A method for monitoring and early warning of geological hazards caused by mountain fissures

Abstract

This application provides a method for monitoring and early warning of geological hazards caused by mountain fissures, comprising: acquiring fissure morphology data of a mountain fissure area, the fissure morphology data including fissure length, width, depth, and orientation, and collecting the data in real time through a pre-set sensor network; acquiring geological characteristic parameters of the fissure area, the geological characteristic parameters including lithology, structure, and stress state, and collecting the data through a pre-deployed geological sensor network; transmitting the fissure morphology data and geological characteristic parameters collected in the field to a monitoring platform in real time; designing a visualization module based on dynamically updated fissure division results to generate a fissure spatial distribution map and attribute characteristic map; and performing correlation analysis between the visualized fissure division results and monitoring information such as rainfall and surface displacement in the monitoring platform to generate a disaster early warning report.

Description

Technical Field

[0001] This invention relates to the field of geological disasters, and in particular to a method for monitoring and early warning of geological disasters caused by mountain fissures. Background Technology

[0002] In monitoring geological hazards caused by mountain fissures, fissure segmentation is a crucial technology. Due to the complex and varied morphology of mountain fissures and the significant differences in geological characteristics across different regions, traditional fissure segmentation methods are insufficient for the needs of refined analysis. Furthermore, the dynamic and uncertain nature of fissure propagation makes it difficult to update the segmentation results in a timely manner, affecting the timeliness of disaster warnings. To achieve accurate crack classification and real-time updates, the following technical bottlenecks need to be overcome: First, how to design a scientific and reasonable classification index system based on crack morphological characteristics such as length, width, depth, and orientation, and establish corresponding classification models and algorithms to achieve automatic crack classification and regional division; second, how to obtain geological characteristic parameters of the crack area, such as lithology, structure, and stress state, and combine them with crack morphological characteristics for comprehensive analysis to improve the accuracy and reliability of classification; third, how to develop an efficient data transmission and processing technology to transmit massive amounts of monitoring data collected in the field to the monitoring platform in real time and dynamically update the classification results to ensure the synchronization and consistency between monitoring data and classification results; finally, how to design a visual display method for classification results to intuitively present the spatial distribution and attribute characteristics of cracks and perform correlation analysis with other monitoring information to provide intuitive and reliable basis for disaster early warning and decision-making. Solving the aforementioned technical problems requires the comprehensive application of knowledge from multiple disciplines, such as geology, rock and soil mechanics, computer science, and communication engineering, to carry out interdisciplinary collaborative research in order to overcome the technical bottlenecks in the accurate division and real-time transmission of cracks, and provide more intelligent and efficient technical support for monitoring geological disasters caused by mountain cracks. Summary of the Invention

[0003] This invention provides a method for monitoring and early warning of geological hazards caused by mountain fissures, mainly including:

[0004] Acquire crack morphology data in the mountain crack area, including crack length, width, depth and orientation, and collect the data in real time through a preset sensor network;

[0005] Based on the crack morphology data, a classification index system is established, which includes thresholds for crack length, width, depth, and orientation. A clustering algorithm is then used to classify the cracks to obtain preliminary crack classification results.

[0006] Geological characteristic parameters of the fracture area are acquired, including lithology, structure and stress state, and data are collected through a pre-deployed network of geological sensors.

[0007] By combining the geological feature parameters with the fracture morphology data, a classification model is constructed using the support vector machine algorithm to comprehensively analyze the fractures and optimize the fracture classification results.

[0008] The crack morphology data and geological feature parameters collected in the field are transmitted to the monitoring platform in real time.

[0009] On the monitoring platform, streaming data processing technology is used to process the received crack morphology data in real time, and the crack division results are dynamically updated in combination with the geological feature parameters.

[0010] Based on the dynamically updated crack segmentation results, a visualization module is designed to generate a crack spatial distribution map and attribute feature map;

[0011] The visualized crack division results are correlated with monitoring information such as rainfall and surface displacement in the monitoring platform to generate a disaster early warning report.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0013] This invention discloses a method for monitoring and early warning of geological hazards caused by mountain fissures. The method uses a sensor network to collect fissure morphology data and geological characteristic parameters in real time, and employs clustering and support vector machine algorithms to classify and comprehensively analyze the fissures. On the monitoring platform, streaming data processing technology is used to process the received data in real time and dynamically update the fissure classification results. This invention also includes a visualization module that generates spatial distribution maps and attribute feature maps of the fissures, and correlates the fissure classification results with monitoring information such as rainfall and surface displacement to generate a disaster early warning report. This method enables real-time monitoring, dynamic analysis, and visualization of mountain fissures, providing effective technical support for early warning of geological hazards such as landslides, improving the accuracy and timeliness of geological hazard warnings, and is of great significance for protecting the lives and property of residents in mountainous areas. Attached Figure Description

[0014] Figure 1 This is a flowchart of a method for monitoring and early warning of geological disasters caused by mountain fissures according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. The described embodiments are merely some embodiments of the present invention.

[0016] like Figure 1 As shown in the figure, the method for monitoring and early warning of geological disasters caused by mountain fissures in this embodiment may specifically include:

[0017] S101. Acquire crack morphology data of the mountain crack area. The crack morphology data includes crack length, width, depth and direction, and is collected in real time through a preset sensor network.

[0018] A preset sensor network layout scheme is obtained, which includes the placement positions of multiple sensor nodes. Based on this scheme, multiple sensor nodes are deployed in the target mountain fissure area to form a sensor network covering the fissure area. Image and depth data of the fissure area are collected by the sensor nodes. The image and depth data collected by the sensor nodes are transmitted to a data processing center. At the data processing center, image segmentation is performed on the image data to obtain an image of the fissure area. Based on the fissure area image, the length and width of the fissure are calculated to obtain fissure size data. Based on the depth data, the depth values ​​at various locations in the fissure area are determined to obtain fissure depth data. The Hough transform algorithm is used to process the fissure area image to calculate the fissure orientation angle, obtaining fissure orientation data. The fissure size data, depth data, and orientation data are fused to obtain fissure morphology data characterizing the fissure shape. Based on the fissure morphology data, the fissure morphology changes in the mountain fissure area are monitored in real time.

[0019] Specifically, obtaining a pre-defined sensor network layout plan is the first step in establishing a mountain fissure monitoring system. This plan is typically developed collaboratively by geologists and engineers, taking into account factors such as mountain topography and fissure distribution. For example, in a mountain fissure area of ​​approximately 5,000 square meters, 50 sensor nodes might be deployed in a grid-like distribution. This layout ensures comprehensive coverage of the entire area.

[0020] When deploying sensor nodes in the target area according to the layout plan, the communication distance between nodes and power supply need to be considered. Solar power and wireless communication technologies can be used to improve the reliability and durability of the system. Each node may be equipped with a high-resolution camera and a laser rangefinder for acquiring image and depth data.

[0021] Data transmission to the processing center is completed via a wireless network. To ensure the security and real-time performance of data transmission, encrypted transmission and edge computing technologies can be employed. At the data processing center, the images are first preprocessed, such as denoising and enhancement, and then deep learning algorithms are used for image segmentation to accurately extract crack regions.

[0022] Calculating crack size involves converting pixels to actual distances. For example, if the image resolution is 0.1 mm / pixel, a crack 100 pixels wide would actually be 10 mm wide. Processing depth data requires considering the accuracy of the laser rangefinder and environmental factors such as temperature and humidity. The Hough transform algorithm is used to calculate the crack's orientation, effectively identifying straight-line features. This method allows us to determine the crack's angle relative to north, such as 30°. This data is crucial for analyzing crack development trends.

[0023] By integrating data on size, depth, and orientation, a three-dimensional model of the crack can be constructed. This comprehensive data not only visually displays the current state of the crack but also allows for prediction of its future development trends through time-series analysis. For example, if the crack width increases by 5 mm, the depth increases by 10 mm, and the orientation angle changes by 2° within a week, this may indicate an increasing risk of landslide.

[0024] Real-time monitoring of crack morphology changes is the core objective of the entire system. By setting early warning thresholds, such as when the crack width increases at a rate exceeding 1 mm / day, the system can automatically issue an alarm. This continuous monitoring and timely early warning mechanism can provide valuable decision support for geological disaster prevention and control, effectively reducing the risk of disasters such as landslides.

[0025] S102. Based on the crack morphology data, establish a classification index system, which includes thresholds for crack length, width, depth, and orientation. Use a clustering algorithm to classify the cracks to obtain preliminary crack classification results.

[0026] Acquire crack morphology data, including crack length, width, depth, and orientation parameters. Set corresponding threshold ranges for the crack length, width, depth, and orientation classification indicators, which serve as the basis for the clustering algorithm. Use the K-means clustering algorithm, taking the crack morphology data as input, and perform cluster analysis based on the threshold settings of the classification indicator system to obtain preliminary crack classification results. Cracks of different categories in the preliminary classification results have similar morphological characteristics. Evaluate the accuracy and rationality of the preliminary classification results. If the preliminary classification results meet the requirements, determine the final crack classification scheme. If the preliminary classification results do not meet the requirements, adjust the threshold settings of the classification indicator system and return to the step of using the K-means clustering algorithm to re-perform the cluster analysis until a crack classification result that meets the requirements is obtained. Based on the final crack classification scheme, formulate targeted crack treatment and maintenance plans.

[0027] Specifically, acquiring crack morphology data is crucial for hill stability assessment. Crack length, width, depth, and orientation parameters collected through a sensor network provide the foundation for subsequent analysis. Setting threshold ranges for these parameters allows for better differentiation of different crack types. For example, length can be categorized as short (<1m), medium (1-5m), and long (>5m); width as narrow (<1cm), medium (1-5cm), and wide (>5cm); depth as shallow (<10cm), medium (10-50cm), and deep (>50cm); and orientation can be classified based on the angle with the principal stress direction.

[0028] The application of the K-means clustering algorithm in crack classification embodies a data-driven analysis approach. This algorithm iteratively groups cracks with similar characteristics into a single class. In practice, crack morphology data can be standardized and used as input vectors. Appropriate initial cluster centers are selected, and classification is performed by calculating Euclidean distance. For example, assuming 100 crack data points, 3-5 main categories might be obtained, each representing a group of cracks with similar morphological features. Evaluating the initial classification results is a crucial step in ensuring classification accuracy. The silhouette coefficient can be used to measure the clustering effect; a value closer to 1 indicates a better clustering effect.

[0029] In addition, clustering results can be visually displayed using methods such as scatter plots or heatmaps. If some cracks are found to be misclassified, or the boundaries between categories are unclear, the threshold settings need to be adjusted and the clustering re-performed. Threshold adjustment is an iterative optimization process. For example, if cracks with lengths between 4.5 and 5.5 meters are frequently found to be divided back and forth between two categories, the medium length range can be adjusted to 1-6 meters. Similarly, the thresholds for other parameters can be fine-tuned according to the actual situation. This process may need to be repeated multiple times until stable and reasonable classification results are obtained.

[0030] The final crack classification scheme provides a basis for developing targeted treatment measures. For example, long and deep cracks may require grouting reinforcement; wide and shallow cracks can be addressed with surface sealing; and cracks oriented in the same direction as the principal stress may require comprehensive reinforcement. This classification method allows for more efficient resource allocation, improving the efficiency and effectiveness of mountain reinforcement.

[0031] This data-driven crack classification method not only objectively reflects the actual situation of cracks in mountains but also continuously optimizes the classification criteria as data accumulates. It provides a scientific basis for mountain stability assessment and early warning, helping to improve the accuracy and effectiveness of geological disaster prevention and control. Furthermore, this method has good scalability and can be appropriately adjusted according to the geological characteristics of different regions, thus enabling its application in a wider range.

[0032] S103. Obtain geological characteristic parameters of the fracture area, including lithology, structure and stress state, by acquiring data through a pre-deployed network of geological sensors.

[0033] The process involves acquiring raw geological data of the fracture area from a pre-deployed network of geological sensors; preprocessing and extracting features from the raw geological data to obtain characteristic parameters representing the geological features of the fracture area, including lithology, structure, and stress state; analyzing the characteristic parameters using a clustering algorithm to divide the fracture area into at least two geological units based on their similarity; establishing a mapping relationship between the characteristic parameters and the degree of fracture development for each of the at least two geological units using a support vector machine algorithm to obtain the fracture development probability of each geological unit; classifying the geological unit as a high-risk fracture area if its fracture development probability exceeds a preset threshold, otherwise classifying it as a low-risk fracture area; generating decision rules for fracture prevention measures using a decision tree algorithm based on the characteristic parameters and the fracture development probability; establishing a prediction model between the characteristic parameters and the fracture propagation trend using a neural network algorithm to predict the fracture development in the future; and visualizing the fracture area division results, the fracture development probability, the decision rules, and the prediction results.

[0034] Specifically, obtaining raw geological data of the fractured area is fundamental to fracture analysis. For example, in a tunnel project in a mountainous area, a network of geological sensors, including strain gauges, inclinometers, and fracture gauges, was deployed. These sensors monitor rock mass deformation, changes in in-situ stress, and the opening and closing of fractures in real time, providing a rich data source for subsequent analysis.

[0035] Preprocessing and feature extraction of raw data are crucial steps in improving analytical efficiency. Raw data can be cleaned using methods such as filtering and denoising, and outlier removal. Then, characteristic parameters such as lithology (e.g., rock type, weathering degree), structure (e.g., fault and fold distribution), and stress state (e.g., principal stress direction, stress concentration areas) can be extracted. Clustering algorithms are used to analyze these characteristic parameters, effectively identifying regions with similar geological features. Using the K-means clustering algorithm, three geological units were identified: a hard rock mass zone, a fractured zone, and a weak interlayer zone. This division lays the foundation for subsequent fracture risk assessment.

[0036] Support vector machine (SVM) algorithms can establish a mapping relationship between feature parameters and the degree of fracture development. For each geological unit, inputting feature parameters such as lithology, structure, and stress state, the trained SVM model can output the fracture development probability of that unit. For example, due to its lower rock mass strength and higher stress concentration, the fracture development probability of a fracture zone may reach 0.8, exceeding the preset threshold of 0.7, and therefore it is identified as a high-risk fracture zone.

[0037] Decision tree algorithms can generate decision rules for crack prevention measures based on feature parameters and crack development probabilities. In this example, the decision tree might generate rules such as: if the rock mass strength is below 30 MPa and the stress concentration factor is greater than 2, then systematic anchor bolt support should be adopted; if the crack development probability is greater than 0.7 and groundwater is abundant, then curtain grouting should be implemented, etc. These rules provide engineers with clear decision-making basis. Neural network algorithms perform well in predicting crack propagation trends.

[0038] By establishing the relationship between characteristic parameters and crack propagation rates, crack development over a future period can be predicted. For example, in weak interlayer areas, a neural network model predicts that the length of the main crack may increase by 20% and the width by 15% within the next three months. This prediction provides an important reference for timely preventative measures. Finally, visualizing the analysis results can intuitively present the risk distribution and development trend of the cracked area. For example, GIS technology can be used to overlay information such as the geological unit division along the tunnel, the probability distribution of crack development, and recommended preventative measures onto a three-dimensional geological model, and display the predicted crack propagation results through dynamic charts. This visualization method helps engineers quickly understand complex geological conditions and potential risks, thereby developing more scientific and reasonable construction plans and monitoring schedules.

[0039] S104. Combine the geological feature parameters with the fracture morphology data, use the support vector machine algorithm to construct a classification model, perform comprehensive analysis on the fractures, and optimize the fracture classification results.

[0040] Geological feature parameters and fracture morphology data are acquired and fused to obtain a comprehensive feature dataset. Based on this dataset, data preprocessing techniques are used to clean and normalize it, resulting in standardized feature vectors. For these standardized feature vectors, a feature selection algorithm is employed to extract a subset of features with a correlation to fracture classification exceeding a preset threshold. This subset is then input into a support vector machine (SVM) algorithm to train a fracture classification model, which is subsequently optimized using cross-validation. Next, fracture data to be analyzed is acquired, and its geological feature parameters and morphology data are extracted to construct a feature vector. This feature vector is then input into the fracture classification model to obtain the classification results. Based on these classification results, combined with geological knowledge and experience, the preliminary fracture classification results are optimized and adjusted to obtain the final comprehensive fracture analysis results.

[0041] Specifically, acquiring geological characteristic parameters and fracture morphology data is fundamental to fracture analysis. Geological characteristic parameters, including lithology, structure, and stress state, can be collected through geological sensor networks. Fracture morphology data, including fracture length, width, and orientation, can be obtained through image recognition technology. Fusing these two types of data yields a more comprehensive feature dataset.

[0042] Data preprocessing is a crucial step in improving the quality of analysis. Cleaning the comprehensive feature dataset can remove outliers; for example, data with crack widths exceeding 10 cm are considered outliers and removed. Normalization unifies data with different dimensions to the same scale; for instance, lithological hardness indices and crack lengths are scaled to the range of 0-1, facilitating subsequent analysis.

[0043] Feature selection algorithms help extract the most relevant subset of features. For example, by calculating the correlation coefficient between each feature and crack classification, features with a correlation coefficient greater than 0.7 are selected. This may reveal that lithology and stress state have a significant impact on crack classification, while some minor factors can be ignored.

[0044] Support Vector Machines (SVMs) are powerful classification algorithms. By constructing an optimal hyperplane from the training data, different types of cracks can be separated. Cross-validation methods, such as K-fold cross-validation, can be used to optimize model parameters and improve classification accuracy.

[0045] Feature extraction and classification of the crack data to be analyzed are crucial for model application. For example, for a newly discovered crack, features such as the lithology (e.g., sandstone) and stress state (e.g., compressive stress of 20 MPa) of its location are extracted to construct a feature vector. Inputting this vector into a trained model yields the crack classification result, such as classifying it as a high-risk crack.

[0046] Finally, it is crucial to optimize and adjust the classification results by incorporating geological knowledge and experience. For example, if the model classifies a crack as low-risk, but geological experts, based on historical data and structural characteristics of the area, believe that its actual risk is higher, the result should be adjusted accordingly.

[0047] This human-machine collaborative approach fully leverages the efficiency of algorithms and the experience of experts to obtain more reliable comprehensive crack analysis results. This series of steps forms a complete crack analysis workflow, from data acquisition, preprocessing, and feature extraction to model training and application, and finally optimization using expert knowledge. This method not only improves the accuracy and efficiency of the analysis but also provides reliable decision support for crack prevention and control.

[0048] S105. Transmit the crack morphology data and geological characteristic parameters collected in the field to the monitoring platform in real time.

[0049] Raw monitoring data is obtained by acquiring crack morphology data and geological feature parameters. Data preprocessing techniques are used to remove noise and outliers from this raw data, resulting in cleaned, valid data. The cleaned data is packaged and encapsulated according to a predefined data transmission protocol and format specification to form standardized data packets. These standardized data packets are then transmitted in real-time to a remote monitoring platform via a constructed wireless transmission channel using encrypted transmission. Upon receiving the data packets, the monitoring platform uses data parsing techniques to extract crack morphology features and geological parameters, obtaining structured monitoring data. This structured monitoring data is input into a pre-built machine learning model, and a support vector machine algorithm is used for feature analysis to determine the degree of danger of the crack morphology. Based on the results of the crack morphology danger analysis and a predefined warning threshold, an automatic early warning of crack disasters is achieved using a decision tree algorithm, and the warning information is pushed to relevant departments.

[0050] Specifically, acquiring crack morphology data and geological characteristic parameters is a crucial step in monitoring rock cracks. This data may include morphological features such as crack length, width, and depth, as well as geological parameters such as rock type, geological structure, and stress state. For example, in a mountain quarry, a laser scanner can be used to acquire the three-dimensional morphology of cracks, while a strain gauge can be used to measure the stress state of the rock mass.

[0051] Raw monitoring data often contains noise and outliers, requiring preprocessing. Median filtering can be used to remove impulse noise, and moving averages can be used to smooth the data curves. Outliers can be removed by setting thresholds. For example, if a crack width data point deviates significantly in a monitoring session, it may be due to equipment malfunction and should be removed. After data cleaning, it needs to be packaged according to a standard format. JSON format can be used to organize crack morphology and geological parameters into structured data. For example,

[0052] {“crack_length”:2.5,“crack_width”:0.03,“rock_type”:“granite”,“stress_state”:“compressive”}. This format facilitates data transmission and subsequent processing.

[0053] Data transmission employs encryption to ensure security. The AES encryption algorithm can be used, with data packets encrypted using a pre-shared key. The transmission channel can be a 4G / 5G network or satellite communication, enabling remote real-time monitoring. After receiving the data, the monitoring platform performs parsing and structured processing. A JSON parser can be used to extract various parameters and store them in a relational database for subsequent analysis and querying. Machine learning models are used to analyze the severity of cracks. Support vector machines (SVMs) can effectively handle high-dimensional features, establishing a mapping between crack features and severity levels through training on historical data. For example, crack length, width, and depth can be used as input features to output a severity score.

[0054] Finally, disaster early warning is generated based on the analysis results. Multiple warning thresholds can be set; for example, a hazard score exceeding 0.7 triggers a yellow warning, and exceeding 0.9 triggers a red warning. The decision tree algorithm can comprehensively judge based on multiple factors, such as considering crack hazard, rainfall, and seismic activity, to generate more accurate early warning information. Early warning information can be pushed to relevant departments via SMS, email, etc., ensuring timely implementation of protective measures. This system, through the entire process of data collection, processing, transmission, and analysis, achieves intelligent monitoring and early warning of rock cracks, effectively improving the ability to prevent and control geological disasters.

[0055] S106. On the monitoring platform, the received crack morphology data is processed in real time using streaming data processing technology, and the crack division results are dynamically updated in combination with the geological feature parameters.

[0056] The system acquires fracture morphology data received from the monitoring platform. For each fracture morphology data point, morphological feature parameters are extracted, including the fracture length, width, depth, and orientation, to obtain a fracture morphology feature vector. Geological feature parameters corresponding to the fracture morphology data are also acquired, including lithology, stratigraphic age, and fault distribution. These geological feature parameters are then fused with the fracture morphology feature vector to construct a fracture morphology-geological feature fusion vector. Based on a pre-established fracture classification model, a support vector machine algorithm is used to classify the fracture morphology-geological feature fusion vector, classifying the fractures into tensile fractures and shear fractures, thus obtaining an initial fracture classification result. During the streaming data processing, the system continuously receives the latest fracture morphology data transmitted from the monitoring platform. Data is collected, morphological feature parameters are extracted, and fused with corresponding geological feature parameters to obtain a new fracture morphology-geological feature fusion vector. This new vector is then input into the fracture classification model, which is updated in real-time using an incremental learning algorithm to obtain dynamically adjusted fracture classification results. Statistical analysis is performed on the dynamically adjusted fracture classification results to calculate the number and length of each type of fracture, generating a fracture distribution statistical report, which is then sent to the monitoring platform for display. If a sudden increase in the number or length of fractures in a certain area is detected, an early warning mechanism is triggered. A time series analysis algorithm is used to predict the fracture development trend in that area, assess potential geological hazard risks, and push the early warning information to relevant departments.

[0057] Specifically, acquiring crack morphology data is a crucial step in geological disaster monitoring. By extracting morphological feature parameters such as crack length, width, depth, and orientation, a crack morphology feature vector can be constructed. For example, a crack was detected in a mountainous area with a length of 10 meters, a width of 0.5 meters, a depth of 2 meters, and an orientation of 45 degrees northeast. These data constitute the morphological feature vector of this crack.

[0058] The fusion of geological characteristic parameters can provide more comprehensive information for fracture analysis. Factors such as lithology, stratigraphic age, and fault distribution directly affect the formation and development of fractures. For example, in a Mesozoic strata composed of alternating sandstone and shale, if there is a north-south trending fault, this information, combined with the fracture morphology characteristic vector, forms a richer fracture morphology-geological characteristic fusion vector.

[0059] The application of support vector machine (SVM) algorithms in crack classification demonstrates the importance of machine learning in geological disaster early warning. By training the model to identify the characteristics of tensile and shear cracks, newly observed cracks can be quickly classified. For example, if a crack has a wide opening and is perpendicular to the principal stress direction, it is likely to be classified as a tensile crack; while if the crack exhibits shear displacement, it may be identified as a shear crack.

[0060] The introduction of incremental learning algorithms enables the crack classification model to be continuously optimized. With the continuous input of new data, the model can adjust its classification boundaries and improve accuracy. For example, if the initial model misclassifies cracks with certain features as tensile, incremental learning can gradually correct this bias and improve the recognition rate of shear cracks.

[0061] The generation of crack distribution statistics reports provides decision-makers with an intuitive risk assessment tool. By calculating the number and total length of cracks of various types, geological stability can be quickly determined. For example, if the total length of tension cracks suddenly increases from 100 meters to 500 meters within a 1-square-kilometer area, this may indicate that the geological structure of the area is undergoing significant changes.

[0062] Time series analysis algorithms play a crucial role in predicting crack development trends. By analyzing the changes in crack parameters over time, the potential development of cracks in the future can be estimated. For example, if a crack's width is observed to be widening at a rate of 0.5 cm per day over the past week, time series analysis can predict the crack's potential width for the following week, thereby assessing the potential risk of landslides or collapses. This comprehensive crack monitoring and analysis system can significantly improve the accuracy and timeliness of geological disaster early warnings. Through real-time data acquisition, intelligent algorithm analysis, and dynamic model updates, the system can identify danger signals before a disaster occurs, providing a valuable time window for relevant departments to take preventative measures, thus effectively reducing casualties and property losses caused by geological disasters.

[0063] S107. Based on the dynamically updated crack division results, design a visualization module to generate a crack spatial distribution map and attribute feature map.

[0064] The process involves: obtaining crack segmentation results, including spatial coordinate information and attribute information of the cracks; storing the crack segmentation results in a database; retrieving the spatial coordinate information and attribute information of the cracks from the database; generating a three-dimensional spatial distribution model of the cracks using three-dimensional modeling technology based on the spatial coordinate information; generating an attribute feature map of the cracks using chart visualization technology based on the attribute information of the cracks, the attribute feature map including statistical distribution maps of crack length, width, and depth; overlaying the three-dimensional spatial distribution model of the cracks with the attribute feature map of the cracks to generate a comprehensive visualization display interface for the cracks; grouping the cracks using a clustering algorithm to obtain crack classification results; associating the crack classification results with the comprehensive visualization display interface of the cracks, identifying different categories of cracks with different colors or patterns to generate a crack classification visualization display interface.

[0065] Specifically, obtaining fracture segmentation results is a crucial step in geological analysis, encompassing spatial coordinates and attribute information. Spatial coordinates may include longitude, latitude, and depth, while attribute information may cover fracture type, length, and orientation. This data is stored in a database, such as using...

[0066] The PostgreSQLwithPostGIS extension enables efficient management of geospatial data.

[0067] 3D modeling technology plays a crucial role in visualizing crack distribution. For example, using graphics libraries such as OpenGL or WebGL, 3D models can be constructed based on the spatial coordinates of cracks. These models visually represent the underground distribution of cracks, helping geological engineers assess potential risk areas. The generation of attribute feature maps relies on chart visualization techniques. For instance, the D3.js library can be used to create interactive statistical distribution maps. A concrete example is plotting a histogram of crack lengths, with the horizontal axis representing length ranges (e.g., 0-5m, 5-10m, etc.) and the vertical axis representing frequency, clearly demonstrating the distribution characteristics of crack lengths.

[0068] The generation of a comprehensive visualization interface involves overlaying a 3D spatial distribution model with attribute feature maps. This can be achieved through a combination of WebGL and SVG technologies. For example, adding clickable hotspots to a 3D model and then clicking them to display the corresponding crack's attribute feature map provides an interactive display that helps analysts gain a more comprehensive understanding of the crack's characteristics.

[0069] Clustering algorithms play a crucial role in fracture classification. Algorithms such as K-means or DBSCAN can group fractures based on their spatial distribution and attribute characteristics. For example, fractures might be classified into categories such as "shallow short fractures" or "deep long fractures." This classification helps identify potential geological structural units or risk areas.

[0070] The generation of a classification visualization interface combines clustering results with a comprehensive visualization. Different colors can be used to identify different categories of cracks, such as red for high-risk cracks and green for low-risk cracks. This visualization method can intuitively show the spatial distribution and risk level of cracks, helping to quickly identify areas requiring key attention. Through this series of steps, complex crack data can be transformed into an intuitive and easy-to-understand visualization. This not only helps geological engineers better understand underground structures but also provides decision-makers with intuitive risk assessment data, enabling the development of more targeted disaster prevention and mitigation measures. For example, if a large number of deep, long cracks are found in an area, it may be necessary to strengthen monitoring or implement reinforcement measures in that area.

[0071] S108. The visualized crack division results are correlated with the monitoring information such as rainfall and surface displacement in the monitoring platform to generate a disaster early warning report.

[0072] The system acquires crack segmentation results and rainfall and surface displacement monitoring information from a monitoring platform; it preprocesses the crack segmentation results and monitoring information to obtain preprocessed crack segmentation results and monitoring information, including data cleaning and data normalization; based on the preprocessed crack segmentation results and monitoring information, it constructs a correlation analysis model using a time series analysis algorithm to analyze the correlation between the crack segmentation results and the rainfall and surface displacement; through the correlation analysis model, it obtains the correlation rules and correlation strength between the crack segmentation results and the rainfall and surface displacement; based on the correlation rules and correlation strength, it determines whether the current crack segmentation results and monitoring information meet preset disaster warning conditions; if the disaster warning conditions are met, it determines the disaster warning level according to preset warning level standards and generates a disaster warning report; it displays the disaster warning report in a visual manner and pushes the warning information to relevant departments and personnel.

[0073] Specifically, obtaining crack segmentation results and monitoring information is a crucial starting point for analyzing geological hazard risks. For example, a monitoring station in a mountainous area recorded that the crack width recently increased from 2 mm to 5 mm, while rainfall reached 150 mm / day and surface displacement reached 20 cm. This raw data needs to be preprocessed before it can be used for subsequent analysis.

[0074] Data preprocessing includes outlier removal and normalization. During the removal process, it may be discovered that rainfall on a certain day is abnormally high, reaching 1000mm, which is clearly erroneous data caused by equipment malfunction and needs to be removed. Normalization unifies data with different dimensions into the 0-1 range, facilitating comparative analysis. For example, a crack width of 5mm is converted to 0.5, rainfall of 150mm is converted to 0.6, and surface displacement of 20cm is converted to 0.4.

[0075] Constructing correlation analysis models is key to revealing the interactions among various factors. Time series analysis algorithms can capture patterns in data changes over time. For example, analysis revealed a positive correlation between crack width and the cumulative rainfall over the preceding three days (correlation coefficient 0.8), and also a positive correlation with surface displacement (correlation coefficient 0.7). This suggests that rainfall and surface displacement may be significant factors contributing to crack expansion. Obtaining correlation rules and correlation strengths helps quantify the influence between factors. For instance, a rule might be derived that when the cumulative rainfall over three days exceeds 200 mm and the surface displacement exceeds 15 cm, the probability of increased crack width reaches 90%. This quantitative relationship provides a scientific basis for early warning.

[0076] Determining early warning conditions requires comprehensive consideration of multiple factors. Let's assume the preset disaster early warning conditions are: cumulative rainfall exceeding 250mm over 3 days, surface displacement exceeding 25cm, and crack width increasing by more than 50%. When monitoring data meets these conditions, the system will trigger an early warning. Early warning levels are typically determined in multiple tiers. For example, four levels of warnings can be set: blue (minor), yellow (moderate), orange (severe), and red (extremely severe). When crack width increases by more than 100%, a red warning may be triggered, indicating an extremely high disaster risk requiring immediate emergency measures.

[0077] The generation and visualization of early warning reports are crucial for conveying information to decision-makers. Reports may include current monitoring data, warning levels, risk assessments, and recommended measures. Visualization can take the form of dashboards, intuitively displaying the real-time status and warning levels of various indicators. Simultaneously, early warning information is pushed to relevant departments and personnel via SMS, email, and other means to ensure timely communication. The establishment of this system helps improve the accuracy and timeliness of geological disaster early warnings. Through data-driven methods, potential risks can be identified earlier, providing a scientific basis and decision support for disaster prevention and mitigation. However, the system's effectiveness requires continuous data accumulation and model optimization to adapt to the geological conditions and climatic characteristics of different regions.

[0078] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A method for monitoring and early warning of geological hazards caused by mountain fissures, characterized in that, The method includes: Acquire crack morphology data in the mountain crack area, including crack length, width, depth and orientation, and collect the data in real time through a preset sensor network; Based on the crack morphology data, a classification index system is established, which includes thresholds for crack length, width, depth, and orientation. A clustering algorithm is then used to classify the cracks to obtain preliminary crack classification results. Geological characteristic parameters of the fracture area are acquired, including lithology, structure and stress state, and data are collected through a pre-deployed network of geological sensors. By combining the geological feature parameters with the fracture morphology data, a classification model is constructed using the support vector machine algorithm to comprehensively analyze the fractures and optimize the fracture classification results. The crack morphology data and geological feature parameters collected in the field are transmitted to the monitoring platform in real time. On the monitoring platform, streaming data processing technology is used to process the received crack morphology data in real time, and the crack division results are dynamically updated in combination with the geological feature parameters. Based on the dynamically updated crack segmentation results, a visualization module is designed to generate a crack spatial distribution map and attribute feature map; The visualized crack division results are correlated with rainfall and surface displacement information in the monitoring platform to generate a disaster early warning report.

2. The method according to claim 1, characterized in that, The acquisition of crack morphology data in the mountain crack area, including crack length, width, depth, and orientation, is performed in real time through a pre-set sensor network, including: Obtain a preset sensor network layout scheme, which includes the placement positions of multiple sensor nodes; According to the sensor network layout scheme, multiple sensor nodes are deployed in the crack area of ​​the target mountain to form a sensor network covering the crack area. The sensor nodes acquire image and depth data of the crack region. The image and depth data collected by the sensor nodes are transmitted to the data processing center. At the data processing center, the image data is segmented to obtain an image of the crack region. Based on the image of the crack area, the length and width of the crack are calculated to obtain the crack size data; Based on the depth data, the depth values ​​at various locations in the crack area are determined to obtain the crack depth data; The Hough transform algorithm is used to process the image of the crack region to calculate the crack direction angle and obtain the crack direction data. By integrating the size data, depth data, and orientation data of the crack, crack morphology data characterizing the crack morphology is obtained; Based on the crack morphology data, the changes in crack morphology in the cracked area of ​​the mountain are monitored in real time.

3. The method according to claim 1, characterized in that, Based on the crack morphology data, a classification index system is established, which includes thresholds for crack length, width, depth, and orientation. A clustering algorithm is then used to classify the cracks, yielding preliminary crack classification results, including: Acquire crack morphology data, which includes crack length, width, depth, and orientation parameters; For the classification indicators of crack length, width, depth and orientation, a corresponding threshold range is set, and the threshold range is used as the judgment basis for the clustering algorithm; The K-means clustering algorithm is used to take the crack morphology data as input and perform cluster analysis according to the threshold setting of the classification index system to obtain preliminary crack classification results. The cracks of different categories in the preliminary crack classification results have similar morphological characteristics. The preliminary division results are evaluated to determine their accuracy and reasonableness. If the preliminary division results meet the requirements, the final crack division scheme will be determined. If the initial classification results do not meet the requirements, the threshold setting of the classification index system is adjusted, and the process returns to the step of using the K-means clustering algorithm to re-perform the clustering analysis until a crack classification result that meets the requirements is obtained. Based on the final crack classification scheme, a targeted crack treatment and repair plan will be developed.

4. The method according to claim 1, characterized in that, The acquisition of geological characteristic parameters of the fracture area, including lithology, structure, and stress state, is achieved through data collection via a pre-deployed geological sensor network, including: Acquire raw geological data of the fracture area, the raw geological data being derived from a pre-deployed network of geological sensors; The original geological data is preprocessed and feature extracted to obtain characteristic parameters that characterize the geological features of the fracture area. These characteristic parameters include lithology, structure, and stress state. Clustering algorithms are used to analyze the feature parameters, and the fracture region is divided into at least two geological units based on the similarity of the feature parameters. For each of the at least two geological units, a mapping relationship between the feature parameters and the degree of fracture development is established using a support vector machine algorithm to obtain the fracture development probability of each geological unit. If the probability of crack development in the geological unit exceeds a preset threshold, the geological unit will be identified as a high-risk crack zone. Otherwise, the geological unit will be classified as a low-risk fracture zone; Based on the characteristic parameters and the crack development probability, a decision tree algorithm is used to generate decision rules for crack prevention measures. A predictive model between the feature parameters and the crack propagation trend is established using a neural network algorithm to predict the development of cracks in the future. The results of the crack region division, the crack development probability, the decision rules, and the prediction results are visualized.

5. The method according to claim 1, characterized in that, The process of combining the geological feature parameters with the fracture morphology data, constructing a classification model using a support vector machine algorithm, comprehensively analyzing the fractures, and optimizing the fracture classification results includes: Geological feature parameters and fracture morphology data are acquired, and the geological feature parameters and fracture morphology data are fused to obtain a comprehensive feature dataset; Based on the comprehensive feature dataset, data preprocessing techniques are used to clean and normalize the comprehensive feature dataset to obtain standardized feature vectors. For the standardized feature vector, a feature selection algorithm is used to extract a subset of features that are more correlated with crack classification than a preset threshold. The feature subset is input into a support vector machine algorithm to obtain a crack classification model through training, and the crack classification model is optimized using a cross-validation method. Acquire the fracture data to be analyzed, extract the geological feature parameters and morphological data of the fracture data to be analyzed, and construct the feature vector to be analyzed; The feature vector to be analyzed is input into the crack classification model to obtain the classification result of the crack data to be analyzed; Based on the classification results, combined with geological knowledge and experience, the preliminary fracture classification results were optimized and adjusted to obtain the final comprehensive fracture analysis results.

6. The method according to claim 1, characterized in that, The process of transmitting crack morphology data and geological feature parameters collected in the field to the monitoring platform in real time includes: Acquire crack morphology data and geological characteristic parameters to obtain raw monitoring data; For the original monitoring data, data preprocessing techniques are used to remove noise interference and outliers to obtain cleaned and effective data. According to the preset data transmission protocol and format specifications, the cleaned valid data is packaged and encapsulated to form a standardized transmission data packet; Through the constructed wireless transmission channel, the standardized transmission data packets are sent to the remote monitoring platform in real time using encrypted transmission. After receiving the transmitted data packet, the monitoring platform uses data parsing technology to extract the crack morphology features and geological parameters to obtain structured monitoring data. The structured monitoring data is input into a pre-built machine learning model, and the support vector machine algorithm is used for feature analysis to determine the degree of danger of the crack morphology. Based on the analysis results of the hazard level of the crack morphology, combined with the preset early warning threshold, the decision tree algorithm is used to realize automatic early warning of crack disasters and push the early warning information to relevant departments.

7. The method according to claim 1, characterized in that, On the monitoring platform, streaming data processing technology is used to process the received fracture morphology data in real time, and the fracture division results are dynamically updated in conjunction with the geological feature parameters, including: Acquire crack morphology data received from the monitoring platform, and for each piece of crack morphology data, extract morphological feature parameters, including the length, width, depth and orientation of the crack, to obtain a crack morphology feature vector. Obtain the geological feature parameters corresponding to the fracture morphology data. The geological feature parameters include lithology, stratigraphic age and fault distribution. Fuse the geological feature parameters with the fracture morphology feature vector to construct a fracture morphology-geological feature fusion vector. Based on the pre-established fracture classification model, the support vector machine algorithm is used to classify the fracture morphology-geological feature fusion vector, dividing the fractures into tensile fractures and shear fractures, thus obtaining the initial fracture classification results. During the streaming data processing, the latest crack morphology data transmitted from the monitoring platform is continuously received, and its morphological feature parameters are extracted. Then, it is fused with the corresponding geological feature parameters to obtain a new fracture morphology-geological feature fusion vector; The new fracture morphology-geological feature fusion vector is input into the fracture division model, and the fracture division model is updated in real time using an incremental learning algorithm to obtain the dynamically adjusted fracture division result. The dynamically adjusted crack division results are statistically analyzed to calculate the number and length of each type of crack, generate a crack distribution statistical report, and send the statistical report to the monitoring platform for display. If a sudden increase in the number or length of cracks is detected in a certain area, an early warning mechanism is triggered. A time series analysis algorithm is used to predict the development trend of cracks in the area, assess the potential geological hazard risk, and push the early warning information to relevant departments.

8. The method according to claim 1, characterized in that, Based on the dynamically updated crack segmentation results, a visualization module is designed to generate a crack spatial distribution map and attribute feature map, including: Obtain the crack segmentation result, which includes the spatial coordinate information and attribute information of the crack; The crack division results are stored in the database; Retrieve the spatial coordinates and attribute information of the crack from the database; Based on the spatial coordinate information of the crack, a three-dimensional spatial distribution model of the crack is generated using three-dimensional modeling technology; Based on the attribute information of the crack, a chart visualization technique is used to generate an attribute feature map of the crack, which includes a statistical distribution map of the crack length, width and depth. The three-dimensional spatial distribution model of the crack is overlaid with the attribute feature map of the crack to generate a comprehensive visualization interface of the crack. The cracks were grouped using a clustering algorithm to obtain the crack classification results; The classification results of the cracks are associated with the comprehensive visualization interface of the cracks, and different categories of cracks are identified by different colors or patterns to generate a classification visualization interface of cracks.

9. The method according to claim 1, characterized in that, The process of correlating the visualized crack segmentation results with rainfall and surface displacement information in the monitoring platform to generate a disaster early warning report includes: Obtain crack segmentation results and rainfall and surface displacement monitoring information from the monitoring platform; The crack division results and the monitoring information are preprocessed to obtain preprocessed crack division results and monitoring information. The data preprocessing includes data cleaning and data normalization. Based on the preprocessed crack division results and monitoring information, a correlation analysis model is constructed. The correlation analysis model uses a time series analysis algorithm to analyze the correlation between the crack division results and the rainfall and surface displacement. The correlation analysis model is used to obtain the correlation rules and correlation strength between the crack division results and the rainfall and surface displacement. Based on the association rules and association strength, determine whether the current crack division results and monitoring information meet the preset disaster early warning conditions; If the disaster warning conditions are met, the disaster warning level is determined according to the preset warning level standards, and a disaster warning report is generated. The disaster early warning report will be displayed in a visual manner, and the early warning information will be pushed to relevant departments and personnel.

Images