Prediction method of water and mud inrush in fault fracture zone based on comprehensive advanced geological prediction

Abstract

The present invention provides a method for predicting water gushing and mud bursting in a fault fracture zone based on comprehensive advanced geological prediction, which relates to the technical field of geological disaster prediction. The method predicts the scale of the fault fracture zone by using a comprehensive prediction model for the fault fracture zone based on polynomial naive Bayes according to the possibility level of the existence of the fault fracture zone, the level of the adjacent precursors, the level of the rock mass integrity, the level of the drilling situation, the level of the seismic wave reflection method and the level of the geological radar method, thereby obtaining the scale of the fault fracture zone. If there is no fault fracture zone, there is no water gushing and mud bursting. If there is a fault fracture zone, the fault gushing and mud bursting level prediction model for the fault fracture zone based on genetic algorithm and support vector machine is used to predict the water gushing and mud bursting level of the fault fracture zone in combination with the water conductivity level of the fault fracture zone, the level of the rock mass integrity level, the level of the groundwater development level and the level of the surface water recharge situation. This solves the problem that the prior art cannot predict the level of water gushing and mud bursting in the fault fracture zone.

Description

Technical Field

[0001] The invention relates to the technical field of geological disaster prediction, and in particular to a method for predicting water and mud bursts in a fault fracture zone based on comprehensive advanced geological prediction. Background Art

[0002] Tunnel advanced geological prediction refers to the prediction of the location, scale and nature of possible adverse geological conditions and the geological disasters caused by them during the tunnel construction phase through ground geological surveys, face geological surveys, advanced horizontal drilling and advanced geophysical prospecting. At present, there are many methods for tunnel advanced geological prediction, which can be mainly divided into two categories: geological analysis method and geophysical method. The geological analysis method mainly infers the geological conditions ahead of the tunnel through geological surveys inside and outside the tunnel, combined with engineering geological and hydrogeological conditions. It can only achieve macroscopic and qualitative predictions, and usually needs to be combined with geophysical methods for comprehensive prediction.

[0003] Different advanced geophysical exploration methods have their own theoretical basis, and the methods of use and the geophysical parameters obtained are also different. Therefore, domestic and foreign scholars have conducted a lot of research on the interpretation and analysis of advanced geophysical exploration methods. The engineering geological conditions in the tunnel site are usually diverse and complex. Therefore, it is difficult to accurately predict the fault fracture zone by relying on a single prediction method. In order to make up for this deficiency, many domestic and foreign scholars have proposed the concept of "comprehensive advanced geological prediction" and have achieved certain research results.

[0004] As the most common geological disaster during tunnel construction, water and mud bursts are mainly caused by the disturbance of rock caused by tunnel excavation, the destruction of rock mass and the combined effect of abundant groundwater. The biggest cause of water and mud bursts is water. When a tunnel passes through a fault fracture zone with poor geological bodies that are prone to water storage, water and mud bursts are very likely to occur. In order to avoid the occurrence of water and mud burst disasters and reduce the consequences, scholars at home and abroad have adopted a series of theories and methods to predict water and mud burst disasters during tunnel construction.

[0005] Many scholars at home and abroad have made a lot of research results on the prediction of water and mud inrush disasters in fault and broken zones of tunnels, which has reduced the occurrence of water and mud inrush disasters in fault and broken zones of tunnels to a certain extent and ensured the safety of tunnel construction. However, due to the limitations of technical means and research methods, there are still some shortcomings:

[0006] (1) Existing comprehensive advanced geological prediction technologies mainly use two or more prediction methods to predict potential fault fracture zones. There are few studies on further predicting tunnel water and mud inrush disasters based on the prediction results. Moreover, most of the prediction results can only determine whether there is a fault fracture zone, but cannot accurately predict its scale or nature.

[0007] (2) In terms of water and mud burst prediction, although corresponding prediction index systems have been established based on geological analysis or case statistics, the consideration of fault fracture zones mostly remains at the level of existence or non-existence, and the prediction results are mainly based on the probability or risk level of disaster occurrence, and the scale of water and mud burst cannot be predicted;

[0008] (3) In the application of machine learning algorithms, although some studies have taken into account the independence of categorical variables, data missing, and sample imbalance during the training phase, they have not conducted in-depth research on the impact of data set partitioning on model performance. There is also little research on the problem of missing data in the application phase of the prediction model, and the actual application effect is not ideal. Summary of the invention

[0009] Technical problem solved by the present invention: The present invention provides a method for predicting water gushing and mud bursting in a fault fracture zone based on comprehensive advanced geological prediction, which solves the problem that the prior art cannot predict the level of water gushing and mud bursting in a fault fracture zone.

[0010] The present invention solves the above technical problems by adopting a technical solution: a method for predicting water and mud bursts in fault fracture zones based on comprehensive advanced geological prediction, comprising the following steps:

[0011] S1. Collecting and preprocessing the input data for fault fracture zone prediction; the input data for fault fracture zone prediction includes the possibility level of the existence of the fault fracture zone, the level of the near precursor, the level of the rock mass integrity, the level of the drilling situation, the level of the seismic wave reflection method and the level of the geological radar method;

[0012] S2, input the pre-processed fault fracture zone prediction input data into the fault fracture zone comprehensive prediction model based on polynomial naive Bayes to obtain the predicted fault fracture zone scale level. If there is no fault fracture zone, there is no water inrush and mud inrush. If there is a fault fracture zone, enter S3;

[0013] S3, collecting and preprocessing the water gushing and mud bursting prediction input data; the water gushing and mud bursting prediction input data includes the fault fracture zone water conductivity grade, rock mass integrity grade, groundwater development grade, surface water recharge grade and the predicted fault fracture zone scale grade;

[0014] S4. Input the preprocessed water gushing and mud burst prediction input data into a fault fracture zone water gushing and mud burst grade prediction model based on genetic algorithm and support vector machine to obtain the predicted fault fracture zone water gushing and mud burst grade.

[0015] Furthermore, the preprocessing includes feature encoding, and the feature encoding includes one-hot encoding.

[0016] Furthermore, the fault rupture zone comprehensive prediction model based on polynomial naive Bayes is trained by collecting existing fault rupture zone sample data, wherein the fault rupture zone sample data includes fault rupture zone prediction input data and fault rupture zone scale level.

[0017] Furthermore, before training, the existing fault zone sample data are processed for missing values, sample imbalance and feature encoding.

[0018] Furthermore, the fault fracture zone water and mud inrush grade prediction model based on genetic algorithm and support vector machine is trained by collecting existing water and mud inrush sample data, wherein the water and mud inrush sample data includes water and mud inrush prediction input data and the fault fracture zone water and mud inrush grade.

[0019] Furthermore, before training, the existing water and mud burst sample data are processed for missing values, sample imbalance and feature encoding.

[0020] Furthermore, the missing value processing includes deletion method, manual filling method, special value filling method, interpolation filling method and prediction filling method; the sample imbalance processing includes sampling, category weighting and generating synthetic samples, and the generating synthetic samples uses SMOTE algorithm; the feature encoding uses one-hot encoding.

[0021] Furthermore, the possibility level of fault shattering zone is divided according to the signs of the existence of fault shattering zone; the level of proximity precursors is divided according to the number and obviousness of precursors; the level of rock integrity is divided according to rock structure, number of rock volume joints and rock integrity index; the level of drilling conditions is divided according to the number of drilling anomalies; the level of seismic wave reflection method is divided according to the difference between the interpretation result and the longitudinal wave velocity of the predicted mileage section; the level of geological radar method is divided according to the interpretation result and the two-way travel time; the level of fault shattering zone scale is divided according to the width of the fault shattering zone; the level of fault shattering zone water conductivity is divided according to the compactness of the fault shattering zone and the nature of the internal filling; the level of groundwater development is divided according to the level of groundwater development; the level of surface water recharge is divided according to the surface water recharge; the level of water and mud inrush in the fault shattering zone is divided according to the amount of water and mud inrush.

[0022] The beneficial effects of the present invention are as follows: the present invention provides a method for predicting water gushing and mud bursting in a fault fracture zone based on comprehensive advanced geological prediction. The fault fracture zone scale level is predicted by using a comprehensive prediction model for the fault fracture zone based on polynomial naive Bayes according to the possibility level of the existence of the fault fracture zone, the level of nearby precursors, the level of rock integrity, the level of drilling conditions, the level of seismic wave reflection method and the level of geological radar method, thereby obtaining the scale of the fault fracture zone. If there is no fault fracture zone, there is no water gushing and mud bursting. If there is a fault fracture zone, the fault gushing and mud bursting level is predicted by using a fault gushing and mud bursting level prediction model based on genetic algorithm and support vector machine in combination with the level of water conductivity of the fault fracture zone, the level of rock integrity, the level of groundwater development and the level of surface water recharge conditions. This solves the problem that the prior art cannot predict the level of water gushing and mud bursting in a fault fracture zone. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic flow chart of a method for predicting water gushing and mud bursting in fault fracture zones based on comprehensive advanced geological prediction provided by the present invention. DETAILED DESCRIPTION

[0024] Aiming at the problem that the prior art cannot predict the level of water inrush and mud inrush in fault fracture zones, the present invention provides a method for predicting water inrush and mud inrush in fault fracture zones based on comprehensive advanced geological prediction, such as Figure 1 As shown, the following steps are included:

[0025] S1. Collecting and preprocessing the input data for fault fracture zone prediction; the input data for fault fracture zone prediction includes the possibility level of the existence of the fault fracture zone, the level of the near precursor, the level of the rock mass integrity, the level of the drilling situation, the level of the seismic wave reflection method and the level of the geological radar method;

[0026] S2, input the pre-processed fault fracture zone prediction input data into the fault fracture zone comprehensive prediction model based on polynomial naive Bayes to obtain the predicted fault fracture zone scale level. If there is no fault fracture zone, there is no water inrush and mud inrush. If there is a fault fracture zone, enter S3;

[0027] S3, collecting and preprocessing the water gushing and mud bursting prediction input data; the water gushing and mud bursting prediction input data includes the fault fracture zone water conductivity grade, rock mass integrity grade, groundwater development grade, surface water recharge grade and the predicted fault fracture zone scale grade;

[0028] S4. Input the preprocessed water gushing and mud burst prediction input data into a fault fracture zone water gushing and mud burst grade prediction model based on genetic algorithm and support vector machine to obtain the predicted fault fracture zone water gushing and mud burst grade.

[0029] Therefore, the method for predicting water gushing and mudslides in fault fracture zones based on comprehensive advanced geological prediction provided by the present invention first obtains the predicted scale grade of the fault fracture zone through a comprehensive prediction model of the fault fracture zone based on polynomial naive Bayes; if a fault fracture zone exists, the predicted water gushing and mudslide grade of the fault fracture zone is obtained by using a prediction model of water gushing and mudslide grade of the fault fracture zone based on genetic algorithm and support vector machine, thereby solving the problem that the prior art cannot predict the water gushing and mudslide grade of the fault fracture zone.

[0030] Furthermore, the preprocessing includes feature encoding, and the feature encoding includes one-hot encoding. The one-hot encoding is converted into a format suitable for processing by a machine learning algorithm, which effectively eliminates the size relationship between categories and also expands the feature space.

[0031] Furthermore, a comprehensive fault rupture zone prediction model based on polynomial naive Bayes is trained by collecting existing fault rupture zone sample data, wherein the fault rupture zone sample data includes fault rupture zone prediction input data and fault rupture zone scale level. During training, the model prediction accuracy is used as the goal of parameter optimization, and the optimization range of the hyperparameter is set to [0, 1], the step size is 0.0001, and 10-fold cross validation is used to evaluate the model prediction accuracy. The hyperparameter with the highest accuracy is selected as the optimal hyperparameter, such as 0.8631, and the corresponding model is determined to be the optimal model.

[0032] Furthermore, a fault fracture zone water and mud inrush grade prediction model based on genetic algorithm and support vector machine is trained by collecting existing water and mud inrush sample data, wherein the water and mud inrush sample data include water and mud inrush prediction input data and the fault fracture zone water and mud inrush grade. During model training, the support vector machine model has two hyperparameters, namely, a penalty coefficient and a radial basis kernel function parameter. The penalty coefficient is used to balance the weights of classification intervals and misclassified samples, and the radial basis kernel function parameter affects the distribution of data after being mapped to high-dimensional space. Typically, the penalty coefficient ranges from 0.1 to 100, and the radial basis kernel function parameter ranges from 0. .0001 to 1. By adjusting the random seed when dividing the data, the overall training effect of the model can be improved. Therefore, the optimization range of the random seed is set to 0 to 100, the step size is 1, the population size of the genetic algorithm is set to 100, the maximum number of iterations is 100, the mutation probability is 0.1, the crossover probability is 0.8, and the optimization range is: the optimization range of the penalty coefficient is 0.1 to 100, and the optimization range of the radial basis kernel function parameter is 0.0001 to 1. The best random seed is 82, the penalty coefficient is 9.5335, the radial basis kernel function parameter is 0.0565, and the prediction accuracy is 85.19%.

[0033] Furthermore, for the existing fault fracture zone sample data and the existing water gushing and mudslide sample data collected through literature collection and network retrieval, there are some data missing and sample imbalance problems. Therefore, before training, the existing fault fracture zone sample data and the existing water gushing and mudslide sample data are all subjected to missing value processing, sample imbalance processing and feature encoding. The missing value processing includes deletion method, manual filling method, special value filling method, interpolation filling method and prediction filling method; the sample imbalance processing includes sampling, category weighting and generating synthetic samples, and the generating synthetic samples uses the SMOTE algorithm; the feature encoding adopts one-hot encoding.

[0034] Specifically, the possibility level of the existence of a fault fracture zone is divided according to the signs of the existence of the fault fracture zone, and the signs of the existence of the fault fracture zone include stratigraphic signs, fault surface signs, rock signs, structural signs and geomorphic signs. Stratigraphic signs: mainly manifested as the absence, duplication of strata or sudden interruption of strata in the same layer along the stratum dip. Stratigraphic signs are the inevitable signs of the existence of a fault. Fault surface signs: The fault surface is the geological structural surface along which the two plates of the fault are displaced and dislocated. The exposure of the main section and the fault scratches are the inevitable signs of the existence of a fault. Rock signs: that is, fault structural rock signs, such as structural breccia, breccia, mylonite, etc. The structural rocks near the fault surface and the fault fracture zone composed of them are the inevitable signs of the existence of a fault. Structural signs: A series of small structures such as traction folds, dense joint zones and disappearance of stratigraphic bedding often appear on both sides of the fault surface, which are reliable signs of the existence of a fault. Geomorphic signs: negative geomorphic features such as valleys and ridge saddles, but not all fault zones appear as negative geomorphic features. Negative geomorphic features are only reference signs of the existence of faults. Therefore, the possibility of the existence of fault zones is divided into the following levels: the first level, there is no sign of the existence of fault zones; the second level, there are only geomorphic signs or only structural signs; the third level, there are geomorphic signs and any other signs, or there are structural signs and any other signs; the fourth level, there are any three or more signs.

[0035] The level of near precursors is divided according to the number and obviousness of precursors. There are five types of precursors of fault fracture zones. The first type is that the number of joint groups increases sharply. When approaching the fault fracture zone, the number of joint groups can reach 6 to 12 groups; the second type is the traction folds and traction wrinkles of the rock strata; the third type is the small spiral structure or reverse joints composed of arc joints; the fourth type is the zone with obvious reduction in rock strength; the fifth type is the appearance of fractured rocks and shattered rocks near the reverse faults, and the joint density represented by joints increases significantly near the strike-slip faults. Therefore, the results of the classification of near precursor levels are as follows: the first level, no precursors are found; the second level, there is one precursor and the phenomenon is not obvious; the third level, there are two precursors and only one precursor phenomenon is obvious; the fourth level, there are two precursors and the phenomenon is obvious, or there are three or more precursors.

[0036] The rock mass integrity level is divided according to the rock mass structure, the number of rock mass volume joints and the rock mass integrity index. With reference to the Engineering Rock Mass Classification Standard (GB 50218-2014), the rock mass integrity level is divided as follows: the first level, the rock mass structure is an integral structure, the number of rock mass volume joints is less than 3, and the rock mass integrity index is greater than 0.75; the second level, the rock mass structure is a blocky structure, the number of rock mass volume joints is between 3 and 10, and the rock mass integrity index is between 0.55 and 0.75; the third level, the rock mass structure is a layered structure, the number of rock mass volume joints is between 10 and 20, and the rock mass integrity index is between 0.35 and 0.55; the fourth level, the rock mass structure is a fragmented structure, the number of rock mass volume joints is between 20 and 35, and the rock mass integrity index is between 0.15 and 0.35; the fifth level, the rock mass structure is a loose structure, the number of rock mass volume joints is not less than 35, and the rock mass integrity index is not greater than 0.15.

[0037] The drilling situation level is divided according to the number of drilling anomalies. There are three types of drilling anomalies: the first is that the drilling speed and torque increase significantly, and the rotation speed decreases; the second is that the flushing fluid is turbid and mixed with mud, debris, breccia, etc.; the third is the occurrence of drill sticking. The drilling situation level division results are: the first level, there is no drilling anomaly; the second level, there is one drilling anomaly and the phenomenon is not obvious; the third level, there are two drilling anomalies and only one drilling anomaly is obvious; the fourth level, there are two drilling anomalies and the phenomenon is obvious, or there are three drilling anomalies.

[0038] The level of the seismic wave reflection method is divided according to the difference between the interpretation result and the longitudinal wave velocity of the predicted mileage section. The results of the seismic wave reflection method are divided as follows: the first level, the interpretation result is no; the second level, the interpretation result is yes, and the difference in the longitudinal wave velocity of the predicted mileage section is less than 281; the third level, the interpretation result is yes, and the difference in the longitudinal wave velocity of the predicted mileage section is between 281 and 485; the fourth level, the interpretation result is yes, and the difference in the longitudinal wave velocity of the predicted mileage section is between 485 and 671; the fifth level, the interpretation result is yes, and the difference in the longitudinal wave velocity of the predicted mileage section is greater than 671.

[0039] The levels of the geological radar method are divided according to the interpretation results and the two-way travel time. The results of the geological radar method are as follows: the first level, the interpretation result is no; the second level, the interpretation result is yes, and the two-way travel time is less than 160; the third level, the interpretation result is yes, and the two-way travel time is between 160 and 220; the fourth level, the interpretation result is yes, and the two-way travel time is between 220 and 260; the fifth level, the interpretation result is yes, and the two-way travel time is greater than 260.

[0040] The scale of the fault shattering zone is divided according to the width of the fault shattering zone. The results of the fault shattering zone scale classification are as follows: the first level, there is no fault shattering zone; the second level, the width of the fault shattering zone is between 0 and 2 meters; the third level, the width of the fault shattering zone is between 2 and 5 meters; the fourth level, the width of the fault shattering zone is between 5 and 10 meters; the fifth level, the width of the fault shattering zone is greater than 10 meters.

[0041] The water conductivity grade of the fault fracture zone is divided according to the compactness of the fault fracture zone and the nature of the internal filling. The results of the fault fracture zone water conductivity grade classification are as follows: the first grade, the fault fracture zone is compacted and the internal filling is densely bonded; the second grade, the fault fracture zone is compacted and the internal filling is generally bonded; the third grade, the fault fracture zone is generally bonded and the internal filling is generally bonded; the fourth grade, the fault fracture zone is loosely bonded and the internal filling is generally bonded; the fifth grade, the fault fracture zone is loosely bonded and the internal filling is loosely bonded.

[0042] The groundwater development level is divided into the following levels: the first level, the groundwater is scarce; the second level, the groundwater is relatively developed; the third level, the groundwater is developed; the fourth level, the groundwater is relatively abundant; the fifth level, the groundwater is abundant.

[0043] The surface water recharge situation is divided into levels according to the surface water recharge situation. The results of the surface water recharge situation are as follows: Level 1, surface water is scarce; Level 2, surface water is relatively developed; Level 3, surface water is developed; Level 4, surface water is relatively abundant; Level 5, surface water is abundant.

[0044] The levels of water and mud inrush in the fault fracture zone are divided according to the amount of water and mud inrush, and the results of the classification of water and mud inrush in the fault fracture zone are as follows: the first level, the daily amount of water and mud inrush is less than 100 cubic meters; the second level, the daily amount of water and mud inrush is between 100 cubic meters and 1,000 cubic meters; the third level, the daily amount of water and mud inrush is between 1,000 cubic meters and 10,000 cubic meters; the fourth level, the daily amount of water and mud inrush is between 10,000 cubic meters and 100,000 cubic meters; the fifth level, the daily amount of water and mud inrush is greater than 100,000 cubic meters.

[0045] The present invention converts geological data into grades, and then obtains the predicted fault fracture zone scale grade after preprocessing by a comprehensive fault fracture zone prediction model based on polynomial naive Bayes. If a fault fracture zone exists, the predicted fault fracture zone water and mud inrush grade is obtained by a fault fracture zone water and mud inrush grade prediction model based on genetic algorithm and support vector machine, thereby obtaining the daily water and mud inrush amount corresponding to the fault fracture zone water and mud inrush grade.

Claims

1. A method for predicting water and mud inrush in fault fracture zones based on comprehensive advanced geological prediction, characterized in that: The following steps are involved: S1. Collecting and preprocessing the input data for fault fracture zone prediction; the input data for fault fracture zone prediction includes the possibility level of the existence of the fault fracture zone, the level of the near precursor, the level of the rock mass integrity, the level of the drilling situation, the level of the seismic wave reflection method and the level of the geological radar method; S2, input the pre-processed fault fracture zone prediction input data into the fault fracture zone comprehensive prediction model based on polynomial naive Bayes to obtain the predicted fault fracture zone scale level. If there is no fault fracture zone, there is no water inrush and mud inrush. If there is a fault fracture zone, enter S3; S3, collecting and preprocessing the water gushing and mud bursting prediction input data; the water gushing and mud bursting prediction input data includes the fault fracture zone water conductivity grade, rock mass integrity grade, groundwater development grade, surface water recharge grade and the predicted fault fracture zone scale grade; S4. Input the preprocessed water gushing and mud burst prediction input data into a fault fracture zone water gushing and mud burst grade prediction model based on genetic algorithm and support vector machine to obtain the predicted fault fracture zone water gushing and mud burst grade.

2. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 1 is characterized in that: The preprocessing includes feature encoding, and the feature encoding includes one-hot encoding.

3. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 1 is characterized in that: The fault rupture zone comprehensive prediction model based on polynomial naive Bayes is trained by collecting existing fault rupture zone sample data, wherein the fault rupture zone sample data includes fault rupture zone prediction input data and fault rupture zone scale level.

4. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 3 is characterized in that: Before training, the existing fault zone sample data are processed for missing values, sample imbalance and feature encoding.

5. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 1 is characterized in that: The fault fracture zone water and mud inrush grade prediction model based on genetic algorithm and support vector machine is trained by collecting existing water and mud inrush sample data, wherein the water and mud inrush sample data includes water and mud inrush prediction input data and the fault fracture zone water and mud inrush grade.

6. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 5 is characterized in that: Before training, the existing water and mud burst sample data were processed for missing values, sample imbalance and feature encoding.

7. The method for predicting water inrush and mud burst in fault fracture zones based on comprehensive advanced geological prediction according to claim 4 or 6, characterized in that: The missing value processing includes deletion method, manual filling method, special value filling method, interpolation filling method and prediction filling method; the sample imbalance processing includes sampling, category weight and generation of synthetic samples, and the generation of synthetic samples uses the SMOTE algorithm; the feature encoding uses one-hot encoding.

8. The method for predicting water and mud inrush in fault fracture zones based on comprehensive advanced geological prediction according to any one of claims 1 to 6, characterized in that: The possibility level of fault fracture zone existence is divided according to the signs of fault fracture zone existence; the level of adjacent precursors is divided according to the number and obviousness of precursors; the level of rock integrity is divided according to rock structure, number of rock volume joints and rock integrity index; the level of drilling conditions is divided according to the number of drilling anomalies; the level of seismic wave reflection method is divided according to the difference between the interpretation result and the longitudinal wave velocity of the predicted mileage section; the level of geological radar method is divided according to the interpretation result and the two-way travel time; the scale level of fault fracture zone is divided according to the width of the fault fracture zone; the water conductivity level of the fault fracture zone is divided according to the compactness of the fault fracture zone and the nature of the internal filling; the level of groundwater development is divided according to the level of groundwater development; the level of surface water recharge is divided according to the surface water recharge; the level of water and mud inrush in the fault fracture zone is divided according to the amount of water and mud inrush.

Images