Unfavorable geological type identification method and system

Abstract

The invention discloses an unfavorable geological type identification method and system, and the method comprises the steps: obtaining various geophysical detection data of a tunnel construction region, carrying out the preprocessing of the geophysical detection data, removing an abnormal value and noise interference, and obtaining the geophysical detection data of the tunnel construction region, a standardized geophysical parameter data set is obtained; in the tunnel face construction process, a seepage water sample is collected in real time for a key monitoring area, the pH value, conductivity and anion and cation concentration hydrochemical parameters of the seepage water sample are measured, and dynamic water ion characteristic data in the tunnel construction process are obtained; and for the determined potential water inrush point, combining with a pre-established water ion feature knowledge base of different types of unfavorable geologic bodies, classifying the water ion feature data of the potential water inrush point by adopting a support vector machine algorithm, and judging the unfavorable geologic type of the potential water inrush point.

Description

Technical Field

[0001] The invention belongs to the field of information technology, and in particular relates to a method and system for identifying adverse geological types. Background Art

[0002] In the process of tunnel construction, advanced geological prediction is a key technology, and its purpose is to timely discover the possible adverse geological conditions ahead of the tunnel, such as fault fracture zones, karst development zones, etc. However, the existing advanced geological prediction technology has some limitations. First, the geophysical parameter information obtained by different geophysical detection methods is different. How to comprehensively analyze and interpret these parameters to accurately predict the geological conditions ahead is an urgent problem to be solved. Secondly, water inrush problems are often encountered during tunnel construction. The water ion characteristics of the water inrush point are different from those of the surrounding rock mass, but there is currently no effective method to judge the adverse geological type, location and scale of the water inrush point based on these differences. Furthermore, the water ion characteristics corresponding to different types of adverse geological bodies are different, and the water ion characteristics of the same type of adverse geological body under different geological conditions may also be different. Therefore, how to establish the corresponding relationship between adverse geological types and water ion characteristics, and further establish an evaluation index system, is also an urgent problem to be solved. Finally, how to effectively manage and utilize the water ion characteristic data of various adverse geological types and build a complete water ion characteristic database of adverse geological types is also a problem worthy of in-depth study. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for identifying adverse geological types.

[0004] To achieve the above object, the present invention adopts the following technical solution:

[0005] A method for identifying adverse geological types, comprising:

[0006] Obtain geophysical survey data for the tunnel construction area;

[0007] Based on the geophysical exploration data of the tunnel construction area, the dynamic water ion characteristic data during the tunnel construction process is obtained;

[0008] The dynamic water ion characteristic data is analyzed by a cluster analysis algorithm, and the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass is calculated. If the Euclidean distance exceeds a preset threshold, it is determined that the difference between the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the location is determined to be a potential water inrush point;

[0009] For the determined potential water inrush points, combined with the pre-established knowledge base of water ion characteristics of different types of adverse geological bodies, the support vector machine algorithm is used to classify the water ion characteristic data of the potential water inrush points to determine the adverse geological type of the potential water inrush points.

[0010] Preferably, the geophysical exploration data include seismic wave velocity, resistivity and density parameters.

[0011] Preferably, obtaining dynamic water ion characteristic data during tunnel construction includes:

[0012] Preprocessing the geophysical exploration data to remove outliers and noise interference to obtain a standardized geophysical parameter data set;

[0013] According to a pre-established knowledge base of correspondence between geophysical parameters and geological types, a pattern recognition algorithm is used to classify the standardized geophysical parameter data set, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones;

[0014] During the tunnel face construction process, seepage water samples are collected in real time from key monitoring areas, and the pH value, electrical conductivity and anion and cation concentration water chemical parameters of the seepage water samples are measured to obtain dynamic water ion characteristic data during the tunnel construction process.

[0015] Preferably, the method further includes: comprehensively analyzing the predicted results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information, generating a comprehensive geological forecast result in front of the tunnel by weighted superposition, comparing the comprehensive geological forecast result with actual construction data, and updating the water ion characteristic database of adverse geological types.

[0016] The present invention also provides a system for identifying adverse geological types, comprising:

[0017] A first processing device for acquiring geophysical detection data of the tunnel construction area;

[0018] The second processing device is used to obtain dynamic water ion characteristic data during the tunnel construction process based on the geophysical detection data of the tunnel construction area;

[0019] The third processing device is used to analyze the dynamic water ion characteristic data through a cluster analysis algorithm, calculate the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass, and if the Euclidean distance exceeds a preset threshold, it is judged that the difference between the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the place is determined to be a potential water inrush point;

[0020] The fourth processing device is used to classify the water ion characteristic data of the potential water inrush point determined by the potential water inrush point in combination with the pre-established water ion characteristic knowledge base of different types of adverse geological bodies, and use the support vector machine algorithm to determine the adverse geological type of the potential water inrush point.

[0021] Preferably, the geophysical exploration data include seismic wave velocity, resistivity and density parameters.

[0022] Preferably, the second processing device comprises:

[0023] A first processing module is used to pre-process the geophysical detection data, remove outliers and noise interference, and obtain a standardized geophysical parameter data set;

[0024] The second processing module is used to classify the standardized geophysical parameter data set using a pattern recognition algorithm according to a pre-established knowledge base of correspondence between geophysical parameters and geological types, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones;

[0025] The third processing module is used to collect seepage water samples in real time in key monitoring areas during the construction of the tunnel face, measure the pH value, conductivity and anion and cation concentration water chemical parameters of the seepage water samples, and obtain dynamic water ion characteristic data during the tunnel construction process.

[0026] Preferably, it also includes: an updating device for comprehensively analyzing the predicted results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information, generating a comprehensive geological forecast result in front of the tunnel by weighted superposition, comparing the comprehensive geological forecast result with the actual construction data, and updating the water ion characteristic database of adverse geological types.

[0027] The present invention first obtains geophysical detection data of the tunnel construction area, including seismic wave velocity, resistivity and density parameters, and pre-processes and standardizes the data. Then, according to the corresponding relationship between geophysical parameters and geological types, a pattern recognition algorithm is used to classify the data, and a preliminary prediction of the unfavorable geological types that may exist in front of the tunnel is made. During the construction process, seepage water samples are collected in real time and their water chemical parameters are measured to obtain dynamic water ion characteristic data. The water ion characteristic data is analyzed by a cluster analysis algorithm to determine potential water inrush points. For potential water inrush points, a support vector machine algorithm is used to judge the unfavorable geological type in combination with a water ion characteristic knowledge base of unfavorable geological bodies. Finally, the prediction results, water inrush risks and other information are comprehensively analyzed to generate a comprehensive geological forecast result in front of the tunnel, and the database is updated by comparison with the actual construction data. The present invention can accurately predict the geological conditions and water inrush risks in tunnel construction by combining multiple data analysis and machine learning algorithms, thereby improving construction safety and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The figure is a flow chart of the method for identifying adverse geological types of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0030] Embodiment 1:

[0031] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying adverse geological types, including:

[0032] S101. Acquire a variety of geophysical exploration data of the tunnel construction area, wherein the geophysical exploration data include seismic wave velocity, resistivity and density parameters; preprocess the geophysical exploration data to remove outliers and noise interference to obtain a standardized geophysical parameter data set.

[0033] A variety of geophysical detection data such as seismic wave velocity, resistivity and density in the tunnel construction area are obtained to construct a geophysical parameter data set. The geophysical parameter data set is preprocessed, and the median filter algorithm is used to remove outliers. The wavelet transform method is used to remove noise interference to obtain a standardized geophysical parameter data set. According to the standardized geophysical parameter data set, a three-dimensional geological model of the tunnel construction area is constructed, which contains attribute parameters such as seismic wave velocity, resistivity and density. The support vector machine algorithm is used to classify the three-dimensional geological model, and the tunnel construction area is divided into different geological units according to the differences in attribute parameters. For each geological unit, the homogeneous area and inhomogeneous area inside the geological unit are identified through attribute parameter clustering analysis. For the identified inhomogeneous area, the Kriging interpolation algorithm is used to estimate the distribution of geophysical parameters inside the inhomogeneous area. According to the results of geological unit division and the estimated results of parameter distribution inside the inhomogeneous area, the geological conditions of the tunnel construction area are comprehensively evaluated, the geological problems that may be encountered in tunnel construction are predicted, and corresponding solutions are proposed.

[0034] Specifically, obtaining geophysical exploration data in the tunnel construction area is the basis for geological assessment. Taking a mountain tunnel as an example, wave velocity, resistivity and density data can be obtained by using methods such as seismic exploration, electrical exploration and gravity exploration. Seismic exploration measures the propagation time of waves in the stratum by exciting and receiving seismic waves, thereby calculating wave velocity. Electrical exploration measures the electric field distribution and calculates resistivity by emitting current underground. Gravity exploration inverts the underground density distribution by measuring the change of gravity field. Preprocessing the raw data can improve the accuracy of subsequent analysis. Median filtering can effectively remove outliers, such as replacing points in wave velocity data that obviously deviate from the surrounding values ​​with the median of neighboring data. Wavelet transform can separate signals and noise, such as performing wavelet decomposition on resistivity data, retaining low-frequency coefficients and setting high-frequency noise coefficients to zero, and then reconstructing the denoised data. Standardization processing can unify different physical quantities to the same scale, which is convenient for comprehensive analysis. The construction of a three-dimensional geological model is the key to data visualization and spatial analysis. The exploration area can be divided into several cubic units by a three-dimensional grid partitioning method, and each unit is assigned a corresponding wave velocity, resistivity and density attribute value. Through the interpolation algorithm, a continuous three-dimensional parameter distribution can be obtained from discrete exploration data. Support vector machine is an effective classification algorithm that can divide the model into different geological units according to multidimensional attribute parameters. By setting appropriate kernel functions and penalty parameters, the optimal classification hyperplane can be found in the high-dimensional feature space. For example, the area with low wave velocity, high resistivity and low density can be divided into a broken zone, and the area with high wave velocity, low resistivity and high density can be divided into a complete rock mass. Cluster analysis can further identify homogeneous and heterogeneous areas within the geological unit. K-means algorithm is a commonly used clustering method. It clusters similar data points into one class by iteratively optimizing the center position of the class. For the identified heterogeneous areas, Kriging interpolation can be used for fine estimation. Kriging method takes into account spatial correlation and can estimate the attribute values ​​of unknown points based on the attribute values ​​and spatial positions of known points. Comprehensive evaluation of geological conditions is crucial for tunnel construction. For example, if it is found that the tunnel line passes through a broken zone, it may encounter problems such as water gushing and collapse, and measures such as advance support and grouting reinforcement should be taken. If a karst development area is found, there may be risks of mud and water inrush, and borehole detection and curtain grouting should be carried out. For high-stress areas, rock bursts may occur, and measures such as pre-splitting blasting and systematic anchor support should be taken. This geological assessment method based on geophysical parameters has many advantages. First, the comprehensive analysis of multiple physical parameters can provide more comprehensive geological information, which is more reliable than a single parameter. Secondly, the three-dimensional model can intuitively display the geological structure, which helps engineers understand complex geological conditions. Furthermore, data-driven classification and clustering methods can reduce the errors caused by subjective judgment and improve the objectivity and accuracy of the assessment. Finally, refined parameter estimation provides a basis for the formulation of construction plans tailored to local conditions, which is conducive to improving construction efficiency and safety. However, this method also has some limitations.Geophysical exploration data may be affected by factors such as surface conditions and instrument accuracy, and there is a certain degree of uncertainty. Assumptions and simplifications in model construction and parameter estimation may also introduce errors. Therefore, in practical applications, direct geological survey methods such as drilling and pit exploration should be combined to verify and correct geophysical interpretation results to obtain more reliable geological assessment results. At the same time, other factors such as hydrogeological conditions and engineering geological conditions should also be considered to comprehensively assess tunnel construction risks.

[0035] S102. Based on a pre-established knowledge base of correspondence between geophysical parameters and geological types, a pattern recognition algorithm is used to classify the standardized geophysical parameter data set, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones.

[0036] According to the geophysical exploration data of the tunnel project, various geophysical parameters reflecting the underground geological conditions are obtained, such as seismic wave velocity, resistivity, density, etc. The obtained geophysical parameter data are preprocessed, including denoising and normalization, to obtain a standardized geophysical parameter data set. According to the pre-established knowledge base of the correspondence between geophysical parameters and geological types, pattern recognition algorithms such as support vector machines are used to classify and train the standardized geophysical parameter data set to establish a discriminant model for geophysical parameters and geological types. Using the trained discriminant model, the geophysical parameter data in front of the tunnel are classified and predicted to determine the possible unfavorable geological types in front of the tunnel, such as fault fracture zones and karst development zones. If the prediction results show that there is unfavorable geology in front of the tunnel, the clustering algorithm is further used to divide the unfavorable geological area according to its type and distribution characteristics to determine its spatial location and range. According to the divided unfavorable geological areas, combined with tunnel construction parameters and empirical knowledge, the decision tree algorithm is used to comprehensively evaluate the geological conditions in front of the tunnel to obtain the preliminary prediction results of the geological type in front of the tunnel. Compare and analyze the preliminary prediction results with the geological conditions revealed by actual construction, continuously optimize and improve the knowledge base and discrimination model of the correspondence between geophysical parameters and geological types, and improve the accuracy and reliability of subsequent tunnel geological predictions.

[0037] Specifically, geophysical exploration of tunnel engineering is an important means to obtain underground geological information. Taking seismic wave velocity as an example, by exciting seismic waves and recording their propagation time, the longitudinal wave velocity of the stratum can be calculated. Different lithologies and geological structures have different wave velocity characteristics, such as hard rocks with higher wave velocity, while broken zones or karst areas have lower wave velocity. Resistivity measurement uses the difference in rock conductivity to obtain the resistivity distribution of the stratum by injecting current into the ground and measuring the potential difference. The resistivity of dense rocks is usually higher, while the resistivity of aquifers or broken zones is lower. Density parameters can be obtained indirectly through gravity measurement, reflecting the mass distribution characteristics of the stratum. Preprocessing the original geophysical data is a key step to ensure the reliability of subsequent analysis. De-noising can use the wavelet transform method to decompose the signal into wavelet coefficients of different scales, and remove high-frequency noise through threshold processing. Normalization processing can unify different physical quantities to the same scale for comprehensive analysis. For example, the seismic wave velocity, resistivity and density data can be divided by their respective maximum values ​​to normalize them to the range of 0-1. Establishing a knowledge base of the correspondence between geophysical parameters and geological types is the basis for intelligent identification. Through a large amount of engineering practice accumulation, the typical characteristics of geophysical parameters under different geological conditions can be summarized. For example, fault fracture zones usually show low wave velocity, low resistivity and low density anomalies, while karst development areas may show low wave velocity and high resistivity. Machine learning algorithms such as support vector machines can learn the nonlinear mapping relationship between geophysical parameters and geological types through training samples based on these prior knowledge. When using the trained model to predict the geology ahead of the tunnel, the sliding window method can be used. A prediction window of a certain length is set along the axis of the tunnel, and the geophysical parameters in the window are comprehensively analyzed to predict the geological type of the area. By gradually moving the prediction window, continuous geological prediction results ahead of the tunnel can be obtained. If the prediction results show that there is poor geology, such as a fault fracture zone, its spatial distribution characteristics need to be further determined. Clustering algorithms such as K-means can be used to cluster areas with similar geophysical characteristics into one category, thereby dividing the spatial range of poor geology. Combined with the divided bad geological areas and tunnel construction parameters, the decision tree algorithm can be used to comprehensively evaluate the geological conditions ahead of the tunnel. Each node of the decision tree can represent different judgment conditions, such as whether the seismic wave velocity is lower than a certain threshold, whether the resistivity is abnormal, etc. Through step-by-step judgment, the prediction result of the geological type ahead of the tunnel is finally obtained. For example, if the wave velocity in a certain area is lower than 3000m / s, the resistivity is lower than 100Ω·m, and the density is less than 2.5g / cm 3, it may be determined to be a fault fracture zone. By comparing and analyzing the prediction results with the geological conditions revealed by actual construction, the prediction model can be continuously optimized. If it is found that the prediction accuracy under a certain type of geological conditions is low, the training samples of this type can be supplemented or the model parameters can be adjusted. Through this feedback mechanism, the accuracy and reliability of geological prediction can be gradually improved, providing more accurate geological information support for tunnel construction.

[0038] According to the demarcated unfavorable geological areas, combined with tunnel construction parameters and empirical knowledge, a decision tree algorithm is used to comprehensively evaluate the geological conditions ahead of the tunnel, and a preliminary prediction result of the geological type ahead of the tunnel is obtained.

[0039] Based on the existing tunnel construction parameters and empirical knowledge, a tunnel geological condition evaluation index system is established, including indicators such as lithology, rock layer thickness, fault fracture zone, etc., and the weight of each indicator is determined. The fuzzy comprehensive evaluation method is used to quantify and score each evaluation indicator to obtain a comprehensive score of the geological conditions in different areas ahead of the tunnel. According to the comprehensive scoring results, the geological conditions ahead of the tunnel are classified into several unfavorable geological areas, such as rock burst area, water gushing area, weak surrounding rock area, etc. Real-time data during tunnel construction, such as excavation speed, initial support deformation, surrounding rock deformation rate, etc., are collected as input features of the decision tree algorithm. According to the divided unfavorable geological areas, the tunnel construction process data is divided into different data subsets and corresponded to the corresponding geological type labels. The ID3 decision tree algorithm is used to generate a tunnel geological condition prediction model with information gain as the division criterion. The geological type ahead of the tunnel is predicted by judging the real-time construction data. The prediction results are compared with the actual geological conditions, and the decision tree model is continuously optimized to improve the accuracy of the geological prediction ahead of the tunnel, providing a reliable geological basis for tunnel construction.

[0040] Specifically, the establishment of a tunnel geological condition evaluation index system is the basis for predicting the geological conditions ahead of the tunnel. Taking a highway tunnel as an example, key indicators such as lithology, rock thickness, and fault fracture zone can be selected. Lithology can be divided into hard rock, soft rock, and extremely soft rock, and different weights are assigned according to their engineering characteristics. Rock thickness can be divided into thin layer, medium-thick layer, thick layer, etc., reflecting the integrity of the rock mass. The presence or absence of fault fracture zone directly affects the safety of tunnel construction and should be given a higher weight. The fuzzy comprehensive evaluation method can effectively deal with the uncertainty in geological evaluation. For example, for lithology indicators, five levels of "very good", "good", "average", "poor", and "very poor" can be set, and the values ​​are assigned to 9, 7, 5, 3, and 1 respectively. The comprehensive score of each area is obtained through expert scoring and fuzzy mathematical operations. Assuming that the comprehensive score of a certain section of the tunnel is 7.5, it can be judged that the geological conditions are good. Dividing the poor geological areas according to the scoring results is helpful for targeted construction. For example, the area in front of a tunnel with a score lower than 4 is divided into a weak surrounding rock area, the area with a score between 4-6 and high ground stress characteristics is divided into a rock burst area, and the area with a score between 5-7 and karst development characteristics is divided into a water gushing area. This division method can provide an important reference for subsequent construction. Real-time data during tunnel construction is an important input to the prediction model. Taking the excavation speed as an example, it is usually slow in hard rock areas and relatively fast in weak surrounding rock areas. The initial support deformation is often large in weak surrounding rock areas and small in intact rock masses. A sudden increase in the surrounding rock deformation rate may indicate that a fault fracture zone is about to be entered. The changing trend of these data can reflect the geological changes in front of the tunnel. The application of decision tree algorithm in tunnel geological prediction reflects the value of machine learning in the engineering field. Taking the ID3 algorithm as an example, it selects the optimal division feature by calculating information gain. Assuming that in the construction of a tunnel, it is found that the excavation speed contributes the most to the discrimination of geological types, then the root node of the decision tree will select the excavation speed as the basis for division. Subsequently, indicators such as initial support deformation and surrounding rock deformation rate may be considered to gradually refine the classification. By continuously comparing the prediction results with the actual geological conditions, the decision tree model can be continuously optimized. For example, if the model repeatedly misjudges high surrounding rock pressure as karst development, the weights or thresholds of related features may need to be adjusted. This dynamic optimization process can improve the adaptability and accuracy of the model, provide a more reliable geological prediction basis for tunnel construction, and thus effectively reduce construction risks and improve construction efficiency.

[0041] S103. During the construction of the tunnel face, collect seepage water samples in real time from key monitoring areas; measure the pH value, electrical conductivity and anion and cation concentration water chemical parameters of the seepage water samples to obtain dynamic water ion characteristic data during the tunnel construction process.

[0042] According to the key monitoring areas during the tunnel construction process, the location and frequency of real-time collection of seepage water samples are determined; the pH value, conductivity and other water chemical parameters of the collected seepage water samples are measured using equipment such as pH meters and conductivity meters; the concentrations of anions and cations in the seepage water samples are measured using equipment such as ion chromatographs; the measured water chemical parameters such as pH value, conductivity, anion and cation concentrations are used as dynamic water ion characteristic data during the tunnel construction process; a dynamic water ion characteristic database during the tunnel construction process is established to store the water chemical parameter data collected in real time; a time series analysis algorithm is used to analyze the changing trends and abnormal conditions of the dynamic water ion characteristic data; if abnormal changes are found in the water ion characteristic data, timely warnings are issued and the tunnel construction plan is adjusted to ensure construction safety and project quality.

[0043] Specifically, during the tunnel construction process, dynamic monitoring of the hydrochemical characteristics of seepage water is crucial for predicting and preventing geological disasters. First, according to the key monitoring areas of tunnel construction, such as fault fracture zones and karst development zones, the collection location and frequency of seepage water samples are determined. For example, a sampling point is set every 50 meters near the fault fracture zone, and water samples are collected every 4 hours. The pH value and conductivity of the water sample are measured on-site using a portable pH meter and conductivity meter. The pH value reflects the acidity and alkalinity of the water body, while the conductivity reflects the total amount of dissolved ions in the water. For example, the pH value of a certain sampling point is 7.2 and the conductivity is 850μS / cm, indicating that the seepage water at this location is weakly alkaline and has a high content of dissolved ions. The concentrations of the main anions and cations in the water sample are determined by ion chromatograph. Common cations include calcium, magnesium, sodium, potassium, etc., and anions include carbonate, sulfate, chloride, etc. For example, the calcium ion concentration in a water sample is measured to be 120mg / L and the sulfate ion concentration is 80mg / L, which may indicate that gypsum dissolution exists in the area. The measured water chemical parameters are used as dynamic water ion characteristic data, and a database is established for storage and management. The database should contain information such as sampling time, location, pH value, conductivity, and concentration of each ion. This makes it easy to track and analyze the spatiotemporal changes of water chemical characteristics. Time series analysis algorithms, such as moving average method and exponential smoothing method, are used to analyze the changing trends and abnormalities of dynamic water ion characteristic data. For example, the sulfate ion concentration change at a certain sampling point was analyzed by the moving average method, and it was found that its 7-day moving average gradually increased from 50 mg / L to 90 mg / L, which may indicate the development of caves in front of the tunnel. If abnormal changes in water ion characteristic data are found, such as a sudden decrease in pH value or a sharp increase in the concentration of a certain ion, timely warnings should be issued and the tunnel construction plan should be adjusted. For example, during the construction of a certain section of the tunnel, the conductivity of the seepage water increased from 600 μS / cm to 1500 μS / cm in a short period of time, and the concentrations of calcium ions and carbonate ions increased significantly, which may indicate that there is strong karstification in the area. At this time, the excavation speed should be slowed down immediately, the advanced geological forecast should be strengthened, and corresponding reinforcement measures should be taken, such as grouting reinforcement or pre-buried pipe sheds. By dynamically monitoring and analyzing the hydrochemical characteristics of tunnel seepage water, potential geological risks can be discovered in time, providing important decision-making basis for tunnel construction. This method can not only improve construction safety, but also optimize construction progress and reduce unnecessary economic losses. At the same time, the long-term accumulated hydrochemical data can also be used to establish a hydrogeological model of the area, providing valuable reference for future engineering construction.

[0044] S104, analyzing the dynamic water ion characteristic data by means of a cluster analysis algorithm, and calculating the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass; if the Euclidean distance exceeds a preset threshold, it is determined that the degree of difference in the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the location is determined to be a potential water inrush point.

[0045] The dynamic water ion characteristic data is obtained, and the data is preprocessed to remove outliers and noise data to obtain a normalized water ion characteristic data set. The K-means clustering algorithm is used to perform cluster analysis on the normalized water ion characteristic data set. According to the clustering results, the data is divided into different categories, and each category corresponds to a water ion characteristic pattern. The Euclidean distance between the water ion characteristics of the seepage water sample and the center point of each category of the surrounding rock mass is calculated to obtain the distance measurement value between the seepage water sample and each category. The Euclidean distance threshold is set. If the distance measurement value between the seepage water sample and a certain category exceeds the threshold, it is judged that the water ion characteristics of the seepage water sample are significantly different from the category. The significant difference between the seepage water sample and each category is counted. If the difference between the seepage water sample and most categories exceeds the threshold, it is judged that the water ion characteristics of the seepage water sample are significantly different from the surrounding rock mass as a whole. According to the spatial location information of the seepage water sample, the seepage water sample with significant difference in water ion characteristics is marked as a potential water inrush point, and a distribution map of potential water inrush points is generated. Evaluate the hazard level of potential water inrush points, calculate the hazard index of each potential water inrush point based on the degree of difference in water ion characteristics and the degree of spatial aggregation of seepage water samples, and generate a water inrush hazard level map to provide a decision-making basis for subsequent prevention and control measures.

[0046] Specifically, during tunnel construction, dynamic monitoring of water ion characteristics of seepage water is crucial for predicting and preventing water inrush disasters. First, the acquired dynamic water ion characteristic data needs to be preprocessed to remove outliers and noise. For example, a tunnel project collected 100 water samples in a week, and the pH value range was 6.5-8.5, but one of the samples had a pH value of 2.3, which was obviously deviated from the normal range and should be removed as an outlier. At the same time, the conductivity data may be affected by equipment jitter and produce noise, which can be smoothed by the sliding average method. The preprocessed data needs to be normalized to make indicators of different dimensions comparable. Common normalization methods include minimum-maximum normalization and Z-score normalization. Taking the calcium ion concentration of seepage water in a tunnel as an example, the original data range is 50-200 mg / L. After minimum-maximum normalization, the data is mapped to the 0-1 interval for subsequent analysis. The K-means clustering algorithm is a commonly used unsupervised learning method that can cluster similar water ion characteristic data into one category. Assuming that the water ion characteristics of a tunnel project include indicators such as pH value, conductivity, calcium ion and sulfate ion concentration, the K-means algorithm can be used to divide the data into different categories, each of which represents a specific hydrogeological condition. For example, the clustering results may show three main types: low-mineralized fresh water, high-sulfate water and high-calcium water, corresponding to different geological environments. Calculating the Euclidean distance between the seepage water sample and the center point of each category can quantitatively evaluate the similarity between the water sample and the known hydrogeological conditions. It is critical to set an appropriate distance threshold. Too large will reduce sensitivity, and too small may produce false alarms. For example, the experience of a tunnel project shows that when the Euclidean distance exceeds 2.5, it can be considered that the water sample characteristics are significantly different from the category. If the distance between a newly collected water sample and all known categories is greater than 2.5, it may indicate that the tunnel is about to enter a new geological unit or encounter abnormal hydrological conditions. Statistical differences between the seepage water sample and each category help to comprehensively evaluate the changes in hydrogeological conditions. If the water samples collected from a certain sampling point for multiple consecutive times show significant differences from most categories, this may indicate that there are special geological structures in the area, such as faults or karst development zones. The advantage of this method is that it can comprehensively consider multiple water chemical indicators and improve the reliability of judgment. According to the spatial distribution of seepage water samples with significant differences in water ion characteristics, a distribution map of potential water inrush points can be drawn. This visualization method intuitively displays high-risk areas and helps engineers quickly identify problems. For example, if multiple sampling points with significant differences in characteristics appear within 50 meters in front of the tunnel, it may indicate that there is a greater risk of water inrush in the area. To evaluate the hazard level of potential water inrush points, it is necessary not only to consider the degree of difference in water ion characteristics, but also to combine the spatial aggregation of seepage water samples. A comprehensive indicator can be designed, such as hazard index = degree of difference × spatial aggregation. The degree of difference can be expressed by Euclidean distance, and the spatial aggregation can be quantified by calculating the number of abnormal points per unit area.For example, if the average Euclidean distance of water ion characteristics in a certain area is 3.0 and there are 5 abnormal points per 100 square meters, then its hazard index is 15, and higher-level prevention and control measures may be required compared with other areas. The final generated water inrush hazard level map can provide an intuitive basis for engineering decision-making. The hazard level can usually be divided into three levels: low, medium, and high, with green, yellow, and red markings respectively. Engineering personnel can take corresponding measures according to different levels, such as advanced drilling, grouting reinforcement, or adjusting the excavation plan for high-risk areas. This data-driven risk assessment method can detect potential dangers in a timely manner and improve the safety and efficiency of tunnel construction.

[0047] S105. For the determined potential water inrush point, a support vector machine algorithm is used to classify the water ion characteristic data of the potential water inrush point in combination with a pre-established knowledge base of water ion characteristics of different types of adverse geological bodies, so as to determine the adverse geological type of the potential water inrush point.

[0048] Based on the pre-collected water ion characteristic data of different types of unfavorable geological bodies, a water ion characteristic knowledge base is established, which contains the typical water ion characteristic value ranges of various types of unfavorable geological bodies. For each potential water inrush point, a water sample at the point is obtained, and the water ion concentration data of the water sample is obtained by chemical analysis. The water ion concentration data of the potential water inrush point is compared with the typical characteristic values ​​in the knowledge base, and the water ion characteristic data is classified by the support vector machine algorithm to determine the unfavorable geological type to which the water sample belongs. If the water ion characteristic data cannot be clearly classified into a certain unfavorable geological type, the potential water inrush point is marked as an unknown type and included in the key monitoring object. According to the judgment result of the unfavorable geological type of the potential water inrush point, corresponding engineering measures are adopted to prevent and control water inrush. For example, for karst-type water inrush, grouting and plugging can be used for treatment. During the excavation process, the changes in parameters such as water pressure and water volume of the potential water inrush point are continuously monitored. Once an abnormality is found, the emergency plan is activated in time to avoid the occurrence of water inrush accidents. Regularly retest identified potential water inrush points, update water ion characteristic data, and combine geological forecast information ahead of the excavation face to dynamically evaluate and adjust the water inrush risk to guide engineering construction.

[0049] Specifically, a water ion characteristic knowledge base is established based on the pre-collected water ion characteristic data of different types of adverse geological bodies, and the knowledge base contains the typical water ion characteristic value ranges of various types of adverse geological bodies. For example, the water ion characteristics of karst areas may be manifested as high concentrations of calcium ions and magnesium ions, while the water ion characteristics of fault zones may be manifested as high sulfate ion concentrations. Through a large amount of field sampling and chemical analysis, the water ion concentration ranges of various types of adverse geological bodies are determined and entered into the knowledge base. For each potential water inrush point, a water sample at the point is obtained, and the water ion concentration data of the water sample is obtained by chemical analysis. Suppose that during the excavation of a coal mine, a seepage water sample is found. Through ion chromatography, the calcium ion concentration of the water sample is 120 mg / L, the magnesium ion concentration is 80 mg / L, and the sulfate ion concentration is 50 mg / L. The water ion concentration data of the potential water inrush point is compared with the typical characteristic values ​​in the knowledge base, and the water ion characteristic data is classified by the support vector machine algorithm. The support vector machine algorithm effectively distinguishes different categories of water ion characteristic data by finding the optimal hyperplane. For example, if the water ion characteristic range of karst-type water inrush in the knowledge base is 100-150 mg / L for calcium ions, 70-90 mg / L for magnesium ions, and 40-60 mg / L for sulfate ions, while the characteristic range of fault-type water inrush is 50-80 mg / L for calcium ions, 30-50 mg / L for magnesium ions, and 70-90 mg / L for sulfate ions, then the water sample is closer to the characteristics of karst-type water inrush. If the water ion characteristic data cannot be clearly classified into a certain adverse geological type, the potential water inrush point is marked as an unknown type and included in the key monitoring object. For example, if the calcium ion concentration of a water sample is 90 mg / L, the magnesium ion concentration is 60 mg / L, and the sulfate ion concentration is 65 mg / L, this data feature between the two types cannot be clearly classified and needs further monitoring and analysis. According to the adverse geological type judgment results of the potential water inrush point, the corresponding engineering measures are adopted to prevent and control water inrush. For example, for karst-type water inrush, grouting and plugging can be used for treatment. In specific operations, appropriate grouting materials and processes can be selected according to the karst development conditions, such as using cement slurry or chemical slurry, which is injected into karst fissures through boreholes to form a curtain to prevent water infiltration. During the excavation process, the water pressure, water volume and other parameter changes of potential water inrush points are continuously monitored. Once an abnormality is found, the emergency plan is activated in time to avoid the occurrence of water inrush accidents. For example, if the water pressure of a potential water inrush point is monitored to suddenly increase and the water volume increases, it may indicate an increased risk of water inrush. It is necessary to stop excavation immediately and take emergency measures such as drainage and reinforcement to ensure construction safety. The identified potential water inrush points should be retested regularly, the water ion characteristic data should be updated, and the water inrush risk should be dynamically evaluated and adjusted in combination with the geological forecast information ahead of the excavation face to guide the construction of the project.For example, water samples are retested at potential water inrush points every quarter. If changes are found in water ion concentration, the adverse geological type and water inrush risk need to be reassessed, and the prevention and control measures should be adjusted according to the new assessment results. Through the above steps, not only can the risk of water inrush be effectively identified and prevented, but also the prevention and control strategy can be dynamically adjusted according to real-time data to improve construction safety and efficiency. Establishing a knowledge base of water ion characteristics and using support vector machine algorithms for classification can improve the accuracy of water inrush prediction; continuous monitoring and dynamic evaluation can ensure the timeliness and effectiveness of prevention and control measures, forming a complete water inrush prevention and control system.

[0050] S106. Comprehensively analyze the predicted results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information, and generate a comprehensive geological prediction result in front of the tunnel by weighted superposition; compare the comprehensive geological prediction result with the actual construction data, and update the water ion characteristic database of the adverse geological type for reference in subsequent tunnel construction.

[0051] Step 1: Obtain the prediction results of the geological type ahead of the tunnel, the location coordinates of the potential water inrush point, and the water inrush risk level information. Step 2: According to the pre-set weight coefficient, perform weighted superposition calculation on the geological prediction results ahead of the tunnel, the location of the potential water inrush point, and the water inrush risk level information to obtain the comprehensive geological prediction results ahead of the tunnel. Step 3: Compare and analyze the comprehensive geological prediction results with the geological information obtained during the actual construction process to determine the accuracy and reliability of the prediction results. Step 4: If there is a large deviation between the prediction results and the actual construction data, update and optimize the water ion feature database of the bad geological type according to the actual construction data. Step 5: Train the updated water ion feature database of the bad geological type through the machine learning algorithm to improve the accuracy of the subsequent geological prediction of the tunnel construction. Step 6: Apply the optimized water ion feature database of the bad geological type to the subsequent tunnel construction to provide data support for the geological prediction and risk assessment of the construction site. Step 7: Based on the actual geological information obtained during the subsequent tunnel construction process, continuously update and improve the water ion characteristic database of adverse geological types to form a dynamically optimized geological prediction model to provide continuous geological risk warning and decision-making reference for tunnel construction.

[0052] Specifically, in tunnel construction, obtaining the prediction results of the geological type ahead, the location of potential water inrush points, and the risk level information is the key first step. These data can be obtained through geological drilling, geophysical exploration, and other means. For example, in a tunnel project, through advance geological drilling, it was found that there was a karst development area 50 meters ahead, the potential water inrush point was located at pile number K15+200, and the water inrush risk level was high. When weighted superposition calculation of this information is performed, the weight coefficient can be set according to engineering experience. For example, the weight of the geological type prediction result is 0.4, the weight of the potential water inrush point location is 0.3, and the weight of the water inrush risk level is 0.3. The comprehensive geological forecast results are obtained by calculation to provide a basis for construction decisions. Comparing and analyzing the forecast results with the actual construction data is an important step in verifying the accuracy of the forecast. If the forecast shows that the front is a karst development area, but the actual excavation finds that it is a complete rock mass, it means that the forecast is biased. At this time, it is necessary to update the database according to the actual situation, such as adjusting the water ion characteristic range of the karst development area. Training the updated database with a machine learning algorithm can improve the forecast accuracy. For example, the support vector machine algorithm is used to train the classification model with water ion concentration as the feature vector and bad geological type as the label. Through continuous iterative optimization, the prediction accuracy of the model can be increased from the initial 70% to more than 90%. The application of the optimized database in subsequent construction can provide more reliable data support for geological prediction. For example, when predicting the risk of water inrush in a certain section of the tunnel, the possible bad geological type can be quickly judged according to the ion concentration of the water sample, and corresponding prevention and control measures can be taken in time. Continuous updating and improvement of the database is the key to ensure the dynamic optimization of the prediction model. As the tunnel construction progresses, new geological information and water sample data are continuously collected and included in the database. This dynamic optimization mechanism can make the prediction model more in line with the actual engineering situation and improve the accuracy and timeliness of the warning. Through the above steps, a dynamically optimized geological prediction model can be constructed to provide continuous risk warning and decision support for tunnel construction. This method can not only improve construction safety, but also optimize the construction plan and reduce unnecessary economic losses. For example, in a tunnel project, the model successfully predicted a potential water inrush point, and grouting reinforcement measures were taken in time to avoid possible construction delays and economic losses.

[0053] The updated database of water ion characteristics of adverse geological types is trained through machine learning algorithms to improve the accuracy of geological forecasts for subsequent tunnel construction.

[0054] Obtain an updated database of water ion characteristics of adverse geological types as training data for machine learning algorithms; select a suitable machine learning algorithm, such as support vector machine, random forest or neural network, based on the characteristics of adverse geological types; divide the training data into a training set and a validation set, and preprocess the data, such as normalization and feature selection; use the training set data to train the selected machine learning model, and evaluate the model performance through the validation set; adjust the model hyperparameters based on the model performance evaluation results, and repeat the training and validation process until the optimal model is obtained; apply the trained machine learning model to tunnel construction geological prediction, and predict adverse geological types based on water ion characteristic data; combine the prediction results with expert experience to formulate a tunnel construction plan to improve construction safety and efficiency and reduce the risk of geological disasters.

[0055] Specifically, in tunnel construction, using machine learning algorithms to predict adverse geological types is an important means to improve construction safety and efficiency. First, it is necessary to obtain an updated database of water ion characteristics of adverse geological types as training data. These data contain various water ion concentrations, pH values, conductivity and other indicators, as well as corresponding geological type labels. For example, high calcium ion concentrations may be associated with karst geology, while high sulfate ion concentrations may indicate the presence of coal-bearing strata. Selecting a suitable machine learning algorithm is a key step. Support vector machines are suitable for processing high-dimensional feature spaces and can effectively distinguish the boundaries of different geological types. Random forests can handle nonlinear relationships and provide feature importance ranking. Neural networks perform well in processing complex patterns, especially when there is a large amount of training data. Data preprocessing is crucial to model performance. Normalization can unify features of different dimensions to the same scale to avoid certain features dominating model decisions. Feature selection helps reduce noise and improve model generalization ability. For example, through correlation analysis, it may be found that calcium ions and carbonate ions are highly correlated, and only one of them can be retained as a feature. Model training and validation is an iterative process. Taking random forest as an example, the model can be optimized by adjusting hyperparameters such as the number of trees and the maximum depth. Using cross-validation can better evaluate the generalization ability of the model. For example, on a data set containing 1,000 samples, 5-fold cross-validation can be used, using 800 samples for training and 200 samples for validation each time. The trained model can be applied to actual tunnel construction. Suppose that in the construction of a tunnel, water samples were collected and the calcium ion concentration was measured to be 120 mg / L, the sulfate ion concentration was 80 mg / L, and the pH value was 7.2. After inputting these data into the model, the model predicts that the probability of karst geology in the area is 85%. This prediction result provides an important reference for engineers. Combining the prediction results with expert experience to formulate a construction plan is the key to ensuring safety. For the predicted karst geology, engineers may take measures such as pre-grouting reinforcement and advanced geological forecasting. At the same time, the frequency of hydrological monitoring can be increased to pay close attention to groundwater changes. These measures can effectively reduce the risks of geological disasters such as water inrush and landslides and improve construction efficiency. By continuously collecting actual geological data during the construction process and constantly updating and improving the water ion characteristic database, a dynamically optimized geological prediction model can be formed. This model can not only adapt to the geological characteristics of different regions, but also continuously improve the prediction accuracy as data accumulates, providing more reliable geological risk warning and decision-making support for tunnel construction.

[0056] Embodiment 2:

[0057] The embodiment of the present invention further provides a system for identifying adverse geological types, including:

[0058] A first processing device for acquiring geophysical detection data of the tunnel construction area;

[0059] The second processing device is used to obtain dynamic water ion characteristic data during the tunnel construction process based on the geophysical detection data of the tunnel construction area;

[0060] The third processing device is used to analyze the dynamic water ion characteristic data through a cluster analysis algorithm, calculate the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass, and if the Euclidean distance exceeds a preset threshold, it is judged that the difference between the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the place is determined to be a potential water inrush point;

[0061] The fourth processing device is used to classify the water ion characteristic data of the potential water inrush point determined by the potential water inrush point in combination with the pre-established water ion characteristic knowledge base of different types of adverse geological bodies, and use the support vector machine algorithm to determine the adverse geological type of the potential water inrush point.

[0062] As an implementation manner of the embodiment of the present invention, the geophysical detection data includes seismic wave velocity, resistivity and density parameters.

[0063] As an implementation of the embodiment of the present invention, the second processing device includes:

[0064] A first processing module is used to pre-process the geophysical detection data, remove outliers and noise interference, and obtain a standardized geophysical parameter data set;

[0065] The second processing module is used to classify the standardized geophysical parameter data set using a pattern recognition algorithm according to a pre-established knowledge base of correspondence between geophysical parameters and geological types, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones;

[0066] The third processing module is used to collect seepage water samples in real time in key monitoring areas during the construction of the tunnel face, measure the pH value, conductivity and anion and cation concentration water chemical parameters of the seepage water samples, and obtain dynamic water ion characteristic data during the tunnel construction process.

[0067] As an implementation mode of an embodiment of the present invention, it also includes: an updating device, which is used to comprehensively analyze the prediction results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information, generate a comprehensive geological prediction result in front of the tunnel by weighted superposition, compare the comprehensive geological prediction result with the actual construction data, and update the water ion characteristic database of the adverse geological type.

[0068] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying adverse geological types, characterized in that: include: Obtain geophysical survey data for the tunnel construction area; Based on the geophysical exploration data of the tunnel construction area, the dynamic water ion characteristic data during the tunnel construction process is obtained; The dynamic water ion characteristic data is analyzed by a cluster analysis algorithm, and the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass is calculated. If the Euclidean distance exceeds a preset threshold, it is determined that the difference between the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the location is determined to be a potential water inrush point; For the determined potential water inrush points, combined with the pre-established knowledge base of water ion characteristics of different types of adverse geological bodies, the support vector machine algorithm is used to classify the water ion characteristic data of the potential water inrush points to determine the adverse geological type of the potential water inrush points.

2. The method for identifying adverse geological types according to claim 1, characterized in that: The geophysical exploration data include seismic wave velocity, resistivity and density parameters.

3. The method for identifying adverse geological types according to claim 2, characterized in that: The dynamic water ion characteristic data obtained during the tunnel construction process include: Preprocessing the geophysical exploration data to remove outliers and noise interference to obtain a standardized geophysical parameter data set; According to a pre-established knowledge base of correspondence between geophysical parameters and geological types, a pattern recognition algorithm is used to classify the standardized geophysical parameter data set, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones; During the tunnel face construction process, seepage water samples are collected in real time from key monitoring areas, and the pH value, electrical conductivity and anion and cation concentration water chemical parameters of the seepage water samples are measured to obtain dynamic water ion characteristic data during the tunnel construction process.

4. The method for identifying adverse geological types according to claim 3, characterized in that: Also includes: The predicted results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information are comprehensively analyzed, and a comprehensive geological prediction result in front of the tunnel is generated through weighted superposition. The comprehensive geological prediction result is compared with the actual construction data, and the water ion characteristic database of the adverse geological type is updated.

5. A system for identifying adverse geological types, characterized in that: include: A first processing device for acquiring geophysical detection data of the tunnel construction area; The second processing device is used to obtain dynamic water ion characteristic data during the tunnel construction process based on the geophysical detection data of the tunnel construction area; The third processing device is used to analyze the dynamic water ion characteristic data through a cluster analysis algorithm, calculate the Euclidean distance between the water ion characteristics of the seepage water sample and the surrounding rock mass, and if the Euclidean distance exceeds a preset threshold, it is judged that the difference between the water ion characteristics of the seepage water sample and the surrounding rock mass exceeds the preset threshold, and the place is determined to be a potential water inrush point; The fourth processing device is used to classify the water ion characteristic data of the potential water inrush point determined by the potential water inrush point in combination with the pre-established water ion characteristic knowledge base of different types of adverse geological bodies, and use the support vector machine algorithm to determine the adverse geological type of the potential water inrush point.

6. The adverse geological type identification system according to claim 5, characterized in that: The geophysical exploration data include seismic wave velocity, resistivity and density parameters.

7. The adverse geological type identification system according to claim 6, characterized in that: The second processing device comprises: A first processing module is used to pre-process the geophysical detection data, remove outliers and noise interference, and obtain a standardized geophysical parameter data set; The second processing module is used to classify the standardized geophysical parameter data set using a pattern recognition algorithm according to a pre-established knowledge base of correspondence between geophysical parameters and geological types, determine the unfavorable geological types that may exist in front of the tunnel, and obtain a preliminary prediction result of the geological type in front of the tunnel, wherein the unfavorable geological types include fault fracture zones and karst development zones; The third processing module is used to collect seepage water samples in real time in key monitoring areas during the construction of the tunnel face, measure the pH value, conductivity and anion and cation concentration water chemical parameters of the seepage water samples, and obtain dynamic water ion characteristic data during the tunnel construction process.

8. The adverse geological type identification system according to claim 7, characterized in that: Also includes: The updating device is used to comprehensively analyze the predicted results of the geological type in front of the tunnel, the location of potential water inrush points and the water inrush risk level information, generate a comprehensive geological prediction result in front of the tunnel by weighted superposition, compare the comprehensive geological prediction result with the actual construction data, and update the water ion characteristic database of the adverse geological type.

Images