CN-121808571-B - Method and system for identifying collapse hidden danger based on geological environment multi-factor data

CN121808571BCN 121808571 BCN121808571 BCN 121808571BCN-121808571-B

Abstract

The invention discloses a method and a system for identifying collapse hidden danger based on geological environment multi-factor data, and relates to the technical field of geological disaster monitoring and early warning. A system for identifying potential collapse hazards based on geological environment multi-factor data comprises a data acquisition and processing module, a potential hazard target area identification module, a three-dimensional geological modeling module, a model construction and prediction module and a risk classification and early warning module. According to the method and the system for identifying the potential collapse hazards based on the geological environment multi-factor data, the potential index of each part of the monitored area is comprehensively calculated based on the preprocessed geological environment multi-factor data by adopting the analytic hierarchy process and the information quantity process, so that not only can the expert experience and the historical data statistics rule be comprehensively synthesized, but also the weight coefficient and the information quantity contribution value of each factor can be quantitatively calculated, thereby realizing scientific quantification and accurate partitioning of the potential of the monitored area, and improving the potential collapse hazard identification effect of the method and the system for identifying the potential collapse hazards based on the geological environment multi-factor data.

Inventors

CAO JIAN
JIA CHAO
CHE LUKUAN
XIN HAO
XU FENG
WANG ZHIYUAN
LIANG HAO
PENG XIN
LIU XICHAO
PIAN LEI
ZHAN JIAN
HOU KAILUN
Gu Congnan
YANG JIANWEI
TIAN QI
CHEN JUNJIE

Assignees

天津市地质研究和海洋地质中心

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (7)

1. A method for identifying potential collapse hazards based on geological environment multi-factor data, comprising the steps of: Collecting geological environment multi-factor data of a monitoring area, wherein the geological environment multi-factor data comprises topographic data, geological structure data, meteorological hydrographic data, human activity data and deformation monitoring data, and preprocessing the geological environment multi-factor data to obtain preprocessed geological environment multi-factor data; comprehensively calculating potential indexes of all parts of a monitoring area based on preprocessed geological environment multi-factor data by adopting a hierarchical analysis method and an information quantity method, and dividing the target area of the monitoring area according to the potential indexes and a preset dividing rule to obtain a high potential target area, a medium potential target area and a low potential target area; carrying out multi-scale refined investigation on a high-potential target area in a monitoring area, constructing a quantifiable three-dimensional geomechanical model, and acquiring key early warning factor data of the high-potential target area based on preprocessed geological environment multi-factor data and the three-dimensional geomechanical model, wherein the key early warning factor data comprises development factor data, basic factor data and initiation factor data; constructing a target zone collapse risk prediction model based on a physical information neural network, preprocessing key early warning factor data, and inputting the key early warning factor data into the trained target zone collapse risk prediction model to obtain a collapse risk prediction probability of a high-potential target zone; Setting an initial grading threshold of the collapse risk prediction probability according to the historical data cumulative distribution function, adopting a Bayesian optimization algorithm to carry out self-adaptive dynamic adjustment on the initial grading threshold to obtain a current grading threshold of the collapse risk prediction probability, grading the collapse risk of the high potential target area based on the current grading threshold and the collapse risk prediction probability, and sending out an alarm of a corresponding grade.
2. The method of claim 1, wherein the topographical data includes DEM data, slope data and elevation data, the geological formation data includes formation lithology data, fault distance data and joint crack development density data, the meteorological hydrographic data includes rainfall intensity data, accumulated rainfall data and ground mass saturation data, the human activity data includes slope change rate data, road construction degree data and slope building degree data, the deformation monitoring data includes GNSS displacement data, inclination change rate data and crack change data, the development factor data includes critical rock mass stability index data and structural face adverse combination index data, the base factor data includes structural stress coefficient data, rock mass integrity coefficient data and hydrographic condition index data, and the initiation factor data includes rainfall intensity data and human activity intensity data.
3. The method for identifying potential collapse hazards based on geological environment multi-factor data according to claim 2, wherein the specific steps of preprocessing the geological environment multi-factor data to obtain preprocessed geological environment multi-factor data are as follows: Converting the geological environment multi-factor data into a time sequence format by taking the date as an index, unifying the initial reference time points of all time sequences, unifying the time frequencies of all time sequences, and filling the missing values of the time sequences with low time frequencies at new time points by using a linear interpolation or polynomial interpolation method; Processing null values or missing values existing in the time sequence, and filling the missing values by adopting a linear interpolation or polynomial interpolation method, or directly filling the missing values by using the average value or the median of the time sequence; Identifying outliers in the time series that do not conform to the expected pattern by a Z-Score method or an IQR method, repairing the outliers by deleting the outliers and replacing the outliers with the mean or median of the time series or using an interpolation method; the data of different dimensions are standardized through the minimum-maximum standardization, wherein the calculation formula of the minimum-maximum standardization is as follows: wherein The value of the original data is represented by, Representing the minimum value in the time series, Represents the maximum value in the time series, Representing the normalized data values.
4. The method for identifying potential collapse hazards based on geological environment multi-factor data according to claim 3, wherein the specific steps of comprehensively calculating potential indexes of each part of the monitored area by adopting a hierarchical analysis method and an information volume method based on the preprocessed geological environment multi-factor data are as follows: Each item of data in the preprocessed geological environment multi-factor data is respectively regarded as a type of factor, and the weight coefficients of each type of factor are obtained through an analytic hierarchy process; historical collapse disaster data of each part of the monitoring area are obtained, information quantity values of various factors on the collapse disasters are calculated by adopting an information quantity method, and a calculation formula is specifically as follows: Wherein the method comprises the steps of Represent the first The information magnitude of the class factors on the collapse disasters, Indicating the condition of collapse The probability of the occurrence of the class factor, Represent the first The probability of the occurrence of the class factor in the monitored area; And carrying out weighted summation on the information magnitude of the collapse disasters by various factors based on the weight coefficients of the various factors to obtain potential degree indexes of the monitored area.
5. The method for identifying potential collapse hazards based on geological environment multi-factor data according to claim 4, wherein the target zone collapse risk prediction model based on the physical information neural network comprises an input layer, three full-connection layers, a physical constraint layer and an output layer, wherein the input layer is used for receiving the preprocessed key early warning factor data, the first full-connection layer is used for receiving the output of the input layer and performing feature extraction to capture nonlinear relations, the second full-connection layer is used for receiving the output of the first full-connection layer and performing feature dimension reduction to extract key patterns, the third full-connection layer is used for receiving the output of the second full-connection layer and performing high-layer semantic mapping, the physical constraint layer is used for receiving the output of the third full-connection layer and embedding physical equation residuals to constrain network output to conform to mechanical mechanisms, and the output layer is used for receiving the output of the physical constraint layer and outputting the collapse risk prediction probability through a Sigmoid activation function.
6. The method for identifying potential collapse hazards based on geological environment multi-factor data according to claim 5, wherein the specific steps of setting an initial grading threshold of the collapse risk prediction probability according to the historical data accumulation distribution function, and adopting a Bayesian optimization algorithm to carry out self-adaptive dynamic adjustment on the initial grading threshold to obtain the current grading threshold of the collapse risk prediction probability are as follows: Acquiring the collapse risk prediction probability of each monitoring area before collapse from historical collapse disaster data of each monitoring area, respectively calculating 25 quantiles, 50 quantiles and 75 quantiles of the collapse risk prediction probability through a historical data accumulation distribution function, and taking the 25 quantiles, the 50 quantiles and the 75 quantiles as initial grading thresholds of the collapse risk prediction probability; Acquiring past collapse disaster data of a monitoring area, extracting early warning event samples from the past collapse disaster data, calculating early warning accuracy, early warning omission rate and early warning false alarm rate according to the early warning event samples, constructing an objective function according to the early warning accuracy, early warning omission rate and early warning false alarm rate, carrying out iterative optimization on an initial grading threshold value based on the objective function by adopting a Bayesian optimization algorithm until the objective function reaches a preset value or the variation of the objective function is smaller than the preset threshold value for a plurality of times, and obtaining the current grading threshold value of the collapse risk prediction probability.
7. A system for identifying potential collapse hazards based on geological environment multi-factor data, which is applied to the method for identifying potential collapse hazards based on geological environment multi-factor data according to any one of claims 1 to 6, and is characterized by comprising the following steps: The data acquisition and processing module is used for acquiring geological environment multi-factor data of the monitoring area, wherein the geological environment multi-factor data comprises topographic data, geological structure data, meteorological hydrologic data, human activity data and deformation monitoring data, and preprocessing the geological environment multi-factor data to obtain preprocessed geological environment multi-factor data; the hidden danger target area identification module is used for comprehensively calculating the potential index of each part of the monitoring area based on the preprocessed geological environment multi-factor data by adopting a hierarchical analysis method and an information quantity method, and carrying out target area division on the monitoring area according to the potential index and a preset division rule to obtain a high potential target area, a medium potential target area and a low potential target area; The three-dimensional geological modeling module is used for carrying out multi-scale refined investigation on the high potential target area in the monitoring area and constructing a quantifiable three-dimensional geomechanical model, and acquiring key early warning factor data of the high potential target area based on the preprocessed geological environment multi-factor data and the three-dimensional geomechanical model, wherein the key early warning factor data comprises development factor data, basic factor data and initiation factor data; the model construction and prediction module is used for constructing a target zone collapse risk prediction model based on a physical information neural network, preprocessing key early warning factor data, inputting the key early warning factor data into the trained target zone collapse risk prediction model, and obtaining the collapse risk prediction probability of the high-potential target zone; the risk classification and early warning module is used for setting an initial classification threshold value of the collapse risk prediction probability according to the historical data accumulation distribution function, carrying out self-adaptive dynamic adjustment on the initial classification threshold value by adopting a Bayesian optimization algorithm to obtain a current classification threshold value of the collapse risk prediction probability, classifying the collapse risk of the high-potential target area based on the current classification threshold value and the collapse risk prediction probability, and sending out an alarm of a corresponding level.

Description

Method and system for identifying collapse hidden danger based on geological environment multi-factor data Technical Field The invention relates to the technical field of geological disaster monitoring and early warning, in particular to a method and a system for identifying collapse hidden danger based on geological environment multi-factor data. Background The traditional geological disaster monitoring and early warning method mainly depends on single earth surface deformation measurement (such as GNSS punctiform monitoring) or periodic remote sensing investigation, lacks comprehensive consideration of multidimensional factors such as geological structure, meteorological hydrology, human activities and the like, is difficult to capture the full chain evolution process of the collapse disaster, is focused on impending disaster early warning in the prior art, is weak in early recognition capability of hidden danger points, cannot realize early discovery of hidden danger and early warning of risks, causes long-term stability at the album points and frequent occurrence of phenomena of disaster burst at the album points, and meanwhile, the early warning threshold value in the prior art is mainly statically set, cannot be dynamically adjusted according to geological environment evolution and real-time monitoring data, is difficult to adapt to the multi-scale and nonlinear collapse evolution characteristics of complex mountain areas, and causes poor recognition effect of the existing method and system on the collapse hidden danger. Based on the above situation, the invention provides a method and a system for identifying the hidden danger of collapse based on geological environment multi-factor data, which have good hidden danger identification effect. Disclosure of Invention The invention provides a method and a system for identifying potential collapse hazards based on geological environment multi-factor data, which are good in potential collapse hazard identification effect and are used for overcoming the defects that the existing geological disaster monitoring and early warning method mainly depends on single earth surface deformation measurement (such as GNSS punctiform monitoring) or periodic remote sensing investigation, lacks comprehensive consideration of multi-dimensional factors such as geological structure, weather hydrology and human activity, is difficult to capture the full chain evolution process of the potential collapse hazards, is focused on impending disaster early warning in the prior art, is weak in early identification capability of potential hazard points, cannot realize early detection of the potential hazards and early warning of risks, and causes the phenomena of long-term stability at the album points and non-sudden disaster formation at the album points. A method for identifying potential collapse hazards based on geological environment multi-factor data comprises the following steps: Collecting geological environment multi-factor data of a monitoring area, wherein the geological environment multi-factor data comprises topographic data, geological structure data, meteorological hydrographic data, human activity data and deformation monitoring data, and preprocessing the geological environment multi-factor data to obtain preprocessed geological environment multi-factor data; comprehensively calculating potential indexes of all parts of a monitoring area based on preprocessed geological environment multi-factor data by adopting a hierarchical analysis method and an information quantity method, and dividing the target area of the monitoring area according to the potential indexes and a preset dividing rule to obtain a high potential target area, a medium potential target area and a low potential target area; carrying out multi-scale refined investigation on a high-potential target area in a monitoring area, constructing a quantifiable three-dimensional geomechanical model, and acquiring key early warning factor data of the high-potential target area based on preprocessed geological environment multi-factor data and the three-dimensional geomechanical model, wherein the key early warning factor data comprises development factor data, basic factor data and initiation factor data; constructing a target zone collapse risk prediction model based on a physical information neural network, preprocessing key early warning factor data, and inputting the key early warning factor data into the trained target zone collapse risk prediction model to obtain a collapse risk prediction probability of a high-potential target zone; Setting an initial grading threshold of the collapse risk prediction probability according to the historical data cumulative distribution function, adopting a Bayesian optimization algorithm to carry out self-adaptive dynamic adjustment on the initial grading threshold to obtain a current grading threshold of the collapse risk prediction probability, grading the collapse risk of