Disaster prediction method based on geological safety

Abstract

The invention relates to the technical field of geological disaster prediction, and discloses a disaster prediction method based on geological safety, and the method comprises the steps: constructing a multi-source heterogeneous historical monitoring data set; obtaining a geological safety index based on the geological safety monitoring data set; constructing a geological safety prediction model according to the geological safety index; and inputting real-time monitoring data into the geological safety prediction model to obtain a disaster risk prediction result, and triggering a corresponding early warning mechanism.

Description

Technical Field

[0001] This invention relates to the field of geological disaster prediction technology, specifically a disaster prediction method based on geological safety. Background Technology

[0002] Frequent geological disasters pose a serious threat to people's lives and property, as well as the operational safety of critical infrastructure. Currently, the methods for predicting and warning of geological disasters in this field have the following main shortcomings: (1) Data utilization is singular and lacks comprehensive evaluation. Existing methods mostly rely on a single type or a few types of monitoring data for judgment. Such methods cannot fully reflect the "health status" of the complex geological system under the coupling of multiple internal and external dynamics. They ignore the comprehensive influence of multiple sources of information such as geological conditions and human activities, resulting in an incomplete understanding of the disaster gestation process and insufficient basis for prediction.

[0003] (2) The analysis methods are static and lack dynamic foresight. Most prediction models are based on historical static data for statistical analysis or machine learning, which cannot respond to and evaluate the dynamic changes in the geological environment in real time and continuously. They also lack the ability to effectively predict future development trends and are unable to provide sufficient early warning time before disasters occur.

[0004] (3) Traditional statistical models or simple machine learning models are difficult to capture the complex nonlinear relationships in the geological disaster system. At the same time, early warnings are mostly based on fixed physical quantity thresholds and lack a dynamic adjustment mechanism that takes into account the geological body’s previous “stress history” and current “safety status”. This “one-size-fits-all” early warning mode has a high rate of false alarms and missed alarms in complex and ever-changing geological environments, which can easily lead to early warning failure or the “crying wolf” effect, reducing the authority of early warnings and public trust.

[0005] Therefore, this application proposes a disaster prediction method based on geological safety to overcome the above-mentioned shortcomings. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this application provides a disaster prediction method based on geological safety.

[0007] (II) Technical Solution To address the above problems, this application provides the following technical solution: A disaster prediction method based on geological safety includes: Construct a multi-source heterogeneous historical monitoring dataset; The geological safety index is obtained based on the geological safety monitoring dataset. A geological safety prediction model is constructed based on the geological safety index. Real-time monitoring data is input into the geological safety prediction model to obtain disaster risk prediction results and trigger the corresponding early warning mechanism.

[0008] Preferably, the construction of the multi-source heterogeneous historical monitoring dataset specifically includes: The historical monitoring dataset includes the following sub-data feature types: geological ontology data, environmentally induced data, seismic activity data, geological deformation data, and human activity data; Obtain geological data of the target area within a period, including the properties of soil and rock, structural features, and topography. Acquire environmental induced data for the target area within a given period, including meteorological data and hydrological data; the meteorological data includes rainfall, temperature, or number of freeze-thaw cycles; the hydrological data includes changes in groundwater level or pore water pressure. Acquire seismic activity data for the target region within a period, including microseismic frequency, magnitude, or focal depth; Acquire geological deformation data of the target area within a period, including surface displacement, deep displacement, or building deformation; Acquire human activity data for the target area within a period, including excavation disturbance or drainage activities; The geological data, environmentally induced data, seismic activity data, geological deformation data, and human activity data are cleaned, time-aligned, and standardized to obtain the historical monitoring dataset.

[0009] Preferably, obtaining the geological safety index based on the geological safety monitoring dataset specifically includes: The geological safety index includes deformation stability index, hydraulic stability index, mechanical stability index, structural stability index, and activity disturbance index; The deformation stability values ​​are obtained based on geological deformation data, and the mathematical expression for the deformation stability index is: (1) In formula (1), The deformation index is... , , , The attenuation coefficient is... For displacement acceleration, This is the cumulative displacement. The historical maximum safe displacement threshold within the target area. The coefficient for the trend term; The hydraulic stability index is obtained based on environmentally induced data, and its mathematical expression is as follows: (2) In formula (2), The hydraulic stability index, The hydrological sensitivity coefficient for the target area. This represents the effective rainfall over the past 7 days. The critical effective rainfall amount, For real-time groundwater levels, This is the critical groundwater level. The threshold for normal water level fluctuations used for normalization. and These are the weighting coefficients. This refers to the number of freeze-thaw cycles. To measure the pore water pressure, Displacement reference hole pressure; The mechanical stability index is obtained based on seismic activity data, and its mathematical expression is as follows: (3) In formula (3), This is the mechanical stability index. To accumulate the energy release from microseismic events, the energy of all microseismic events during the monitoring period is summed and calculated. This is the critical cumulative energy release threshold. This is the amplification factor, used to adjust the degree of nonlinear response.

[0010] The structural stability index is obtained based on geological ontology data, and its mathematical expression is as follows: (4) In formula (4), It is a structural stability index. , and These are the weighting coefficients. For rock mass quality rating, This is the value after normalizing the terrain slope. To score the relationship between the structural plane and the slope aspect combination. , , ; The activity interference index is obtained based on human activity data, and its mathematical expression is as follows: (5) In formula (5), The activity interference index, To determine the intensity of excavation disturbance. For drainage volume, and Let be the disturbance attenuation coefficient; The mathematical expression for the geological safety state vector is: (6) In formula (6), Let be the geological safety state vector at time t; The mathematical expression for the geological safety index is: (7) In formula (7), Let be the geological safety index at time t. For time t, the first Environmental condition coefficient of individual safety index Geological safety state vector The Middle The value of the individual safety index at time t.

[0011] Preferably, the step of constructing a geological safety prediction model based on the geological safety index specifically includes: The historical monitoring feature vector is obtained by extracting time-series features from the historical monitoring dataset. With the time prediction window set to T, the mathematical expression for the historical monitoring feature vector is: (8) In formula (8), For historical monitoring feature vectors, For geological deformation data, For environmentally induced data, For earthquake activity data, For geological ontology data, Data on human activities; Mathematical expressions for gating calculations of each sub-data type in the historical monitoring dataset: (9) In formula (9), , , , , The weight matrix for gating each sub-data feature type. , , , , This is the bias vector for gating each sub-data feature type. , , , , Gating for each sub-data feature type; The data input to the geological safety prediction model is: (9) The mathematical expression for the input gate is: (10) In formula (10), , , , , For each sub-data feature type of g, the feature parameters are: The weight matrix from the input features to the input gate. The weight matrix from the hidden state to the input gate. In hidden state, This is the bias vector for the input gate. For input gates; The mathematical expression for the forgetting gate is: (11) In formula (11), For the Gate of Oblivion The weight matrix from the input features to the forget gate. This is the weight matrix from the hidden state to the forget gate. This is the bias vector for the forget gate; The mathematical expression for the output gate is: (12) In formula (12), For output gate, The weight matrix from the input features to the output gate. The weight matrix from the hidden state to the output gate. This is the bias vector for the output gate; The mathematical expression for the candidate cell state is: (13) In formula (13), Candidate cell state, This is the weight matrix for cell states; The mathematical expression for obtaining cell state using a multi-scale memory mechanism is as follows: (14) In formula (14), This is a short-term cellular state. This is the intermediate cell stage. For long-term cell state, , and These are forgetting gates for short-term, medium-term, and long-term scales, respectively. , and These are input gates for short-term, medium-term, and long-term scales, respectively. The mathematical expression for the scale fusion gating weight vector is: (15) In formula (15), For scale fusion gating weight vector, The weight matrix for scale fusion gating. The bias vector for scale fusion gating; The mathematical expression for fusing short-term, medium-term, and long-term memory is: (16) In formula (16), To achieve fusion of cell states, , and These are the short-term, medium-term, and long-term memory weight vectors, respectively. This is the long-term memory retention factor; The mathematical expression for risk assessment of geological safety is: (17) In formula (17), The geological safety risk level within the target area. As risk assessment weights, Risk assessment bias; The mathematical expression for the final hidden state, derived from the risk assessment, is as follows: (18) In formula (18), This is the final hidden state. For risk sensitivity coefficient, The fusion coefficients are the attention context vectors. The mathematical expression for the geological safety prediction index sequence is: (19) In formula (19), To output the weight matrix, For output bias, This is a sequence of geological safety prediction indices.

[0012] Preferably, the step of constructing a geological safety prediction model based on the geological safety index further includes: A geological time-series attention mechanism is introduced to adaptively assign weights to hidden states at different historical moments based on real-time monitoring data and historical trends when the geological safety prediction model makes predictions, and to generate context vectors that integrate key historical information.

[0013] Preferably, the step of inputting real-time monitoring data into the geological safety prediction model to obtain disaster risk prediction results and triggering a corresponding early warning mechanism specifically includes: Real-time monitoring data is input into the geological safety prediction model to obtain a geological safety prediction index sequence for future periods; Geological disaster risk levels are determined based on the geological safety prediction index in the geological safety prediction index sequence. When the geological safety prediction index exceeds a preset threshold, an early warning instruction corresponding to the risk level is automatically triggered, specifically including: When the geological safety prediction index is lower than or equal to the first threshold And above the second threshold At that time, a warning is triggered. When the geological safety prediction index is lower than or equal to the second threshold And above the third threshold When this happens, an alert or warning is triggered. When the geological safety prediction index is lower than or equal to the third threshold At that time, a danger warning is triggered; in, .

[0014] (III) Beneficial Effects Compared with existing technologies, this application provides a disaster prediction method based on geological safety, which has the following beneficial effects: 1. This method integrates multiple data types and proposes a geological safety index formed by the weighted fusion of multiple sub-safety indices, which realizes a comprehensive quantitative assessment of geological stability, overcomes the one-sidedness of a single data source, and makes the results of geological disaster prediction more reliable. 2. This method uses an improved LSTM to construct a geological safety prediction model, which incorporates multi-source feature gating, multi-scale memory mechanism and geological time series attention, enabling it to better capture the complex nonlinear relationships in the geological disaster system, while ensuring that the prediction results conform to physical laws, thus significantly improving the accuracy and reliability of the model. 3. The early warning trigger of this method is based on the predicted GSI sequence and its dynamic changes (such as the value falling below the threshold, the appearance of an accelerated decline inflection point, etc.), forming a dynamic and intelligent early warning mechanism, realizing a fundamental transformation from "critical alarm" to "trend early warning", and winning crucial advance preparation time for disaster avoidance and relief. Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0015] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a schematic diagram of a disaster prediction method based on geological safety proposed in this application. Detailed Implementation

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0017] The terms "first" and "second" in the specification and claims of this application may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise stated, "multiple" means two or more. Furthermore, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0018] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0019] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0020] This application provides a new technical solution: a disaster prediction method based on geological safety, comprising: Construct a multi-source heterogeneous historical monitoring dataset; The geological safety index is obtained based on the geological safety monitoring dataset. A geological safety prediction model is constructed based on the geological safety index. Real-time monitoring data is input into the geological safety prediction model to obtain disaster risk prediction results and trigger the corresponding early warning mechanism.

[0021] In this invention, the construction of a multi-source heterogeneous historical monitoring dataset specifically includes: The historical monitoring dataset includes the following sub-data feature types: geological ontology data, environmentally induced data, seismic activity data, geological deformation data, and human activity data; Obtain geological data of the target area within a period, including the properties of soil and rock, structural features, and topography. Acquire environmental induced data for the target area within a given period, including meteorological data and hydrological data; the meteorological data includes rainfall, temperature, or number of freeze-thaw cycles; the hydrological data includes changes in groundwater level or pore water pressure. Acquire seismic activity data for the target region within a period, including microseismic frequency, magnitude, or focal depth; Acquire geological deformation data of the target area within a period, including surface displacement, deep displacement, or building deformation; Acquire human activity data for the target area within a period, including excavation disturbance or drainage activities; The geological data, environmentally induced data, seismic activity data, geological deformation data, and human activity data are cleaned, time-aligned, and standardized to obtain the historical monitoring dataset.

[0022] In this invention, obtaining the geological safety index based on the geological safety monitoring dataset specifically includes: The geological safety index includes deformation stability index, hydraulic stability index, mechanical stability index, structural stability index, and activity disturbance index; The deformation stability values ​​are obtained based on geological deformation data, and the mathematical expression for the deformation stability index is: (1) In formula (1), The deformation index is... , , , The attenuation coefficient is... For displacement acceleration, This is the cumulative displacement. The historical maximum safe displacement threshold within the target area. The coefficient for the trend term; The hydraulic stability index is obtained based on environmentally induced data, and its mathematical expression is as follows: (2) In formula (2), The hydraulic stability index, The hydrological sensitivity coefficient for the target area. This represents the effective rainfall over the past 7 days. The critical effective rainfall amount, For real-time groundwater levels, This is the critical groundwater level. The threshold for normal water level fluctuations used for normalization. and These are the weighting coefficients. This refers to the number of freeze-thaw cycles. To measure the pore water pressure, Displacement reference hole pressure; The mechanical stability index is obtained based on seismic activity data, and its mathematical expression is as follows: (3) In formula (3), This is the mechanical stability index. To accumulate the energy release from microseismic events, the energy of all microseismic events during the monitoring period is summed and calculated. This is the critical cumulative energy release threshold. This is the amplification factor, used to adjust the degree of nonlinear response.

[0023] The structural stability index is obtained based on geological ontology data, and its mathematical expression is as follows: (4) In formula (4), It is a structural stability index. , and These are the weighting coefficients. For rock mass quality rating, This is the value after normalizing the terrain slope. To score the relationship between the structural plane and the slope aspect combination. , , ; The activity interference index is obtained based on human activity data, and its mathematical expression is as follows: (5) In formula (5), The activity interference index, To determine the intensity of excavation disturbance. For drainage volume, and Let be the disturbance attenuation coefficient; The mathematical expression for the geological safety state vector is: (6) In formula (6), Let be the geological safety state vector at time t; The mathematical expression for the geological safety index is: (7) In formula (7), Let be the geological safety index at time t. For time t, the first Environmental condition coefficient of individual safety index Geological safety state vector The Middle The value of the individual safety index at time t.

[0024] In this invention, the step of constructing a geological safety prediction model based on the geological safety index specifically includes: The historical monitoring feature vector is obtained by extracting time-series features from the historical monitoring dataset. With the time prediction window set to T, the mathematical expression for the historical monitoring feature vector is: (8) In formula (8), For historical monitoring feature vectors, For geological deformation data, For environmentally induced data, For earthquake activity data, For geological ontology data, Data on human activities; Mathematical expressions for gating calculations of each sub-data type in the historical monitoring dataset: (9) In formula (9), , , , , The weight matrix for gating each sub-data feature type. , , , , This is the bias vector for gating each sub-data feature type. , , , , Gating for each sub-data feature type; The data input to the geological safety prediction model is: (9) The mathematical expression for the input gate is: (10) In formula (10), , , , , For each sub-data feature type of g, the feature parameters are: The weight matrix from the input features to the input gate. The weight matrix from the hidden state to the input gate. In hidden state, This is the bias vector for the input gate. For input gates; The mathematical expression for the forgetting gate is: (11) In formula (11), For the Gate of Oblivion The weight matrix from the input features to the forget gate. This is the weight matrix from the hidden state to the forget gate. This is the bias vector for the forget gate; The mathematical expression for the output gate is: (12) In formula (12), For output gate, The weight matrix from the input features to the output gate. The weight matrix from the hidden state to the output gate. This is the bias vector for the output gate; The mathematical expression for the candidate cell state is: (13) In formula (13), Candidate cell state, This is the weight matrix for cell states; The mathematical expression for obtaining cell state using a multi-scale memory mechanism is as follows: (14) In formula (14), This is a short-term cellular state. This is the intermediate cell stage. For long-term cell state, , and These are forgetting gates for short-term, medium-term, and long-term scales, respectively. , and These are input gates for short-term, medium-term, and long-term scales, respectively. The mathematical expression for the scale fusion gating weight vector is: (15) In formula (15), For scale fusion gating weight vector, The weight matrix for scale fusion gating. The bias vector for scale fusion gating; The mathematical expression for fusing short-term, medium-term, and long-term memory is: (16) In formula (16), To achieve fusion of cell states, , and These are the short-term, medium-term, and long-term memory weight vectors, respectively. This is the long-term memory retention factor; The mathematical expression for risk assessment of geological safety is: (17) In formula (17), The geological safety risk level within the target area. As risk assessment weights, Risk assessment bias; The mathematical expression for the final hidden state, derived from the risk assessment, is as follows: (18) In formula (18), This is the final hidden state. For risk sensitivity coefficient, The fusion coefficients are the attention context vectors. The mathematical expression for the geological safety prediction index sequence is: (19) In formula (19), To output the weight matrix, For output bias, This is a sequence of geological safety prediction indices.

[0025] In this invention, the step of constructing a geological safety prediction model based on the geological safety index further includes: A geological time-series attention mechanism is introduced to adaptively assign weights to hidden states at different historical moments based on real-time monitoring data and historical trends when the geological safety prediction model makes predictions, and to generate context vectors that integrate key historical information.

[0026] In a specific embodiment, the steps for introducing the geological time-series attention mechanism are as follows: Based on the geological time series correlation, the attention weight of the hidden state at each historical moment to the current prediction is calculated; Based on the attention weights, the historical hidden states are weighted and fused to generate a context vector with key historical information; The context vector is fused with the hidden state obtained from the current cell state to output a hidden state that incorporates historical experience, which is used to predict the geological safety index.

[0027] In this invention, the step of inputting real-time monitoring data into the geological safety prediction model to obtain disaster risk prediction results and triggering a corresponding early warning mechanism specifically includes: Real-time monitoring data is input into the geological safety prediction model to obtain a geological safety prediction index sequence for future periods; Geological disaster risk levels are determined based on the geological safety prediction index in the geological safety prediction index sequence. When the geological safety prediction index exceeds a preset threshold, an early warning instruction corresponding to the risk level is automatically triggered, specifically including: When the geological safety prediction index is lower than or equal to the first threshold And above the second threshold At that time, a warning is triggered. When the geological safety prediction index is lower than or equal to the second threshold And above the third threshold When this happens, an alert or warning is triggered. When the geological safety prediction index is lower than or equal to the third threshold At that time, a danger warning is triggered; in, .

[0028] In a specific embodiment, real-time monitoring data within the target area is collected and input into a trained geological safety prediction model. The geological safety prediction model outputs a geological safety prediction index sequence for a future period of time. The duration of the future period can be 24 hours, 72 hours, one week, or one month, and staff can also customize the prediction duration.

[0029] The rules for determining the risk level are as follows: Risk Level 1: Geological safety prediction index exceeds the first threshold. And there is no significant downward trend; Risk Level 2: Geological safety prediction index is lower than or equal to the first threshold. And above the second threshold Or, the geological safety prediction index sequence shows a sustained and slow decline; Risk Level 3: Geological safety prediction index is lower than or equal to the second threshold. And above the third threshold Or the geological safety prediction index sequence shows an inflection point of accelerated decline; Risk Level 4: Geological safety prediction index is lower than or equal to the third threshold. It can be predicted that the geological safety prediction index will approach 0 in the next few hours.

[0030] Based on the risk level, a visual report containing the forecast results, risk areas, and recommended measures is generated on the early warning information platform and sent to relevant responsible persons via SMS, App push, and other means.

[0031] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0032] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A geological safety-based disaster prediction method, characterized by, The method comprises the following steps: constructing a multi-source heterogeneous historical monitoring data set; obtaining a geological safety index based on the geological safety monitoring data set; constructing a geological safety prediction model according to the geological safety index; inputting real-time monitoring data into the geological safety prediction model to obtain a disaster risk prediction result, and triggering a corresponding early warning mechanism.

2. The geological security based disaster prediction method as claimed in claim 1, wherein, The construction of the multi-source heterogeneous historical monitoring data set specifically comprises: The sub-data feature types included in the historical monitoring data set are geological ontology data, environment-induced data, seismic activity data, geological deformation data and human activity data; obtaining geological ontology data of a target region within a period, wherein the geological ontology data comprises rock mass properties, structural surface characteristics and topography and geomorphology; obtaining environment-induced data of the target region within the period, wherein the environment-induced data comprises meteorological data and hydrological data; the meteorological data comprises rainfall, air temperature or freeze-thaw cycle times; and the hydrological data comprises groundwater level changes or pore water pressure; obtaining seismic activity data of the target region within the period, wherein the seismic activity data comprises microseismic frequency, magnitude or focal depth; obtaining geological deformation data of the target region within the period, wherein the geological deformation data comprises surface displacement, deep displacement or building deformation; obtaining human activity data of the target region within the period, wherein the human activity data comprises excavation disturbance or water pumping behavior; performing data cleaning, time alignment and standardization on the geological ontology data, environment-induced data, seismic activity data, geological deformation data and human activity data to obtain the historical monitoring data set.

3. The method of claim 2, wherein, The geological safety index comprises a deformation stability index, a hydraulic stability index, a mechanical stability index, a structural stability index and an activity interference index. The deformation stability index is obtained based on the geological deformation data, and a mathematical expression of the deformation stability index is: The hydraulic stability index is obtained based on the environment-induced data, and a mathematical expression of the hydraulic stability index is: （1） In formula (1), is a deformation index, , , , is a decay coefficient, is a displacement acceleration, is a cumulative displacement amount, is a historical maximum safety displacement threshold in a target area, is a trend item coefficient; The mechanical stability index is obtained based on the seismic activity data, and a mathematical expression of the mechanical stability index is: （2） In formula (2), is the hydraulic stability index, is the target area hydrological sensitivity coefficient, is the effective rainfall in 7 days, is the critical effective rainfall, is the real-time groundwater level, is the critical groundwater level, is the water level normal fluctuation threshold for normalization, and is the weight coefficient, is the number of freeze-thaw cycles, is the measured pore water pressure, displacement reference pore pressure; The structural stability index is obtained based on the geological ontology data, and a mathematical expression of the structural stability index is: （3） In formula (3), is the mechanical stability index, is the cumulative microseismic energy release, which is calculated by summing the energy of all microseismic events in the monitoring period, is the critical cumulative energy release threshold, is the amplification coefficient, which is used to adjust the degree of non-linear response. The activity interference index is obtained based on the human activity data, and a mathematical expression of the activity interference index is: （4） In formula (4), is the structure stability index, , and is the weight coefficient, is the rock mass quality rating, is the normalized value of the terrain slope, is the score for the combination of the structure surface and the slope direction, , , ; A mathematical expression of a geological safety state vector is: （5） In equation (5), is the activity interference index, is the excavation disturbance intensity, is the displacement volume, and is the disturbance attenuation coefficient; A mathematical expression of the geological safety index is: （6） In equation (6), is the geological safety state vector at time t. The geological safety prediction model is constructed according to the geological safety index, and specifically comprises: （7） In formula (7), Let be the geological safety index at time t. For time t, the first Environmental condition coefficient of individual safety index Geological safety state vector The Middle The value of the individual safety index at time t.

5. The method of claim 3, wherein the method further comprises: performing time series feature extraction on the historical monitoring data set to obtain a historical monitoring feature vector, setting a time prediction window as T, and a mathematical expression of the historical monitoring feature vector is: a mathematical expression of sub-data type gating calculation in the historical monitoring data set is: （8） In equation (8), is a historical monitoring feature vector, is a geological deformation data, is an environmental induced data, is a seismic activity data, is a geological ontology data, is a human activity data; data input into the geological safety prediction model comprises: （9） In Equation (9), , , , , is a weight matrix for each sub-data feature type gating, , , , , is a bias vector for each sub-data feature type gating, , , , , is each sub-data feature type gating; a mathematical expression of an input gate is: （9） a mathematical expression of a forget gate is: （10） In Equation (10), , , , , is a feature parameter of each sub-data feature type, is a weight matrix of the input feature to the input gate, is a weight matrix of the hidden state to the input gate, is a hidden state, is a bias vector of the input gate, is an input gate; a mathematical expression of an output gate is: （11） In Equation (11), is a forget gate, is a weight matrix of input features to the forget gate, is a weight matrix of hidden states to the forget gate, is a bias vector of the forget gate; a mathematical expression of a candidate cell state is: （12） In equation (12), is an output gate, is a weight matrix of input features to the output gate, is a weight matrix of hidden states to the output gate, is a bias vector of the output gate; a mathematical expression of a cell state obtained by using a multi-scale memory mechanism is: （13） In equation (13), is a candidate cell state, is a weight matrix for cell states; a mathematical expression of a scale fusion gating weight vector is: （14） In equation (14), is the short-term cell state, is the medium-term cell state, is the long-term cell state, , and are the short-term, medium-term, and long-term scale forget gates, respectively, , and are the short-term, medium-term, and long-term scale input gates, respectively; ​ （15） In equation (15), is a scale fusion gating weight vector, is a weight matrix of the scale fusion gating, is a bias vector of the scale fusion gating; The mathematical expression for fusing short-term memory, medium-term memory and long-term memory is: （16） In equation (16), is the fusion cell state, , and are short-term, medium-term, and long-term memory weight vectors, respectively, is the long-term memory retention coefficient; The mathematical expression for risk assessment of geological safety is: （17） In equation (17), is a geological safety risk level within the target area, is a risk assessment weight, is a risk assessment bias; The mathematical expression for obtaining the final hidden state according to risk assessment is: （18） In equation (18), is the final hidden state, is the risk sensitivity coefficient, is the fusion coefficient of the attention context vector; The mathematical expression for the geological safety prediction index sequence is: （19） In equation (19), is the output weight matrix, is the output bias, is the geologic safety prediction index sequence.

6. The geological security based disaster prediction method of claim 1, wherein, The geological safety prediction model is constructed according to the geological safety index, and further includes: The geological time sequence attention mechanism is introduced, which is used to assign weights to the hidden states of different historical moments based on real-time monitoring data and historical change trends when the geological safety prediction model is predicting, and generate a context vector that fuses key historical information.

7. The method of claim 5, wherein the method further comprises: The real-time monitoring data is input into the geological safety prediction model to obtain a disaster risk prediction result, and a corresponding warning mechanism is triggered, which specifically includes: The real-time monitoring data is input into the geological safety prediction model to obtain a future period geological safety prediction index sequence; Based on the geological safety prediction index in the geological safety prediction index sequence, the geological disaster risk level is determined; When the geological safety prediction index exceeds a preset threshold, a warning instruction corresponding to the risk level is automatically triggered, which specifically includes: when the geologic safety prediction index is lower than or equal to a first threshold value and higher than a second threshold value a concern alert is triggered; when the geologic safety prediction index is lower than or equal to a second threshold value and higher than a third threshold value an alert warning is triggered; when the geologic safety prediction index is less than or equal to a third threshold value a danger warning is triggered; wherein .

Images