A method and system for assessing the risk of precipitation-induced geological hazards along railway lines.

By combining multi-source data processing and rainfall pattern recognition with a risk assessment method that couples train dynamic loads, the problems of insufficient rainfall pattern differentiation and static parameter deviation in existing technologies have been solved, enabling accurate prediction of geological disasters along railway lines and efficient disaster prevention measures.

CN123117809APending Publication Date: 2026-11-17四川省气候中心(四川省气候变化中心四川省生态气象和卫星遥感中心)
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611344627.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-01
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing geological hazard risk assessment technologies along railway lines cannot accurately distinguish the hazard-causing mechanisms of different rainfall patterns, do not consider the coupling effect of train dynamic loads, and rely on static geological parameters, resulting in insufficient prediction accuracy and failing to meet the refined disaster prevention needs of mountain railways.

Method used

The system employs multi-source data acquisition and standardized preprocessing, combines rainfall time-history curves to identify rainfall patterns, dynamically switches disaster calculation models, and performs disaster probability calibration by combining train operation parameters. Through real-time image correction and two-factor coupled calculation, a two-dimensional risk assessment matrix is ​​constructed to generate differentiated operational handling instructions.

Benefits of technology

It enables accurate risk assessment of different rainfall scenarios, improves the dynamic accuracy and on-site fit of geological disaster prediction, and enhances the safety and emergency response capabilities of railway operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117809A_ABST
    Figure CN123117809A_ABST
Patent Text Reader

Abstract

This invention relates to the field of railway risk early warning technology, and in particular to a method and system for risk assessment of precipitation-induced geological disasters along railway lines. The method includes basic probability calculation of geological disasters and surface condition extrapolation, multi-source image deviation analysis and dynamic correction of geological predictions, disaster probability calibration based on train dynamic load coupling, and two-dimensional risk matrix evaluation and risk level determination. This invention employs a closed-loop calculation scheme combining surface condition extrapolation and multi-source image comparison correction. First, it extrapolates the prediction benchmark values ​​of surface features such as vegetation cover, soil properties, and slope deformation based on the rainfall-induced disaster calculation results. Then, it extracts corresponding measured surface features from inspection-collected images and satellite remote sensing images. Through deviation quantification analysis, it inversely corrects the core parameters of the geological calculations, and finally iteratively updates the geological disaster prediction results. This method can calibrate the deviation between geological model parameters and actual on-site conditions in real time, effectively overcoming prediction errors caused by static parameter distortion.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of railway risk early warning technology, and in particular to a method and system for risk assessment of geological disasters induced by precipitation along railway lines. Background Technology

[0002] my country has a high proportion of mountainous railways, with numerous cutting slopes, natural slopes, and valleys along their routes. Rainfall is the primary trigger for geological disasters such as slope collapses, landslides, and debris flows. These disasters are sudden and highly destructive, easily causing line interruptions and train accidents, seriously threatening railway transport safety and operational efficiency. Therefore, establishing a precise, efficient, and scenario-specific risk assessment system for rainfall-induced geological disasters is a core requirement for railway disaster prevention and mitigation during the flood season.

[0003] Currently, most existing geological hazard risk assessment technologies along railway lines are derived from general regional geological hazard assessment methods. They generally use cumulative rainfall as the core input parameter, combining static geological baseline data with topographic parameters to calculate hazard risk. For example, Chinese invention patent CN111160644A discloses a railway route selection method and device based on geological hazard risk assessment. It constructs a risk assessment model by collecting basic data such as regional geology, topography, and rainfall, thereby supporting the comparison and optimization of railway route options.

[0004] However, the aforementioned existing technologies still have many shortcomings in practical applications: First, they use a single disaster calculation model, failing to distinguish the different disaster-causing mechanisms of different rainfall types such as short-duration heavy rainfall, long-duration continuous rainfall, and intermittent rainfall, resulting in poor adaptability of warning thresholds and significant fluctuations in assessment accuracy under different rainfall scenarios; Second, they treat rainfall as a single disaster factor, failing to consider the coupling disaster-causing effect of train cyclic dynamic load and rainwater softening of soil unique to railway scenarios, leading to generally unsafe assessment results for high-volume trunk lines; Third, they rely on static geological parameters for one-time calculations, without establishing a dynamic correction mechanism based on real-time surface observation data, making prediction accuracy prone to decay with changes in surface environment and soil condition; Fourth, the risk assessment system does not deeply integrate railway operation-specific parameters, and the output risk level cannot directly correspond to operational actions such as inspection and control, train operation adjustments, and emergency resource deployment, limiting the practical application value of the assessment results.

[0005] In summary, existing technologies struggle to balance assessment accuracy, scenario adaptability, and business feasibility, failing to meet the development needs of refined and intelligent disaster prevention and control for mountain railways. Summary of the Invention

[0006] To overcome the problems mentioned in the background art, the present invention proposes a method and system for risk assessment of precipitation-induced geological disasters along railway lines.

[0007] The technical solution of the present invention is as follows: On one hand, a method for risk assessment of precipitation-induced geological disasters along railway lines, comprising the following steps: S1: Multi-source data acquisition and standardized preprocessing: Acquire spatiotemporal rainfall data, spatial image data, railway basic attribute data, train operation data, and geological baseline data; complete data format unification, quality verification, and spatial matching to obtain a standardized input dataset. S2: Intelligent rain pattern recognition and dynamic switching of disaster-causing models. Based on the rainfall time history curve, rainfall characteristic parameters are extracted, the current rainfall rain pattern category is automatically identified, and the corresponding disaster-causing calculation model and early warning threshold are scheduled and matched. S3: Basic probability calculation of geological hazards and surface state extrapolation. Combining geological baseline data with the matched hazard calculation model, the initial probability of geological hazard occurrence in each assessment unit is calculated; at the same time, based on the rainfall-induced hazard calculation results, the predicted characteristic values ​​of surface state are extrapolated. S4: Multi-source image deviation analysis and dynamic correction of geological prediction. Interpretation of real-time inspection images and the latest satellite images to extract measured surface feature values; comparison of deviation between measured surface feature values ​​and predicted surface condition feature values; inversion correction of geological calculation parameters; recalculation of the corrected probability of geological disaster occurrence. S5: Calibration of disaster probability coupled with train dynamic load. Train operation parameters are converted into cyclic dynamic load parameters. Combined with the soil shear strength attenuation law after precipitation infiltration, a slope stability calculation model coupled with two factors of rainfall softening and dynamic load fatigue is constructed. The corrected disaster probability is calibrated to obtain the final disaster probability and classify it. S6: Two-dimensional risk matrix evaluation and risk level determination. A two-dimensional risk evaluation matrix combining probability and consequence dimensions is constructed. The consequence dimension incorporates railway-specific parameters to calculate the severity of consequences. The final risk level is obtained by cross-determining the probability of the final disaster. S7: Generate differentiated operation and disposal instructions. Match the corresponding railway operation business actions according to the final risk level, and output disposal instructions that include inspection requirements, train operation control standards, and emergency resource pre-positioning rules.

[0008] Preferably, step S2 specifically includes: Four characteristic parameters are extracted from the rainfall time-history curve: cumulative rainfall, total rainfall duration, peak hourly rainfall intensity, and percentage of intermittent rainfall. The rainfall process is divided into three types of rainfall by preset classification rules: short-duration heavy rainfall, long-duration continuous rainfall, and intermittent rainfall. Independent disaster calculation models and differentiated rainfall warning thresholds are matched for the three types of rainfall.

[0009] As a preferred option, the disaster calculation models corresponding to the three rainfall types are as follows: Under the short-duration heavy rainfall pattern, the core calculations are surface runoff velocity, slope erosion modulus, and probability of shallow landslides. Under the long-duration continuous rainfall pattern, the core calculations are the infiltration rate of deep soil, the increase in pore water pressure, and the probability of deep landslides. Under intermittent rainfall patterns, based on soil infiltration calculations, the soil shear strength degradation coefficient corresponding to the number of wet-dry cycles is superimposed to calculate the probability of disaster occurrence after degradation correction.

[0010] Preferably, step S4 specifically includes: Image segmentation and feature extraction were performed on real-time inspection images and the latest satellite images to obtain five measured surface feature values: vegetation coverage, soil color characteristics, slope deformation, surface cover type, and slope stone density. The measured surface feature values ​​are compared with the predicted surface state feature values ​​obtained in step S3, and the relative deviation of each indicator is calculated and weighted to obtain the comprehensive deviation coefficient. Establish a mapping relationship between surface feature deviations and geological calculation parameters, and correct key parameters such as soil permeability coefficient, effective shear strength, and slope roughness coefficient in reverse based on the comprehensive deviation coefficient; The disaster-causing calculation model was rerun based on the corrected geological parameters to obtain the corrected probability of geological disaster occurrence.

[0011] Preferably, step S5 specifically includes: The vibration acceleration amplitude at the slope location is calculated based on the train axle load and running speed. The cyclic loading frequency is obtained by combining the daily traffic logarithm. At the same time, the vibration propagation attenuation coefficient is introduced to obtain the cyclic dynamic load intensity and the number of cycles borne by the slope soil. Based on the soil saturation and shear strength attenuation law after precipitation infiltration, the disaster probability increment coefficient under the combined action of rainfall softening and dynamic load fatigue is calculated. Different probability correction ranges are set for different line grades and operation types. The corrected disaster occurrence probability is then calibrated to obtain the final disaster occurrence probability.

[0012] Preferably, in step S6, the railway-specific parameters incorporated into the consequence dimension include four categories: line grade, traffic density, importance of structures, and difficulty of emergency repair; the severity of the consequences is calculated using a weighted scoring method and divided into five levels.

[0013] As a preferred option, in step S7, the final risk level is divided into five levels: low risk, medium risk, relatively high risk, high risk, and extremely high risk. Each risk level is matched with differentiated inspection frequency requirements, driving control standards, and emergency resource pre-positioning rules to form a standardized and implementable risk handling instruction system.

[0014] Preferably, in step S1, the spatiotemporal rainfall data is collected using a fusion acquisition method that combines satellite cloud image inversion with on-site measurements from multiple collection points along the route; the area rainfall data retrieved from the satellite cloud image is accessed, and combined with hourly on-site measurements from automatic rain gauges and soil moisture monitoring points along the route and the rainfall time-history curves of meteorological forecasts, to form a spatiotemporally continuous rainfall dataset.

[0015] Preferably, a model closed-loop iterative optimization step is included after step S7: After each rainfall disaster event, the actual disaster situation, on-site verification data, and response effects are fed back to the model to optimize the rainfall pattern classification threshold, image deviation correction coefficient, dynamic load coupling parameters, and risk matrix weights, thereby achieving continuous iterative optimization of the assessment model.

[0016] On the other hand, a risk assessment system for precipitation-induced geological hazards along railway lines includes: The multi-source data acquisition and management module is used to collect and preprocess spatiotemporal rainfall data, spatial image data, railway basic attribute data, train operation data and geological baseline data, and output a standardized input dataset. The rainfall pattern recognition and model scheduling module is used to extract rainfall characteristic parameters, identify rainfall pattern categories, and dynamically schedule and match corresponding disaster-causing calculation models and early warning thresholds. The geological prediction and image correction module is used to calculate the initial probability of geological disasters, extrapolate the predicted characteristic values ​​of surface conditions, and invert and correct geological parameters and disaster probabilities based on the measured characteristic deviations of multi-source images. The dynamic load coupling calibration module is used to convert the train cyclic dynamic load parameters, construct a two-factor coupled model of rainfall softening and dynamic load fatigue, and calibrate to obtain the final probability of disaster occurrence. The risk assessment and instruction generation module is used to construct a two-dimensional risk assessment matrix incorporating railway-specific parameters, determine the final risk level, and output differentiated operational handling instructions. The visual interaction module is used for displaying risk maps along the route, pushing early warning information, issuing work orders for handling, and querying historical data; The closed-loop iterative optimization module is used to receive the actual disaster occurrence, on-site verification data and disposal effect data after the end of the rainfall disaster process, and update the rainfall type classification threshold, image deviation correction coefficient, dynamic load coupling parameters and risk matrix weights.

[0017] The beneficial effects of this invention are: 1. Compared to existing technologies that use static geological baseline parameters to predict the probability of a single disaster, these methods are unable to adapt to dynamic changes in the surface environment. As vegetation cover, soil physical properties, and surface morphology evolve naturally, and rainfall erosion occurs, the initial geological parameters gradually deviate from the actual on-site conditions, leading to a continuous decline in prediction accuracy over time. Furthermore, there is a lack of effective calibration paths based on real-time observation data. This invention employs a closed-loop calculation scheme combining surface condition extrapolation and multi-source image comparison correction. First, it extrapolates the prediction baseline values ​​for surface features such as vegetation cover, soil properties, and slope deformation based on rainfall-induced disaster calculation results. Then, it extracts corresponding measured surface features from inspection-collected images and satellite remote sensing images. Through deviation quantification analysis, it inverts and corrects the core parameters of the geological calculation, ultimately iteratively updating the geological disaster prediction results. This approach has the advantage of constructing a complete closed loop of prediction, observation, and correction, enabling real-time calibration of the deviation between geological model parameters and actual on-site conditions. It effectively overcomes prediction errors caused by static parameter distortion, significantly improving the dynamic accuracy and on-site fit of geological disaster prediction.

[0018] 2. Compared to existing technologies that use cumulative rainfall as the core input and rely on a single calculation model to assess precipitation-induced geological hazards, this approach suffers from drawbacks such as ignoring the differences in the hazard-causing mechanisms of different rainfall patterns. It cannot distinguish the destructive paths and depths of short-duration heavy rainfall, long-duration continuous rainfall, and intermittent rainfall. A single warning threshold is difficult to adapt to complex and variable rainfall scenarios, leading to significant deviations in assessment results under different rainfall conditions and a persistently high risk of false alarms and missed alarms. This invention employs an automatic rainfall pattern identification and dynamic scheduling scheme based on rainfall time-history curves. It classifies rainfall patterns by extracting multiple characteristic parameters of the rainfall process and matches differentiated hazard-causing calculation models and specific warning thresholds to different rainfall patterns. For short-duration rainfall scenarios, it focuses on slope erosion and shallow landslide risk calculations; for long-duration rainfall scenarios, it focuses on deep soil infiltration and landslide risk calculations; and for intermittent rainfall scenarios, it superimposes the soil strength degradation effect caused by wet-dry cycles. This approach has the advantage of accurately matching various rainfall hazard-causing mechanisms, achieving dynamic adaptation of the risk assessment model to rainfall process characteristics, and comprehensively improving the accuracy of risk calculations and the targeted nature of warnings under different rainfall scenarios.

[0019] 3. Compared to existing technologies that only consider rainfall as a single disaster factor in geological hazard assessment, this approach has the drawback of neglecting the unique load effects of railway scenarios. It ignores the coupled disaster-causing effect of repeated vibrations from train traffic after rainwater softens the soil, which further exacerbates soil shear failure and accelerates slope instability. This results in generally unsafe assessments for freight trunk lines with high traffic density and heavy axle loads, posing significant safety hazards to railway operations. This invention employs a slope stability calculation scheme that couples rainfall softening and dynamic load fatigue. It transforms operational parameters such as train axle load, operating speed, and traffic density into vibration intensity and cyclic loading frequency parameters borne by the slope soil. Combining the soil saturation and shear strength attenuation law after rainwater infiltration, it quantifies the increase in the probability of disaster occurrence under the combined effect of the two factors. Furthermore, it sets differentiated calibration ranges for different levels and types of lines, which has the advantage of closely reflecting the actual disaster-causing mechanism of railway slopes. It fills the technical gap in general geological hazard assessment systems that do not consider train operating loads, and significantly improves the accuracy and safety redundancy of risk assessment for high-volume trunk lines.

[0020] 4. Compared to existing technologies that rely solely on single-point rainfall data collection from ground rain gauges, which suffer from limited monitoring coverage and insufficient data representativeness, mountainous railways often have complex terrain, making monitoring point deployment difficult and resulting in numerous monitoring blind spots. Single-point rainfall data cannot accurately reflect the spatial distribution characteristics of area rainfall, easily leading to distorted assessments in localized heavy rainfall scenarios. This invention employs a rainfall data fusion scheme combining satellite cloud image inversion and multi-point data collection along the railway line. It simultaneously integrates area rainfall data retrieved from satellite cloud images with measured data from ground monitoring points along the line. A weighted fusion algorithm generates a spatiotemporally continuous, high-precision rainfall dataset, balancing the breadth of area coverage with the accuracy of single-point monitoring. This effectively solves the problems of insufficient monitoring points and poor data representativeness along mountainous railways, providing a more reliable input foundation for subsequent disaster calculations. Attached Figure Description

[0021] Figure 1 The diagram shown is the overall flowchart of the method for risk assessment of precipitation-induced geological disasters along railway lines according to the present invention. Figure 2 The diagram shown is an overall architecture diagram of the precipitation-induced geological disaster risk assessment system based on railway lines according to the present invention. Figure 3 The diagram illustrates the principle of multi-source image deviation correction and geological prediction closed-loop update in the risk assessment method for precipitation-induced geological disasters along railway lines of the present invention. Figure 4 The diagram shown is a line graph comparing the deviations in disaster prediction before and after image correction in the method for risk assessment of precipitation-induced geological disasters along railway lines based on the present invention. Figure 5The figure shown is a curve illustrating the relationship between surface characteristic parameters and geological parameters in the method for risk assessment of precipitation-induced geological disasters along railway lines, based on the present invention. Figure 6 The image shown is a radar chart comparing the early warning performance of this invention with that of traditional methods. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] The invention will be further described in detail below using a scenario of assessing precipitation-induced geological hazards along a general mountain railway. In this embodiment, the railway line is divided into several independent assessment units using a 100m×100m grid. Each assessment unit corresponds to a roadbed slope, a cutting slope, or a natural slope adjacent to the railway. This method can be deployed on a railway disaster prevention and control platform via a computer program, and can complete the entire process of automated assessment by combining monitoring equipment along the line, satellite data interfaces, and inspection terminals.

[0024] Please see Figure 1 and Figure 5 In one aspect of this research, a method for risk assessment of precipitation-induced geological hazards along railway lines is provided, and the specific implementation process is as follows: S1: Multi-source data acquisition and standardized preprocessing This step completes the collection, cleaning, and spatial alignment of all the data required for the evaluation, providing standardized input in a unified format for subsequent calculations. The specific process is as follows: S11: Data Collection Range Five types of core data were collected simultaneously, covering all dimensions of disaster-causing factors, disaster-bearing body attributes, baseline conditions, and operational parameters: Spatiotemporal rainfall data: A fusion acquisition scheme of "satellite cloud image inversion + multi-point data collection along the railway line" was adopted. On the one hand, area rainfall raster data retrieved from satellite cloud images was obtained through meteorological satellite data interface, with a spatial resolution of 1km and a temporal resolution of 1 hour; on the other hand, minute-by-minute measured data from automatic rain gauges and soil moisture monitoring points deployed along the railway line were accessed, along with hourly rainfall forecast curves for the next 72 hours issued by the meteorological department.

[0025] Spatial imagery data: The system stores baseline satellite remote sensing images as ground surface baseline data, and also integrates visible light images captured by UAV inspections, on-site images uploaded by on-site inspection terminals, and high-resolution satellite remote sensing images that pass over the area regularly. All images have been pre-orthorectified.

[0026] Railway basic attribute data includes line classification, a register of structures along the line (slopes, tunnel entrances, bridge abutments, stations, etc.), distribution of emergency repair resource points along the line, and terrain slope and traffic accessibility data.

[0027] Train operation data includes dynamic operation plan data such as train axle load, operating speed, number of trains passing through daily, and passenger and freight transport types for the corresponding section.

[0028] Geological baseline data includes basic geological parameters such as soil and rock types, shear strength parameters under natural conditions, soil permeability coefficient, topographic slope, and historical disaster records for each assessment unit.

[0029] S12: Standardized Preprocessing Perform three levels of preprocessing on the above multi-source data: Spatial registration: All spatial data are uniformly converted to the CGCS2000 national geodetic coordinate system. Through spatial overlay analysis, rainfall, imagery, geological, and railway attribute data are matched to the corresponding evaluation units to achieve spatial alignment of multi-source data.

[0030] Data normalization: Parameters with different dimensions are standardized to 0-1 to eliminate the impact of dimensional differences on subsequent weighted calculations.

[0031] Quality verification: Outliers are removed using the Laida criterion, and missing data from ground stations is filled using inverse distance weighted spatial interpolation to ensure the integrity and reliability of the input data.

[0032] S13: Rainfall Data Fusion Calculation A weighted fusion algorithm is used to obtain high-precision rainfall data for each evaluation unit based on satellite-retrieved rainfall and ground station measured data. The fusion formula is as follows: ; in, The hourly rainfall for the merged assessment unit is shown in mm. The hourly rainfall is measured at ground rain gauges and is expressed in mm. The hourly rainfall data for the corresponding assessment unit obtained from satellite cloud image inversion is shown in mm. This is the weighting coefficient for the ground-measured data, with a value ranging from 0.6 to 0.8. The closer the station is to the evaluation unit, the higher the value. The weighting coefficients for satellite inversion data satisfy the following conditions: .

[0033] By integrating the above methods, the single-point accuracy of ground monitoring and the area coverage of satellite data are taken into account, thus solving the problems of insufficient monitoring points and poor data representativeness in mountainous railways.

[0034] S2: Intelligent Rain Pattern Recognition and Dynamic Switching of Disaster Models This step automatically identifies rainfall type based on rainfall time-history curves and schedules and matches the corresponding disaster-causing calculation model and early warning threshold to solve the problem of poor adaptability of a single model. The specific process is as follows: S21: Extraction of Rainfall Feature Parameters Four characteristic parameters were extracted from the hourly rainfall time-history curve: cumulative rainfall, total rainfall duration, peak hourly rainfall intensity, and rainfall interval percentage; among which the rainfall interval percentage is the ratio of the duration of hourly rainfall intensity less than 1 mm to the total duration of rainfall.

[0035] S22: Automatic Rain Type Classification Based on the above characteristic parameters, the rainfall process is divided into three categories according to the preset classification rules: Short-duration heavy rainfall: meets the criteria of total duration ≤ 6h and peak hourly rainfall intensity ≥ 20mm / h, with short-duration high-intensity rainfall and rapid surface runoff as the core characteristics; Long-duration continuous rainfall: meets the requirement of total duration ≥ 24h and intermittent proportion ≤ 10%, with continuous infiltration and deep soil saturation as the core characteristics; Intermittent rainfall: Meeting the criteria of intermittent rainfall with a proportion of ≥20% and a total duration of ≥12 hours, the rainfall process involves multiple alternations between wet and dry conditions, which can easily lead to soil strength fatigue and deterioration.

[0036] S23: Dynamic switching of disaster calculation model Based on the identified rainfall pattern, the system automatically switches between the corresponding disaster calculation model and differentiated early warning threshold. On the one hand, for short-duration heavy rainfall patterns, the disaster-causing mechanism under this pattern is mainly surface runoff erosion and shallow soil saturation landslide. The core calculations are slope erosion intensity and the initial probability of shallow landslide, and the calculation formulas are as follows: ; in, is the initial probability of shallow landslides under a short-duration heavy rainfall pattern, with a value range of [0,1]. This is the slope erosion modulus, expressed in t / (km²). h), is positively correlated with hourly rainfall intensity and slope roughness; The terrain slope of the assessment unit is expressed in degrees (°). This represents the natural cohesion of shallow soil, expressed in kPa. This is the model's empirical coefficient, calibrated based on regional geological data, with a typical value of 0.008.

[0037] In this mode, the hourly rainfall intensity threshold is used as the early warning indicator, and four levels of rainfall intensity thresholds are set to correspond to different early warning levels.

[0038] On the one hand, for long-duration continuous rainfall patterns, the disaster-causing mechanism under this pattern is mainly deep landslides caused by deep rainwater infiltration, increased pore water pressure, and reduced effective stress. The core calculation is the infiltration volume of deep soil and the initial probability of landslides. The calculation formula is as follows: ; in, is the initial probability of deep landslides under a long-duration continuous rainfall pattern, with a value range of [0,1]. This is the pore water pressure in deep soil, expressed in kPa, and is calculated from the cumulative infiltration rate using Darcy's law. This represents the effective self-weight stress of the soil, expressed in kPa. is the internal friction angle of the soil, in degrees; This is the model's empirical coefficient, calibrated based on regional landslide statistics, with a typical value of 0.05.

[0039] This model uses a dual threshold of cumulative rainfall and rainfall duration as early warning indicators.

[0040] On the one hand, for intermittent rainfall patterns, in addition to rainwater infiltration, the soil strength deterioration effect caused by wet-dry cycles is superimposed. The core calculation of the initial probability of disaster after strength deterioration correction is as follows: ; in, This represents the initial probability of disaster occurrence after degradation correction under intermittent rainfall pattern, with a value range of [0,1]. If the calculated result exceeds 1, it is taken as 1. The baseline disaster probability, which does not consider the wet-dry cycle, is calculated based on the cumulative rainfall during the same period. This refers to the number of wet and dry cycles during a rainfall process, i.e., the number of times rainfall and intermittent periods alternate. The strength degradation coefficient is determined by a single wet-dry cycle, and its value is determined according to the soil type. For silty clay, the typical value is 0.12.

[0041] This model uses a dual threshold of cumulative rainfall and number of wet-dry cycles as early warning indicators.

[0042] S3: Basic Probability Calculation of Geological Hazards and Surface Condition Deduction S31: Basic Probability Calculation Based on the disaster-causing calculation model matched in step S2, and combined with the geological background data and rainfall data input in step S1, the initial probability of geological disaster occurrence is calculated unit by unit. This serves as the baseline value for subsequent adjustments.

[0043] S32: Surface Condition Characteristics Deduction Based on the calculation results of rainfall-induced disasters, the predicted characteristic values ​​of the land surface state of the assessment unit under the current rainfall are deduced, serving as a benchmark for subsequent image comparison. A total of five core characteristics are deduced: Vegetation cover prediction: Rainfall-induced slope erosion leads to vegetation loss, and changes in vegetation cover are positively correlated with slope erosion intensity. The derivation formula is as follows: ; in, The current vegetation cover obtained through deduction has a value range of [0,1]. The vegetation cover for the baseline period was extracted from baseline satellite imagery. The slope erosion modulus for this rainfall event is expressed in t / (km²). h); The erosion-vegetation attenuation coefficient is typically 0.0002.

[0044] Soil color feature prediction: Increased soil moisture content leads to a decrease in image grayscale value. Soil color grayscale value is predicted based on soil saturation. Slope deformation and displacement prediction: Based on the soil expansion and sliding trend, the predicted value of the slope surface displacement is deduced; Land cover type prediction: extrapolating the range of changes in land cover types such as exposed land and water cover based on the extent of erosion; Predicted density of rocks on slope: The predicted density of rocks exposed after the surface soil is eroded based on the slope erosion intensity.

[0045] The above five predictive features together constitute the land surface condition prediction baseline set, which is used for subsequent deviation comparison with measured images.

[0046] S4: Multi-source image bias analysis and dynamic correction for geological prediction like Figure 3 and Figure 4 As shown, this step compares the deviation between the measured images and the predicted values, inverts and corrects the geological model parameters, and realizes the dynamic closed-loop update of the prediction results. The specific process is as follows: S41: Feature Extraction from Measured Images Interpretation and processing of real-time drone inspection images, on-site acquired images, and the latest satellite imagery: A semantic segmentation network was used to classify ground features in the images and extract the measured vegetation cover and land cover type. Histogram analysis was used to extract the mean grayscale value and color component characteristics of soil color. The slope deformation was extracted using a feature point matching method. The density of stones on the slope is extracted by analyzing texture features.

[0047] Finally, five measured surface feature values ​​were obtained, which correspond one-to-one with the predicted features in step S3.

[0048] S42: Deviation Quantization Calculation The relative deviation between each measured feature and its corresponding predicted feature is calculated, and then a weighted fusion is performed to obtain the comprehensive deviation coefficient. The calculation formula is as follows: ; in, The comprehensive deviation coefficient is used to evaluate the unit; the larger the value, the higher the deviation between the prediction and the actual value. For the first Measured values ​​of surface features; For the first Predicted values ​​for surface features; For the first The weight coefficients of the feature satisfy the following conditions: Among them, vegetation coverage and soil color have relatively high weights, with typical values ​​of 0.3 and 0.25 respectively, while the other three items each have a weight of 0.15.

[0049] S43: Inversion Correction of Geological Calculation Parameters Establish a mapping relationship between surface feature deviations and geological calculation parameters, and correct the core parameters of the model in reverse based on the comprehensive deviation coefficient: When the measured value of vegetation coverage is lower than the predicted value, it indicates that the slope erosion intensity is higher than expected, and the slope erosion coefficient should be increased and the soil erosion resistance parameter should be reduced accordingly. When the measured value of soil saturation is higher than the predicted value, it indicates that the infiltration rate is faster than expected, and the soil permeability coefficient should be increased and the cohesion parameter should be decreased accordingly. When the measured value of the rock density is higher than the predicted value, it indicates that the surface soil is being eroded more deeply, and the erosion depth parameter should be increased accordingly.

[0050] The general formula for parameter correction is: ; in, These are the corrected geological calculation parameters (such as permeability coefficient, cohesion, erosion coefficient, etc.). These are the baseline geological parameters before the correction. This is the deviation-parameter correction coefficient, with different values ​​corresponding to different parameters. The typical value for the permeability coefficient is 0.8, and the typical value for cohesion is -0.6. This is the overall deviation coefficient.

[0051] S44: Prediction Update Substitute the corrected geological parameters into the disaster-causing calculation model corresponding to step S2, and recalculate the corrected probability of geological disaster occurrence. It completes a closed-loop calculation of "prediction-observation-correction-update", which significantly improves the matching degree between the prediction results and the actual on-site conditions.

[0052] S5: Disaster Probability Calibration of Train Dynamic Load Coupling This step focuses on the unique cyclic dynamic load effect of trains in railway scenarios, quantifies the coupled disaster-causing effect of "rainfall softening + dynamic load fatigue", and performs a secondary calibration on the corrected disaster probability. The specific implementation process is as follows: S51: Cyclic Dynamic Load Parameter Conversion Transform train operating parameters into dynamic load parameters for the slope soil: Calculate the vibration acceleration amplitude at the slope soil based on the train axle load, running speed, and the horizontal distance between the slope and the track. Based on the number of trains passing through each day, the number of daily cyclic loading times is calculated, and the total number of cyclic loading times during the rainfall process is obtained by combining the rainfall duration.

[0053] The formula for calculating the amplitude of vibration acceleration is as follows: ; in, This represents the vibration acceleration amplitude at the slope surface, in m / s². This refers to the axle load of the train, expressed in tons (t). The train's operating speed is expressed in km / h. This is the horizontal distance from the centerline of the track to the toe of the slope, in meters. This is the vibration propagation attenuation coefficient, which is related to the soil type, and typically has a value of 0.03.

[0054] S52: Calculation of Incremental Probability of Two-Factor Coupling Rainwater softening reduces the shear strength of the soil, while cyclic dynamic loading induces cumulative plastic deformation. The coupled effect of these two factors further increases the probability of disasters. A formula for calculating the coupling increment coefficient is constructed as follows: ; in, The probability increment coefficient of disaster caused by dynamic load coupling is dimensionless; This represents the soil saturation after rainfall infiltration, with a value range of [0,1]. The amplitude of vibration acceleration at the slope is expressed in m / s². This represents the total number of loading cycles during the rainfall process, expressed in cycles. The reference vibration acceleration threshold is typically 0.1 m / s². This is the coupling effect coefficient, which is related to the soil type. The typical value for silty clay is 0.002.

[0055] S53: Differential Probability Calibration Different calibration ranges are set for different line grades and operation types: the calibration range for heavy axle load freight trunk lines is increased by 20%, the benchmark value is used for passenger dedicated lines, and the calibration range for low-capacity conventional railways is decreased by 30%.

[0056] The final calibrated formula for calculating the probability of disaster occurrence is: ; in, This represents the final probability of the disaster occurring, with a value range of [0,1]. If the calculated result exceeds 1, it is taken as 1. The probability of geological disaster occurrence after correction in step S4; This is the dynamic load coupling increment coefficient; The calibration coefficient for line type is 1.2 for freight trunk lines, 1.0 for passenger dedicated lines, and 0.7 for conventional railways.

[0057] The final probability is divided into 5 levels: low probability (<0.2), relatively low probability (0.2-0.4), medium probability (0.4-0.6), relatively high probability (0.6~0.8), and high probability (>0.8).

[0058] S6: Two-Dimensional Risk Matrix Evaluation and Risk Level Determination This step constructs a two-dimensional risk assessment matrix based on probability and consequence, incorporates railway-specific parameters to quantify the severity of consequences, and ultimately determines the risk level. The specific implementation process is as follows: S61: Quantification of the severity of consequences A weighted scoring method is adopted, incorporating four types of railway-specific parameters to calculate the comprehensive score. The four types of parameters and scoring rules are as follows: Line classification: High-speed railway 10 points, passenger dedicated line 8 points, freight trunk line 7 points, conventional railway 4 points; Traffic density: 80 or more pairs of vehicles per day: 10 points; 40-80 pairs of vehicles per day: 7 points; 20-40 pairs of vehicles per day: 4 points; less than 20 pairs of vehicles per day: 2 points. Importance of structures: Extra-large bridges / long tunnels / steep slopes 10 points, general bridges and tunnels / stations 7 points, ordinary roadbeds 3 points; Repair difficulty: Steep terrain, no direct road 10 points, average transportation 7 points, convenient transportation 3 points.

[0059] Formula for calculating the overall consequences score: ; in, The score is a comprehensive score for the consequences of the disaster, ranging from 0 to 10 points. Rate the line's quality; Rate traffic density; Assess the importance of the structure; The difficulty of the emergency repair was rated. The weights for the corresponding parameters are 0.4, 0.3, 0.2, and 0.1, respectively, satisfying the condition that the sum of the weights is 1.

[0060] The severity of consequences is divided into 5 levels based on the overall consequence score: minor (<2 points), moderate (2-4 points), serious (4-6 points), serious (6-8 points), and extremely serious (>8 points).

[0061] S62: Two-Dimensional Risk Matrix Rating A 5×5 two-dimensional risk assessment matrix is ​​constructed with the probability of the final disaster occurring as the vertical axis and the severity of the disaster consequences as the horizontal axis. The final risk level is determined by cross-judgment, and is divided into five levels: low risk, medium risk, relatively high risk, high risk, and extremely high risk. The judgment rule follows the principle that "the higher the probability and the more severe the consequences, the higher the risk level." For example, high probability + extremely severe consequences correspond to extremely high risk, and low probability + minor consequences correspond to low risk.

[0062] S7: Generation of Differentiated Operational Handling Instructions Establish a one-to-one correspondence between risk levels and railway operation actions, forming a standardized handling instruction system that is directly pushed to on-site operation and maintenance and dispatch departments. The specific rules are as follows: Low risk: Maintain the regular monthly inspection frequency, operate the vehicle at normal speed, no need for advance emergency resources, only maintain daily monitoring; Medium risk: Increase the frequency of inspections in key sections to once every ten days, increase the frequency of remote automatic monitoring, maintain normal train operation, and ensure that local work areas have emergency supplies and personnel on standby. Higher risk: Daily manual inspections will be increased, key risk points will be staffed with dedicated personnel, trains in relevant sections will be limited to 60km / h, and emergency rescue supplies will be prepositioned in the nearest work area; High risk: Train service in key high-risk areas will be suspended, 24-hour continuous monitoring will be initiated, and emergency rescue teams and equipment will be assembled on-site and ready to respond. Extremely high risk: The corresponding railway section will be closed, personnel and equipment in dangerous areas will be evacuated, the railway geological disaster emergency rescue plan will be officially activated, and emergency management personnel at all levels will be in place to direct operations.

[0063] S8: Model Closed-Loop Iterative Optimization After each rainfall event, the actual disaster occurrence and response data verified on-site are fed back to the model database. The rain pattern classification threshold, image deviation correction coefficient, dynamic load coupling parameters and risk matrix weights are optimized through parameter calibration algorithms to achieve continuous self-iteration of the assessment model and continuously improve the assessment accuracy.

[0064] like Figure 2As shown, the present invention also provides a risk assessment system for precipitation-induced geological disasters along railway lines, used to perform the above assessment method. The system is deployed on the railway disaster prevention and control platform server, and the specific implementation of each module is as follows: Multi-source data acquisition and management module: Configures satellite data interface, ground monitoring station data transmission interface, inspection terminal upload interface, operation management system interface and geological database to realize automatic access, storage and standardized preprocessing of five types of data, and outputs standardized datasets in a unified format; Rainfall pattern recognition and model scheduling module: It has built-in rainfall feature extraction algorithm, rainfall pattern classification rule library and three types of disaster calculation model library, which automatically recognizes rainfall patterns and schedules corresponding calculation models and warning thresholds; Geological prediction and image correction module: It has built-in basic probability calculation unit, surface feature inference unit, image intelligent interpretation unit, deviation analysis and parameter correction unit to complete the closed-loop calculation of prediction-comparison-correction-update; Dynamic load coupling calibration module: It has built-in dynamic load parameter conversion unit, two-factor coupling calculation unit and probability classification unit to realize the coupling disaster calibration of train dynamic load and rainfall; Risk assessment and instruction generation module: It has a built-in two-dimensional risk matrix calculation unit, disposal rule library and instruction push unit, which automatically determines the risk level and generates standardized disposal instructions, and pushes them to relevant business terminals; Visualization and Interaction Module: Based on a GIS map, it displays the risk distribution along the route, provides early warning information pop-ups, work order issuance, and historical data query and statistics functions to support daily operation and maintenance and emergency command. Specific Implementation To further quantify and verify the technical effects of this invention, a comparative calculation verification was conducted using a high slope assessment unit at K123+400 of a freight trunk railway in the mountainous region of southwestern my country as the object. The basic parameters and verification scenario settings for this assessment unit are as follows: Basic geological parameters of the slope: slope height 32m, average slope 35°, soil type is silty clay, natural cohesion. internal friction angle Natural permeability Vegetation coverage in the baseline period ; Railway operating parameters: The line is classified as a freight trunk line, with 45 pairs of trains passing through daily, an average axle load of 25t, an operating speed of 80km / h, and a horizontal distance of 12m from the centerline of the line to the toe of the slope. Structures and repair conditions: This section is a steep road cut slope, adjacent to a large bridge structure. The repair access is a simple mountain road, and the terrain accessibility is generally poor. Rainfall verification scenario: The total duration of this rainfall event was 36 hours, with a cumulative rainfall of 182 mm and a peak hourly rainfall intensity of 16 mm / h. There were two obvious intervals in the rainfall event, with a total interval duration of about 8 hours, which is a typical intermittent rainfall.

[0066] The results of calculations using traditional general geological hazard assessment methods (single cumulative rainfall model, no-rainfall type differentiation, no-dynamic load coupling, and general consequence evaluation) and the method of this invention are compared as follows: Results of traditional methods: Traditional methods use cumulative rainfall as the core input and employ a uniform shallow landslide calculation model. They do not differentiate between rainfall patterns and dynamic loads, and only use general population and economic indicators to evaluate the disaster consequences. The calculated probability of disaster occurrence for this slope is 0.42, classifying it as a medium probability level; the severity of the consequences is classified as a general level, resulting in a final risk level of medium risk. The corresponding response measure is simply "strengthening daily patrols."

[0067] The complete calculation results of the method of this invention are as follows: (1) Multi-source data fusion The assessment unit corresponds to a ground-based rain gauge's measured cumulative rainfall of 178 mm and a satellite cloud image inversion area rainfall of 186 mm. Ground data weights are used. Satellite data weighting The result calculated using the fusion formula is: The fused rainfall data balances the accuracy of individual points with the representativeness of the area, eliminating local errors in single-site data.

[0068] (2) Rain pattern recognition and model switching Extracted rainfall characteristic parameters: total duration 36 hours, percentage of intermittent rainfall. If it conforms to the classification rules for intermittent rainfall, it will automatically switch to the intermittent rainfall-induced disaster calculation model.

[0069] First, calculate the baseline disaster probability without considering the wet-dry cycle. Number of wet and dry cycles during this rainfall event Deterioration coefficient of silty clay in a single wet-dry cycle Substituting into the intermittent probability formula, we obtain the initial disaster probability: If the traditional long-duration continuous rainfall model is mistakenly used for calculation, the initial probability obtained is only 0.48, which deviates from the actual disaster-causing effect of intermittent rainfall by about 14%, verifying the necessity of adaptive switching of rainfall patterns.

[0070] (3) Surface condition projection The slope erosion modulus was obtained from this rainfall calculation. Erosion-vegetation attenuation coefficient The predicted value of current vegetation cover is obtained by extrapolation: Simultaneous simulation yields four other predicted characteristic values, including soil color grayscale and slope deformation, which together form the prediction baseline set.

[0071] (4) Image deviation correction During this rainfall event, drone-collected images yielded a measured vegetation cover of 0.58. Combined with the other four measured characteristics, a comprehensive deviation coefficient was calculated. .

[0072] Taking the soil permeability coefficient as an example, the deviation-parameter correction factor is taken. Corrected permeability coefficient: Similarly, after correcting parameters such as cohesion and slope erosion coefficient, the results are recalculated using the disaster-causing model to obtain the corrected probability of disaster occurrence. Compared to the original value of 0.558, the probability value increased by approximately 12%, which better reflects the actual degree of erosion on site and verifies the accuracy improvement effect of image closed-loop correction.

[0073] (5) Dynamic load coupling calibration Calculations show that the vibration acceleration amplitude at the slope surface is... Total number of cyclic loading times within 36 hours of rainfall duration The result is 68 times.

[0074] Soil saturation Reference vibration acceleration threshold Coupling effect coefficient Substituting into the coupling increment coefficient formula: ; This route is a freight trunk line, and the route calibration coefficient is used. The final calibrated probability of disaster occurrence: The final probability of 0.816 is considered high. Compared to 0.625 without considering dynamic loads, the probability value increases by approximately 30.6%, fully demonstrating the coupled disaster-causing effect of "rainfall softening + dynamic load fatigue." This verifies the accuracy improvement of the risk assessment of busy trunk lines by this invention and avoids the problem of traditional methods being biased towards unsafety.

[0075] (6) Two-dimensional risk matrix rating According to the railway-specific consequence scoring rules, the four parameters are scored as follows: line grade (7 points), traffic density (7 points), importance of structures (7 points), and difficulty of emergency repair (7 points). Substituting these into the comprehensive consequence score formula: The consequences score is 7.0, which falls under the category of serious consequences.

[0076] Based on the high probability level, and through cross-determination using a 5×5 two-dimensional risk matrix, the final risk level is determined to be extremely high.

[0077] (7) Generation of disposal instructions Based on the extremely high risk level, the corresponding response instructions are automatically matched: the corresponding line section is closed, train traffic is suspended, 24-hour uninterrupted monitoring is initiated, emergency rescue teams and equipment are assembled on-site and ready to be deployed, and the geological disaster emergency response is officially launched.

[0078] like Figure 6 As shown above, the comparative calculations in this embodiment can verify that: Improved accuracy in rain pattern adaptation: For intermittent rainfall scenarios, the disaster probability calculation of this invention has a deviation of about 14% compared with a single long-duration model, effectively solving the problem of poor adaptability of a single model; The dynamic correction effect is significant: the multi-source image deviation correction improves the consistency between the prediction results and the actual field conditions by about 12%, overcoming the defect of static geological parameters being distorted by environmental changes; The coupling mechanism is more in line with reality: the coupling of train dynamic loads increases the disaster probability calibration by about 30.6%, filling the gap in traditional assessments that do not consider railway operating loads, and the assessment results are more consistent with the actual disaster-causing laws of railway slopes; The risk management is more practical: the risk matrix, which incorporates railway-specific parameters, corrects the risk level from medium risk to extremely high risk in the traditional method, and upgrades the management measures from strengthening patrols to closing the line and being on standby for emergencies, effectively avoiding the risk of missed alarms and directly supporting the safe management and control of railway operations.

[0079] Those skilled in the art will understand that the above embodiments are merely exemplary general scenario illustrations, and the scope of protection of this invention is determined by the claims. Conventional adjustments and improvements to the parameter values ​​and module divisions made without departing from the principles of this invention are all within the scope of protection of this invention.

Claims

1. A method for assessing the risk of precipitation-induced geological disasters along a railway line, characterized in that, Includes the following steps: S1: Multi-source data acquisition and standardized preprocessing: Acquire spatiotemporal rainfall data, spatial image data, railway basic attribute data, train operation data, and geological baseline data; complete data format unification, quality verification, and spatial matching to obtain a standardized input dataset. S2: Intelligent rain pattern recognition and dynamic switching of disaster-causing models. Based on the rainfall time history curve, rainfall characteristic parameters are extracted, the current rainfall rain pattern category is automatically identified, and the corresponding disaster-causing calculation model and early warning threshold are scheduled and matched. S3: Basic probability calculation of geological hazards and surface state extrapolation. Combining geological baseline data with the matched hazard calculation model, the initial probability of geological hazard occurrence in each assessment unit is calculated; at the same time, based on the rainfall-induced hazard calculation results, the predicted characteristic values ​​of surface state are extrapolated. S4: Multi-source image deviation analysis and dynamic correction of geological prediction. Interpretation of real-time inspection images and the latest satellite images to extract measured surface feature values; comparison of deviation between measured surface feature values ​​and predicted surface condition feature values; inversion correction of geological calculation parameters; recalculation of the corrected probability of geological disaster occurrence. S5: Calibration of disaster probability coupled with train dynamic load. Train operation parameters are converted into cyclic dynamic load parameters. Combined with the soil shear strength attenuation law after precipitation infiltration, a slope stability calculation model coupled with two factors of rainfall softening and dynamic load fatigue is constructed. The corrected disaster probability is calibrated to obtain the final disaster probability and classify it. S6: Two-dimensional risk matrix evaluation and risk level determination. A two-dimensional risk evaluation matrix combining probability and consequence dimensions is constructed. The consequence dimension incorporates railway-specific parameters to calculate the severity of consequences. The final risk level is obtained by cross-determining the probability of the final disaster. S7: Generate differentiated operation and disposal instructions. Match the corresponding railway operation business actions according to the final risk level, and output disposal instructions that include inspection requirements, train operation control standards, and emergency resource pre-positioning rules.

2. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, Step S2 specifically includes: Four characteristic parameters are extracted from the rainfall time-history curve: cumulative rainfall, total rainfall duration, peak hourly rainfall intensity, and percentage of intermittent rainfall. The rainfall process is divided into three types of rainfall by preset classification rules: short-duration heavy rainfall, long-duration continuous rainfall, and intermittent rainfall. Independent disaster calculation models and differentiated rainfall warning thresholds are matched for the three types of rainfall.

3. The method for assessing the risk of precipitation-induced geohazards along a railway according to claim 2, characterized in that, The specific disaster calculation models corresponding to the three types of rainfall are as follows: Under the short-duration heavy rainfall pattern, the core calculations are surface runoff velocity, slope erosion modulus, and probability of shallow landslides. Under the long-duration continuous rainfall pattern, the core calculations are the infiltration rate of deep soil, the increase in pore water pressure, and the probability of deep landslides. Under intermittent rainfall patterns, based on soil infiltration calculations, the soil shear strength degradation coefficient corresponding to the number of wet-dry cycles is superimposed to calculate the probability of disaster occurrence after degradation correction.

4. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, Step S4 specifically includes: Image segmentation and feature extraction were performed on real-time inspection images and the latest satellite images to obtain five measured surface feature values: vegetation coverage, soil color characteristics, slope deformation, surface cover type, and slope stone density. The measured surface feature values ​​are compared with the predicted surface state feature values ​​obtained in step S3, and the relative deviations of each indicator are calculated and weighted to obtain the comprehensive deviation coefficient. Establish a mapping relationship between surface feature deviations and geological calculation parameters, and correct key parameters such as soil permeability coefficient, effective shear strength, and slope roughness coefficient in reverse based on the comprehensive deviation coefficient; The disaster-causing calculation model was rerun based on the corrected geological parameters to obtain the corrected probability of geological disaster occurrence.

5. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, Step S5 specifically includes: The vibration acceleration amplitude at the slope location is calculated based on the train axle load and running speed. The cyclic loading frequency is obtained by combining the daily traffic logarithm. At the same time, the vibration propagation attenuation coefficient is introduced to obtain the cyclic dynamic load intensity and the number of cycles borne by the slope soil. Based on the soil saturation and shear strength attenuation law after precipitation infiltration, the disaster probability increment coefficient under the combined action of rainfall softening and dynamic load fatigue is calculated. Different probability correction ranges are set for different line grades and operation types. The corrected disaster occurrence probability is then calibrated to obtain the final disaster occurrence probability.

6. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, In step S6, the railway-specific parameters incorporated into the consequence dimension include four categories: line grade, traffic density, importance of structures, and difficulty of emergency repair; the severity of the consequences is calculated using a weighted scoring method and divided into five levels.

7. The method for assessing the risk of precipitation-induced geohazards along a railway according to claim 1, characterized in that, In step S7, the final risk level is divided into five levels: low risk, medium risk, relatively high risk, high risk, and extremely high risk. Each risk level is matched with differentiated inspection frequency requirements, vehicle control standards, and emergency resource pre-positioning rules to form a standardized and implementable risk handling instruction system.

8. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, In step S1, the spatiotemporal rainfall data is collected using a fusion acquisition method that combines satellite cloud image inversion with on-site measurements from multiple collection points along the route. The area rainfall data retrieved from the satellite cloud image is then integrated with hourly on-site measurements from automatic rain gauges and soil moisture monitoring points along the route, along with the rainfall time-history curves from meteorological forecasts, to form a spatiotemporally continuous rainfall dataset.

9. The method for risk assessment of precipitation-induced geological hazards along railway lines according to claim 1, characterized in that, The model closed-loop iterative optimization step is also included after step S7: After each rainfall disaster event, the actual disaster situation, on-site verification data, and response effects are fed back to the model to optimize the rainfall pattern classification threshold, image deviation correction coefficient, dynamic load coupling parameters, and risk matrix weights, thereby achieving continuous iterative optimization of the assessment model.

10. A risk assessment system for precipitation-induced geological hazards along railway lines, used to implement the assessment method according to any one of claims 1-9, characterized in that, include: The multi-source data acquisition and management module is used to collect and preprocess spatiotemporal rainfall data, spatial image data, railway basic attribute data, train operation data and geological baseline data, and output a standardized input dataset. The rainfall pattern recognition and model scheduling module is used to extract rainfall characteristic parameters, identify rainfall pattern categories, and dynamically schedule and match corresponding disaster-causing calculation models and early warning thresholds. The geological prediction and image correction module is used to calculate the initial probability of geological disasters, extrapolate the predicted characteristic values ​​of surface conditions, and invert and correct geological parameters and disaster probabilities based on the measured characteristic deviations of multi-source images. The dynamic load coupling calibration module is used to convert the train cyclic dynamic load parameters, construct a two-factor coupled model of rainfall softening and dynamic load fatigue, and calibrate to obtain the final probability of disaster occurrence. The risk assessment and instruction generation module is used to construct a two-dimensional risk assessment matrix incorporating railway-specific parameters, determine the final risk level, and output differentiated operational handling instructions. The visual interaction module is used for displaying risk maps along the route, pushing early warning information, issuing work orders for handling, and querying historical data; The closed-loop iterative optimization module is used to receive the actual disaster occurrence, on-site verification data and disposal effect data after the end of the rainfall disaster process, and update the rainfall type classification threshold, image deviation correction coefficient, dynamic load coupling parameters and risk matrix weights.

Citation Information

Patent Citations

  • Railway line selection method and device based on geological disaster risk assessment

    CN111160644A