rock according to claim 1, characterized in that: The specific process of S2 is: S201. Use Pandas to load the input sample set that affects tunnel settlement deformation collected on site, including detailed data such as tunnel deformation x1, surrounding rock strength x2, cross-section size x3, excavation footage x4, and construction step distance x5, and form time series data in chronological order. S202. During machine learning, the data of the set time step will be used as input to predict and output subsequent data. S203, dividing the data set into a training set and a test set in a ratio of 8:2; S204, converting the data format into the input format of the LSTM layer; S205. Create an LSTM model using the pytorch deep learning framework; S206, import the previously reshaped training and data into the constructed model for iterative training; S207, using the trained model to predict the test set; S208, performing denormalization on the prediction result; S209: Evaluate the performance of the model.
5. A hard rock mass surrounding rock deformation risk early warning method according to claim 4, characterized in that: The number of iterations in S205 is 200 to 400.
6. A hard rock mass surrounding rock deformation risk early warning method according to claim 4, characterized in that: The denormalization formula in S208 is specifically: y’=y×(x max -x min )+x min In the formula, y' is the denormalized standard value, y is the normalized sample standard value, and x max and x min are the maximum and minimum values of the sample values of the influencing parameters that affect the tunnel settlement and deformation.
7. The method for early warning of deformation risk of hard rock mass surrounding rock according to claim 4, characterized in that: The performance evaluation includes root mean square error (RMSE) and goodness of fit (R). 2 ; Specifically: Where n is the number of samples, y i The table shows the actual observed values; is the average value of the actual observations; f i is the predicted value, and i is the index variable of the sample.
8. A hard rock mass surrounding rock deformation risk early warning method according to claim 1; characterized in that: The specific process of S3 is as follows: S301. Construct an early warning classification index F based on the accumulated deformation based on the existing deformation value U and the limit displacement value U0. r : F r =U / U0 Among them, F r The larger the value, the greater the risk; conversely, the smaller the risk; Among them, the fitting formula of U0 is: k=U0e -B / t Among them, k is the tunnel deformation value; B is the parameter to be fitted; t is the time parameter; S302, early warning standard based on deformation rate; S303. Establish comprehensive early warning standards for tunnel deformation.
9. A method for early warning of deformation risk of hard rock mass surrounding rock according to claim 8, characterized in that: The specific process of S302 is as follows: The tunnel large deformation rate sequence is represented as {X1, X2, ..., X n }, and reorder them from large to small, and the new sequence after sorting is {Y1, Y2, ..., Y n }, through the calculation of the two, the development trend evaluation index can be obtained, that is, the rank coefficient r s : R i For X i The rank of X i In the new sequence {Y1, Y2, ..., Y n } is the sorted position; n is the number of samples; r s is the linear relationship between the order sequence and the ascending order of deformation rate; for repeated values in the original data, in the rank R i The average value should be taken as the standard; Using the rank coefficient r s and the corresponding critical value W p The comparative analysis between the two can realize the evaluation of the development trend of the deformation rate. Specifically, when |r s |<W p When r is , it indicates that the development trend of the large deformation rate of the tunnel is not obvious; on the contrary, it indicates that its change trend is more significant, and when r is s When it is greater than 0, it indicates that its development has an increasing trend; On the contrary, it has a decreasing trend; The critical value W p The specific determination method is: When n≤50, check the Spearman rank correlation coefficient cutoff table; when n>50, calculate r s The t value and look up the t value table: