Geological disaster dynamic comprehensive early warning method and system based on three-dimensional monitoring

Abstract

The invention relates to the technical field of intelligent early warning, in particular to a geological disaster dynamic comprehensive early warning method and system based on three-dimensional monitoring, and the method comprises the steps: obtaining a multi-dimensional three-dimensional monitoring data flow of a geological disaster hidden danger area in real time, and carrying out the time sequence alignment and space registration of the multi-dimensional three-dimensional monitoring data flow, obtaining a multi-source fusion data set; identifying key coupling characteristic factors in the multi-source fusion data set; outputting a three-dimensional dynamic risk field model of the geological disaster hidden danger area in a risk probability cloud picture form in a dynamic response process under a multi-field coupling effect; extracting dynamic feature vectors of risk evolution rate and risk space aggregation degree in the three-dimensional dynamic risk field model; performing early warning decision on the geological disaster hidden danger area to obtain an early warning level instruction; parameters in the dynamic response process are reversely optimized, and the optimized parameters are used for construction and updating of the three-dimensional dynamic risk field model of the next round; according to the invention, the efficiency of dynamic comprehensive early warning of geological disasters can be improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent early warning technology, and in particular to a dynamic comprehensive early warning method and system for geological disasters based on three-dimensional monitoring. Background Technology

[0002] Existing geological disaster early warning technologies have significant limitations in data acquisition and processing. They lack the ability to conduct three-dimensional monitoring of potential geological disaster areas, relying heavily on single-dimensional or limited types of monitoring data. Furthermore, they fail to effectively align and spatially register multi-source heterogeneous data. Monitoring data from different sources exhibits temporal asynchrony and inconsistent spatial benchmarks, resulting in low-quality data fusion and an inability to accurately capture the coordinated changes between the surface and interior of geological bodies. Consequently, this hinders a comprehensive perception and assessment of the disaster gestation process.

[0003] Meanwhile, existing technologies fail to accurately identify key coupling factors in the evolution of geological hazards, and lack specificity and accuracy in simulating the dynamic response of geological bodies under multi-field coupling. Risk field models are mostly static or semi-static, making it difficult to reflect the real-time evolution of risks. Furthermore, existing early warning methods lack dynamic parameter optimization mechanisms; model parameters remain fixed and cannot be adaptively adjusted based on early warning results and changes in risk field morphology. This leads to discrepancies between early warning levels and actual risk evolution stages, resulting in insufficient timeliness and reliability of early warnings, failing to meet the actual needs for precise geological hazard early warning. Summary of the Invention

[0004] This invention provides a dynamic comprehensive early warning method and system for geological disasters based on three-dimensional monitoring, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, this invention provides a dynamic comprehensive early warning method for geological disasters based on three-dimensional monitoring, comprising: S1. Deploy and activate the multi-source sensor network deployed in the geological hazard hazard area, acquire the multi-dimensional three-dimensional monitoring data stream of the geological hazard hazard area in real time, and perform time-series alignment and spatial registration on the multi-dimensional three-dimensional monitoring data stream to obtain the multi-source fusion dataset of the geological hazard hazard area. S2. Based on a dynamic Bayesian network, the causal relationship of the geological disaster formation process is mined and the dynamic contribution weight is quantified in the multi-source fusion dataset to obtain the key coupling feature factors in the multi-source fusion dataset. S3. Simulate the dynamic response process of the geological body under multi-field coupling based on the key coupling characteristic factors, and output the three-dimensional dynamic risk field model of the geological hazard area in the form of a risk probability cloud map. S4. Extract the dynamic feature vectors of risk evolution rate and risk spatial clustering degree in the three-dimensional dynamic risk field model; S5. Make an early warning decision on the geological hazard area based on the dynamic feature vector, and obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard area; S6. Based on the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, optimize the parameters in the dynamic response process in reverse, and use the optimized parameters for the construction and updating of the three-dimensional dynamic risk field model in the next round.

[0006] In a preferred embodiment, the deployment and activation of a multi-source sensor network in a geological hazard-prone area allows for the real-time acquisition of multi-dimensional three-dimensional monitoring data streams from the geological hazard-prone area. The multi-dimensional three-dimensional monitoring data streams are then time-series aligned and spatially registered to obtain a multi-source fusion dataset of the geological hazard-prone area, including: Deploy and activate sensing devices located on the surface and in boreholes within geological hazard zones to form a multi-source sensing device network in the geological hazard zone. Receive surface displacement data stream, soil moisture content data stream, underground rock and soil stress data stream, and microseismic signal data stream from the multi-source sensing device network, add a timestamp to each data stream, and obtain the multi-source asynchronous data stream of the geological hazard area; Based on the timestamp, the multi-source asynchronous data stream is interpolated and synchronized to obtain a synchronized monitoring data stream with a completely aligned time series. Based on a preset geographic information system coordinate reference, each monitoring data point in the synchronized monitoring data stream is mapped to a unified three-dimensional spatial coordinate system to obtain the registration data stream of the geological hazard area; The registered data stream is restructured according to the time-space dimension to obtain a multi-source fusion dataset of the geological hazard hazard area.

[0007] In a preferred embodiment, the step of mining causal relationships and quantifying dynamic contribution weights of the geological disaster gestation process in the multi-source fusion dataset based on a dynamic Bayesian network to obtain key coupling feature factors in the multi-source fusion dataset includes: The multi-source fusion dataset is subjected to anomaly cleaning and standardization normalization to obtain multi-source monitoring time-series data of the geological hazard hazard area; From the multi-source monitoring time series data, the correlation between the rate of change of surface displacement and the trend of stress change in underground rock and soil is extracted to obtain the displacement-stress covariance characteristics of the geological hazard zone. From the multi-source monitoring time-series data, the correlation between the spatiotemporal variation of soil moisture content and the occurrence sequence of microseismic signal events is extracted to obtain the seepage-microseismic correlation characteristics of the geological hazard area. Based on the prior knowledge of the pre-set geological hazard mechanics model, the contribution of the displacement-stress covariance characteristics and the seepage-microseismic correlation characteristics to the geological hazard incubation stage is evaluated. Based on the contribution, features with a contribution higher than the set standard are selected from the displacement-stress covariance features and the seepage-microseismic correlation features to obtain the key coupling feature factors of the geological hazard zone.

[0008] In a preferred embodiment, the contribution of the displacement-stress covariance characteristic is calculated using the following formula: ; In the formula, The contribution of the displacement-stress covariance characteristic. The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, Let be the cross-adjustment coefficient, and let be , be , be ; The contribution formula for the seepage-microseismic correlation feature is as follows: ; In the formula, The contribution of the seepage-microseismic correlation characteristics, The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, This is the cross-adjustment coefficient.

[0009] In a preferred embodiment, the step of simulating the dynamic response process of a geological body under multi-field coupling based on the key coupling characteristic factors includes: The constitutive relation set of the interaction between stress field, seepage field and deformation field is encapsulated into a coupled analysis engine; The key coupling feature factors are configured as boundary conditions and driving parameters into the corresponding input interface of the coupling analysis engine; In the coupling analysis engine, the time step and total simulation duration are set for simulating the dynamic response process; The coupling analysis engine is driven to run iteratively according to the time step within the total simulation duration; After the coupled analysis engine completes its iterative run, it extracts and outputs the simulation state parameters updated by each iteration step, thus obtaining the spatiotemporal evolution sequence of the geological body.

[0010] In a preferred embodiment, driving the coupling analysis engine to run iteratively according to the time step within the total simulation duration includes: In the coupling analysis engine, the initial state vector of the geological body is initialized based on the key coupling feature factors and preset initial geological parameters; Starting with the initial state vector, an iterative loop is initiated for the total simulation duration; Based on the key coupling feature factor corresponding to the current iteration step and the current geological body state vector, the constitutive relation set is invoked to generate the change in the geological body state under the current multi-field coupling effect; Based on the change, update the current geological body state vector to obtain the updated geological body state vector for the next iteration. The updated geological body state vector is used as the new current geological body state vector for the next iteration cycle; When the iteration loop reaches the number of iteration steps corresponding to the total simulation duration, the loop terminates and all the geological body state vectors generated in all iteration steps are output.

[0011] In a preferred embodiment, outputting the three-dimensional dynamic risk field model of the geological hazard hazard area in the form of a risk probability cloud map includes: Assign state parameter values ​​corresponding to the current simulation moment in the spatiotemporal evolution sequence to each grid cell in the three-dimensional spatial grid model of the geological hazard potential area; Based on preset instability criteria and risk assessment rules, the probability value of instability and failure of the grid cell in the next simulation period is predicted; The risk probability values ​​are smoothed to obtain the risk probability field data of the geological hazard hazard area; By using different colors and transparency to characterize the different risk probability levels and spatial distributions in the risk probability field data, a three-dimensional dynamic risk field model of the geological hazard hazard area is obtained.

[0012] In a preferred embodiment, extracting the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model includes: Extract the risk probability time series of each spatial unit within a preset time window from the three-dimensional dynamic risk field model; The rate of change of risk probability of each spatial unit in the risk probability time series is used as a quantitative indicator of the risk evolution rate in the geological hazard hazard area. Identify the spatial clustering index of high-risk probability regions in the three-dimensional dynamic risk field model; By combining the quantitative index of risk evolution rate with the index of spatial clustering, a dynamic feature vector of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model is obtained.

[0013] In a preferred embodiment, the step of back-optimizing the parameters in the dynamic response process based on the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model includes: Based on the level of the early warning instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, an optimization target for improving the accuracy of the next simulation cycle is generated. Based on the optimization objective, a set of parameters to be adjusted associated with the optimization objective is identified from the constitutive relation set of the coupling analysis engine; The set of parameters to be adjusted is fine-tuned according to a preset adjustment strategy to generate optimized parameters.

[0014] To address the aforementioned problems, the present invention also provides a dynamic integrated early warning system for geological disasters based on three-dimensional monitoring, the system comprising: The multi-source data acquisition and fusion module is used to deploy and activate a network of multi-source sensing devices deployed in geological hazard hazard areas, acquire multi-dimensional three-dimensional monitoring data streams of the geological hazard hazard areas in real time, and perform temporal alignment and spatial registration on the multi-dimensional three-dimensional monitoring data streams to obtain multi-source fusion datasets of the geological hazard hazard areas. The key feature factor identification module is used to mine the causal relationship of the geological disaster gestation process and quantify the dynamic contribution weight based on the dynamic Bayesian network to obtain the key coupling feature factors in the multi-source fusion dataset. The dynamic risk field simulation and visualization module is used to simulate the dynamic response process of geological bodies under multi-field coupling based on the key coupling characteristic factors, and output the three-dimensional dynamic risk field model of the geological hazard area in the form of a risk probability cloud map. The risk evolution feature extraction module is used to extract the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model. The intelligent early warning decision module is used to make early warning decisions for the geological hazard hazard area based on the dynamic feature vector, and obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard hazard area; The model parameter adaptive optimization module is used to optimize the parameters in the dynamic response process in reverse according to the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, and use the optimized parameters for the construction and update of the three-dimensional dynamic risk field model in the next round.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention deploys a network of multi-source sensors covering the surface and interior of geological hazard-prone areas to collect multi-dimensional monitoring data in real time. Through temporal alignment, spatial registration, and structured recombination, a high-quality multi-source fusion dataset is formed, enabling comprehensive perception and precise capture of the geological body's state. Simultaneously, through anomaly data cleaning, standardization, and contribution quantification analysis, key coupling characteristic factors are accurately identified, providing precise and reliable core inputs for subsequent dynamic response simulation of geological hazards, effectively improving the scientific rigor and relevance of hazard evolution characteristic analysis.

[0016] 2. This invention utilizes a three-dimensional dynamic risk field model constructed based on key coupled feature factors. It visually presents the spatiotemporal distribution and evolution of risks in the form of a risk probability cloud map. Combining dynamic feature vectors of risk evolution rate and spatial clustering, it enables early warning decision-making, outputting early warning level instructions highly matched to the current risk evolution stage, significantly improving the accuracy and timeliness of early warning decisions. Furthermore, by back-optimizing parameters in the dynamic response process and using them for the next round of model updates, it achieves adaptive iterative upgrades of the early warning model, continuously enhancing the stability and applicability of the early warning method, and providing efficient technical support for geological disaster prevention and control. Attached Figure Description

[0017] Figure 1 A flowchart illustrating a dynamic integrated early warning method for geological disasters based on three-dimensional monitoring, provided in an embodiment of the present invention; Figure 2 A functional module diagram of a dynamic integrated early warning system for geological disasters based on three-dimensional monitoring, provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a dynamic comprehensive early warning method for geological disasters based on three-dimensional monitoring, according to an embodiment of the present invention. In this embodiment, the dynamic comprehensive early warning method for geological disasters based on three-dimensional monitoring includes: S1. Deploy and activate the multi-source sensor network deployed in the geological hazard hazard area, acquire the multi-dimensional three-dimensional monitoring data stream of the geological hazard hazard area in real time, and perform time-series alignment and spatial registration on the multi-dimensional three-dimensional monitoring data stream to obtain the multi-source fusion dataset of the geological hazard hazard area. In this embodiment of the invention, the deployment and activation of the multi-source sensing device network in the geological hazard hazard area, the real-time acquisition of multi-dimensional three-dimensional monitoring data streams of the geological hazard hazard area, and the temporal alignment and spatial registration of the multi-dimensional three-dimensional monitoring data streams to obtain a multi-source fusion dataset of the geological hazard hazard area, including: Deploy and activate sensing devices located on the surface and in boreholes within geological hazard zones to form a multi-source sensing device network in the geological hazard zone. Receive surface displacement data stream, soil moisture content data stream, underground rock and soil stress data stream, and microseismic signal data stream from the multi-source sensing device network, add a timestamp to each data stream, and obtain the multi-source asynchronous data stream of the geological hazard area; Based on the timestamp, the multi-source asynchronous data stream is interpolated and synchronized to obtain a synchronized monitoring data stream with a completely aligned time series. Based on a preset geographic information system coordinate reference, each monitoring data point in the synchronized monitoring data stream is mapped to a unified three-dimensional spatial coordinate system to obtain the registration data stream of the geological hazard area; The registered data stream is restructured according to the time-space dimension to obtain a multi-source fusion dataset of the geological hazard hazard area.

[0021] Sensing devices are deployed in key surface areas and boreholes within the geological hazard zone. Surface-mounted devices must be evenly distributed according to the terrain features and monitoring needs of the hazard zone to ensure coverage of the core monitoring areas. Borehole-mounted devices must be deployed in layers according to different depths determined by geological surveys, with each layer corresponding to a different depth within the underground rock and soil. The deployed devices include displacement sensors for collecting surface displacement data, moisture sensors for collecting soil moisture data, stress sensors for collecting underground rock and soil stress data, and microseismic sensors for collecting microseismic signal data. After deployment, the power supply system and communication module of each device are checked for proper functioning. Each device is initialized and calibrated to ensure its acquisition accuracy meets testing standards. By starting the acquisition program and verifying the data transmission link's smoothness, all sensors form a collaborative multi-source sensor network, enabling comprehensive acquisition of multi-dimensional data from the surface and underground of the geological hazard zone.

[0022] Through the communication modules of each sensor in the multi-source sensor network, the system receives real-time data streams from displacement sensors (surface displacement), moisture sensors (soil moisture content), stress sensors (underground soil and rock stress), and microseismic sensors (microseismic signal). Each sensor automatically records the precise time of data acquisition upon receiving the raw data, using its built-in clock. When acquiring each data stream, the receiving end converts the acquisition time of each raw data point into a standard timestamp using a unified UTC time format, accurate to the millisecond level, ensuring the consistency and accuracy of the time information. The converted standard timestamp is then bound to the corresponding raw data entry, giving each data stream a unique and standardized time identifier. This results in a multi-source asynchronous data stream containing four types of data: surface displacement, soil moisture content, underground soil and rock stress, and microseismic signals, with each data point appended with a standard timestamp.

[0023] First, the acquisition frequency of various data streams in the multi-source asynchronous data stream is analyzed to determine the highest acquisition frequency among all data streams. The time interval corresponding to this acquisition frequency is then used as the unified target time interval to construct a complete time series axis. This time series axis covers all target time points from the start of data acquisition to the current moment, ensuring that each target time point is evenly distributed on the time series axis. Then, each type of data in the multi-source asynchronous data stream is traversed, and all data points with timestamps in each type are extracted. These data points are then mapped onto the constructed time series axis according to their timestamp order. For a given target time point, if a certain type of data stream has a direct... The collected data is directly retained as valid data for that time point. If no data is directly collected for that time point, two adjacent valid data points are found before and after the target time point. Based on the magnitude and time interval of these two data points, the missing data value corresponding to the target time point is determined by linear estimation. That is, assuming that the data between two adjacent valid data points shows a uniform trend, the estimated data for the target time point is calculated according to the time proportion. This operation is repeated until all target time points of each type of data stream have corresponding valid data on the time series axis, and finally a synchronized monitoring data stream with complete time series alignment is obtained.

[0024] The National Geodetic Coordinate System 2000 was pre-determined as the coordinate benchmark for the geographic information system. This benchmark has unified ellipsoidal parameters and projection rules, ensuring the accuracy and consistency of spatial coordinates. During the sensor deployment phase, the latitude, longitude, and elevation data of all surface sensors were acquired using GPS positioning technology. For sensors inside boreholes, the three-dimensional spatial coordinates of the sensors were obtained through geometric calculations based on the borehole's latitude, longitude, borehole inclination, and the depth of the sensors within the borehole. These coordinate data were stored as the original location information for each sensor. When processing the synchronized monitoring data stream, for each monitoring data point, the corresponding original location information of the sensor was extracted. According to the projection transformation rules of the National Geodetic Coordinate System 2000, the original location information was converted into standard three-dimensional coordinates under this coordinate system. The converted standard three-dimensional coordinates were then bound to the corresponding monitoring data points, giving each monitoring data point a clear spatial location identifier. All monitoring data points were orderly distributed in a unified three-dimensional spatial coordinate system, ultimately resulting in the registration data stream for the geological hazard area.

[0025] Using time as the vertical organizational basis, all data points in the registered data stream are sorted according to the order of timestamps. Monitoring data from all spatial locations corresponding to the same timestamp are grouped into a time slice. Each time slice contains various monitoring data from all monitoring points in the hazard area at that time, ensuring the continuity and integrity of data in the time dimension. Using space as the horizontal organizational basis, monitoring data from the same spatial location at different time points are linked together to form independent time series data for each spatial location, clearly showing the changing trends of various monitoring indicators at that location. Subsequently, the time slices and spatial series data are correlated and integrated to construct a structured data framework with three dimensions: time, space, and monitoring indicators. The time dimension covers all target time points, the spatial dimension covers the three-dimensional coordinate positions of all sensing devices, and the monitoring indicator dimension covers four types of data: surface displacement, soil moisture content, underground rock and soil stress, and microseismic signals. Each data unit corresponds to a unique time, spatial location, and monitoring indicator in the framework, achieving the organic integration of various types of data. Finally, a multi-source fusion dataset of geological hazard areas with a standardized structure, complete dimensions, and close data correlation is obtained.

[0026] The beneficial effects include: by deploying multiple types of sensors in key surface areas and underground boreholes within geological hazard-prone zones and constructing a collaborative multi-source sensor network, comprehensive coverage of surface displacement, soil moisture content, underground rock and soil stress, and microseismic signals is achieved. This ensures that the monitoring data fully reflects the geological environment of the surface and underground areas of the hazard zone, avoiding information gaps caused by single monitoring dimensions. By attaching standardized millisecond-level timestamps to various raw data streams and performing interpolation synchronization processing, the originally asynchronously acquired multi-source data achieves complete time series alignment, eliminating time-series deviations caused by differences in acquisition frequencies of different devices. This ensures the consistency and comparability of various monitoring data in the time dimension, laying a temporal foundation for subsequent multi-dimensional data correlation analysis. Based on a unified geographic information system coordinate benchmark, spatial registration of monitoring data points is completed, mapping all monitoring data to the same three-dimensional spatial coordinate system. This clarifies the precise spatial location corresponding to each monitoring data item, solving the problem of scattered location information from different devices and the inability to accurately correlate them, thus achieving standardized spatial positioning of the data.

[0027] By reorganizing data in a structured manner across time and space to form a multi-source fusion dataset, the data becomes continuous and orderly in the time dimension, clearly located in the spatial dimension, and clearly classified in the indicator dimension. This constructs a three-dimensional data framework that links time, space, and monitoring indicators, allowing various types of monitoring data to be organically integrated and closely related. This not only improves the usability and standardization of the data, but also provides comprehensive, accurate, and structured high-quality data support for the early identification, dynamic tracking, and risk assessment of geological disaster hazards, effectively assisting in the precise implementation of geological disaster early warning work.

[0028] S2. Based on a dynamic Bayesian network, the causal relationship of the geological disaster formation process is mined and the dynamic contribution weight is quantified in the multi-source fusion dataset to obtain the key coupling feature factors in the multi-source fusion dataset. In this embodiment of the invention, the step of mining the causal association of the geological disaster gestation process and quantifying the dynamic contribution weights of the multi-source fusion dataset based on a dynamic Bayesian network to obtain key coupling feature factors in the multi-source fusion dataset includes: The multi-source fusion dataset is subjected to anomaly cleaning and standardization normalization to obtain multi-source monitoring time-series data of the geological hazard hazard area; From the multi-source monitoring time series data, the correlation between the rate of change of surface displacement and the trend of stress change in underground rock and soil is extracted to obtain the displacement-stress covariance characteristics of the geological hazard zone. From the multi-source monitoring time-series data, the correlation between the spatiotemporal variation of soil moisture content and the occurrence sequence of microseismic signal events is extracted to obtain the seepage-microseismic correlation characteristics of the geological hazard area. Based on the prior knowledge of the pre-set geological hazard mechanics model, the contribution of the displacement-stress covariance characteristics and the seepage-microseismic correlation characteristics to the geological hazard incubation stage is evaluated. Based on the contribution, features with a contribution higher than the set standard are selected from the displacement-stress covariance features and the seepage-microseismic correlation features to obtain the key coupling feature factors of the geological hazard zone.

[0029] The contribution of the displacement-stress covariance characteristic is calculated using the following formula: ; In the formula, The contribution of the displacement-stress covariance characteristic. The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, Let be the cross-adjustment coefficient, and let be , be , be ; The contribution formula for the seepage-microseismic correlation feature is as follows: ; In the formula, The contribution of the seepage-microseismic correlation characteristics, The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, This is the cross-adjustment coefficient.

[0030] The source of the self-variation sensitivity is multi-source monitoring time-series data and the displacement-stress covariance features and seepage-microseismic correlation features extracted from them. The acquisition process involves extracting complete time-series data sequences of displacement-stress covariance features and seepage-microseismic correlation features from the multi-source monitoring time-series data. The time-series data sequences of these two features are then normalized. During the processing, the maximum and minimum values ​​in the time-series data sequences of each feature are first determined. The minimum value of the corresponding sequence is subtracted from the value of each data point in the sequence. The difference is then divided by the difference between the maximum and minimum values ​​of the corresponding sequence. Through this process, the time-series data of both features are transformed to the same numerical range. The normalized data corresponding to the displacement-stress covariance features and the normalized data corresponding to the seepage-microseismic correlation features are the self-variation sensitivity.

[0031] The dynamic coupling degree is derived from time-series data of displacement-stress covariance characteristics and seepage-microseismic correlation characteristics. The acquisition process involves aligning the time-series data of the two characteristics along the time axis, comparing and analyzing the direction and magnitude of change of the two characteristics at each time point, counting the number of times and duration of coordinated changes of the two characteristics rising and falling synchronously within the same time interval, calculating the proportion of coordinated change duration to the total analysis time, and quantifying the ratio of the change magnitude of the two characteristics in the coordinated change stage. The number of coordinated changes, the proportion of coordinated duration, and the ratio of change magnitude are integrated to obtain an index that reflects the dynamic change of the interaction strength of the two characteristics over time. This index is the dynamic coupling degree.

[0032] The coupling enhancement coefficient and cross-adjustment coefficient are derived from prior knowledge of the geological disaster mechanics model and historical geological disaster monitoring data. The acquisition process involves collecting multi-source monitoring time-series data from a large number of historical geological disaster cases, extracting displacement-stress covariance characteristics, seepage-microseismic correlation characteristics, and corresponding disaster occurrence situations. Combined with the theoretical basis of the geological disaster mechanics model regarding the impact of displacement-stress coupling and seepage-microseismic coupling on disaster incubation, the degree of matching between the contribution calculated under different coefficient values ​​and the actual disaster incubation stage characteristics is analyzed. By adjusting the coefficient values ​​and repeatedly verifying, a fixed value that can accurately reflect the actual impact of the characteristics on the disaster incubation stage is determined. The final fixed values ​​are used as the coupling enhancement coefficient and cross-adjustment coefficient, respectively.

[0033] The contribution of displacement-stress covariance characteristics increases with the increase of their own change sensitivity. With other parameters remaining constant, the greater the change sensitivity, the more significant the change in displacement-stress covariance characteristics, the more obvious the impact on the geological disaster incubation stage, and the greater the corresponding contribution value.

[0034] The contribution of displacement-stress covariance characteristics increases with the increase of dynamic coupling degree. When other parameters are fixed, the higher the dynamic coupling degree, the stronger the synergistic effect between displacement-stress covariance characteristics and seepage-microseismic correlation characteristics. This synergistic effect will enhance the influence of displacement-stress covariance characteristics on disaster gestation, thus increasing its contribution degree accordingly.

[0035] The contribution of displacement-stress covariant characteristics increases with the increase of coupling enhancement coefficient. Under the condition that other parameters remain unchanged, the increase of coupling enhancement coefficient can enhance the contribution of dynamic coupling degree to displacement-stress covariant characteristics, thereby increasing the value of contribution.

[0036] The contribution of the displacement-stress covariance characteristic decreases as the cross-adjustment coefficient increases. When other parameters are fixed, an increase in the cross-adjustment coefficient will increase the value of the denominator in the formula. With the numerator remaining unchanged, the overall contribution value calculated will decrease accordingly.

[0037] The contribution of seepage-microseismic correlation characteristics increases with the increase of their own sensitivity to change. When other parameters remain unchanged, the greater the sensitivity to change, the more prominent the change in seepage-microseismic correlation characteristics, the more critical the role in the geological disaster incubation stage, and the greater the corresponding contribution value.

[0038] The contribution of seepage-microseismic correlation features increases with the increase of dynamic coupling. With other parameters fixed, the higher the dynamic coupling, the stronger the synergistic effect of seepage-microseismic correlation features and displacement-stress covariance features. This synergistic effect will enhance the influence of seepage-microseismic correlation features on disaster gestation, resulting in a corresponding increase in its contribution.

[0039] The contribution of seepage-microseismic correlation characteristics increases with the increase of coupling enhancement coefficient. When other parameters remain unchanged, the increase of coupling enhancement coefficient can enhance the effect of dynamic coupling degree on the contribution of seepage-microseismic correlation characteristics, thereby increasing the value of contribution.

[0040] The contribution of seepage-microseismic correlation characteristics decreases as the cross-adjustment coefficient increases. When other parameters are fixed, increasing the cross-adjustment coefficient will increase the value of the denominator in the formula. With the numerator unchanged, the overall calculated contribution value decreases accordingly.

[0041] Each data stream in the multi-source fusion dataset, including surface displacement, soil moisture content, underground soil and rock stress, and microseismic signal data stream, was checked item by item. Based on the historical valid data range in the field of geological disaster monitoring, reasonable numerical ranges were determined for each of the four data types: surface displacement (0-50 cm), soil moisture content (10%-60%), underground soil and rock stress (0-500 kPa), and microseismic signal (0-1000 mV). Values ​​exceeding these reasonable ranges were directly identified as extreme anomalies and removed. For missing values ​​encountered during data acquisition, two adjacent valid data points were identified. Based on the values ​​and time interval of these two valid data points, a linear estimation method was used to supplement the missing data, i.e., by calculating the difference between the values ​​of the two valid data points. The difference is allocated according to the proportion of time in which the missing data is located, thus obtaining the supplementary value for the missing data. If the difference between a data point value and the data at adjacent time points exceeds 10% of the maximum value of the reasonable range for that type of data, it is judged as a mutation data. It is verified by combining the corresponding monitoring data of other surrounding sensors at the same time point. If no similar changes are found in the data of surrounding devices and there is no clear geological cause record, the mutation data is deleted, and the abnormal data cleaning is completed. Then, standardization and normalization processing is performed. For each type of cleaned monitoring data, the maximum and minimum values ​​of that type of data are found. The minimum value of that type of data is subtracted from each specific value in that type of data. The difference is then divided by the difference between the maximum and minimum values ​​of that type of data. Through this operation, all monitoring data are uniformly transformed to the value range of 0 to 1, and finally, multi-source monitoring time series data of the geological disaster hazard area are obtained.

[0042] All surface displacement data were extracted from multi-source monitoring time-series data and arranged chronologically to form a surface displacement time-series sequence. The difference between the surface displacement values ​​at two adjacent time points was calculated, and this difference was divided by the time interval between the two time points to obtain the surface displacement change rate at each time point. The surface displacement change rates at all time points were then arranged chronologically to form a surface displacement change rate time-series sequence. Simultaneously, all underground soil and rock stress data were extracted from the multi-source monitoring time-series data and arranged chronologically to form a stress time-series sequence. The stress values ​​at adjacent time points were compared segment by segment. When the stress value at a later time point was higher than that at a previous time point and the difference was greater than 0, it was determined to be a stress increase phase. When the interval is low and the difference is less than 0, it is determined to be the stress decrease stage. When the difference is equal to 0, it is determined to be the stress stability stage. The overall change trend of underground rock and soil stress is clarified by segment-by-segment analysis. The time series of surface displacement change rate is completely correlated with the stress change trend of underground rock and soil according to the time axis. The number of time periods with the surface displacement change rate increasing and the stress in the rising stage and the surface displacement change rate decreasing and the stress in the decreasing stage are counted for each time period. The duration of each time period is recorded. At the same time, the ratio of surface displacement change rate to stress change amplitude in each time period is calculated. The correlation information such as the number of time periods, duration and change amplitude ratio are integrated to obtain the displacement-stress covariance characteristics of the geological disaster hazard area.

[0043] Based on the spatial location of the sensing devices recorded in the multi-source monitoring time-series data, soil moisture content data is divided into several groups, each corresponding to a fixed spatial location. The data in each group are arranged chronologically, and the values ​​of adjacent data points are compared at each time point. When the value at a later time point is higher than that at a previous time point, it is marked as a period of increasing soil moisture content at that spatial location; when the value at a later time point is lower than that at a previous time point, it is marked as a period of decreasing soil moisture content; when the value remains unchanged, it is marked as a stable period. Simultaneously, the temporal sequence of soil moisture content changes at different spatial locations is compared horizontally, and the number of spatial locations showing increases or decreases in moisture content within the same time period is counted. When more than 50% of the spatial locations show both increases and decreases in moisture content within a certain time period, that area is identified as a concentrated area of ​​soil moisture content changes, thus clarifying the spatiotemporal variation pattern of soil moisture content. Microseismic signal data were extracted from multi-source monitoring time-series data. The occurrence time and intensity of each microseismic signal were recorded in chronological order to form a microseismic signal event sequence. The spatiotemporal variation of soil moisture content was correlated with the microseismic signal event sequence. The proportion of microseismic signal events occurring during periods of increasing soil moisture content was calculated to determine the frequency of microseismic events as a function of 10% increase in soil moisture content. The spatial overlap between the location of microseismic events and the area of ​​concentrated soil moisture content variation was compared, and the proportion of overlapping locations to the total number of microseismic event locations was recorded. These correlation information, including temporal trigger ratios, frequency changes, and spatial overlap ratios, were integrated to obtain the seepage-microseismic correlation characteristics of geological hazard areas.

[0044] Prior knowledge of geological hazard mechanics models is collected, including the synergistic mechanism of displacement and stress during soil and rock deformation, the influence of seepage caused by changes in soil moisture content on the stability of soil and rock, and the correlation principle between microseismic signal generation and soil and rock fracture development—all clearly defined knowledge verified by experiments and derived theoretically. Based on the synergistic mechanism of displacement and stress, the degree of synergy between stress and deformation of soil and rock reflected by the displacement-stress covariance characteristics is analyzed. The more synergistic time periods, the longer the duration, and the more stable the ratio of variation amplitude, the more important this characteristic is for predicting the geological hazard gestation stage. The stronger the action, the higher the contribution level of the feature is determined. Based on the influence of seepage on the stability of soil and rock and the correlation principle between microseismic events and soil and rock fracturing, the correlation strength between seepage and microseismic events reflected by the seepage-microseismic correlation feature is analyzed. The higher the proportion of microseismic events occurring during periods of increased soil moisture content, the stronger the correlation between the magnitude of moisture content change and the frequency of microseismic events, and the higher the spatial overlap ratio, the more critical the influence of the feature on the geological disaster incubation stage is. Thus, the contribution level of the feature is determined, and the contribution of the two types of features in the geological disaster incubation stage is evaluated.

[0045] Based on the practical application needs of geological disaster monitoring and early warning, and the verification results of the effectiveness of features in historical disaster cases, a contribution screening standard was set. The lowest contribution level that can effectively indicate the geological disaster gestation state was set as the screening threshold. This threshold was verified by more than 100 sets of historical monitoring data to ensure that features above this threshold can accurately reflect the key coupling relationships in the geological disaster gestation process. The evaluation contribution level of displacement-stress covariance features was compared with the set screening threshold. If the contribution level of a feature is higher than the screening threshold, it is included as a candidate key coupling feature factor. In the same comparison method, the evaluation contribution level of seepage-microseismic correlation features was compared with the screening threshold. If it meets the condition of being higher than the screening threshold, it is also included as a candidate key coupling feature factor. All candidate features that meet the screening standard are integrated to ensure that the integrated features can comprehensively cover the core coupling relationships in the geological disaster gestation process, and finally, the key coupling feature factors of the geological disaster hazard area are obtained.

[0046] The beneficial effects include: by performing anomaly cleaning and standardization on multi-source fusion datasets, extremely abnormal and abrupt data are accurately removed, and missing data is supplemented; the dimensional differences between different monitoring indicators are eliminated, resulting in standardized and anomaly-free multi-source monitoring time-series data, laying a reliable data foundation for subsequent feature extraction. By systematically mining the correlation between the rate of change of surface displacement and the trend of stress change in underground soil and rock, displacement-stress covariance characteristics are extracted, accurately capturing the core coupling mechanism of stress and deformation in soil and rock. Simultaneously, the spatiotemporal correspondence between the spatiotemporal changes in soil moisture content and the sequence of microseismic signal events is analyzed, generating seepage-microseismic correlation characteristics, clearly revealing the influence path of seepage on the stability of soil and rock. Based on the prior knowledge of geological hazard mechanics models, scientific evaluation of the two types of features is conducted to clarify their impact on the geological hazard incubation stage, avoiding the omission of key information due to blind screening. By setting reasonable screening criteria, integrating and retaining features that meet the contribution standards to form key coupled feature factors, and eliminating irrelevant or low-impact features, we can ensure that the feature factors accurately point to the core coupling relationship of geological disasters. This provides targeted and reliable feature support for subsequent geological disaster hazard identification and risk assessment, effectively improving the scientific nature and accuracy of geological disaster monitoring and early warning.

[0047] S3. Simulate the dynamic response process of the geological body under multi-field coupling based on the key coupling characteristic factors, and output the three-dimensional dynamic risk field model of the geological hazard area in the form of a risk probability cloud map. In this embodiment of the invention, the step of simulating the dynamic response process of a geological body under multi-field coupling based on the key coupling characteristic factors includes: The constitutive relation set of the interaction between stress field, seepage field and deformation field is encapsulated into a coupled analysis engine; The key coupling feature factors are configured as boundary conditions and driving parameters into the corresponding input interface of the coupling analysis engine; In the coupling analysis engine, the time step and total simulation duration are set for simulating the dynamic response process; The coupling analysis engine is driven to run iteratively according to the time step within the total simulation duration; After the coupled analysis engine completes its iterative run, it extracts and outputs the simulation state parameters updated by each iteration step, thus obtaining the spatiotemporal evolution sequence of the geological body.

[0048] The method of outputting the three-dimensional dynamic risk field model of the geological hazard hazard area in the form of a risk probability cloud map includes: Assign state parameter values ​​corresponding to the current simulation moment in the spatiotemporal evolution sequence to each grid cell in the three-dimensional spatial grid model of the geological hazard potential area; Based on preset instability criteria and risk assessment rules, the probability value of instability and failure of the grid cell in the next simulation period is predicted; The risk probability values ​​are smoothed to obtain the risk probability field data of the geological hazard hazard area; By using different colors and transparency to characterize the different risk probability levels and spatial distributions in the risk probability field data, a three-dimensional dynamic risk field model of the geological hazard hazard area is obtained.

[0049] This study collects constitutive relations of the interactions between stress field, seepage field, and deformation field in the field of geological hazard mechanics. These relations include the driving law of stress field changes on deformation field, the influence mechanism of seepage field on rock mass pore pressure, and the feedback principle of deformation field reaction on stress field. All constitutive relations are determined based on laboratory geotechnical mechanics test data, field geological hazard monitoring verification results, and derivation from classical mechanics theory, ensuring their accuracy and applicability. These constitutive relations are categorized and organized by function to form a logically coherent set of constitutive relations. Each relation clearly defines input variables, output variables, and interaction logic. Subsequently, the set of constitutive relations is converted into executable code modules through programming. Each module contains a complete process of data reception, logical operation, and result output. Standardized input and output interfaces are designed. The input interface is used to receive subsequent boundary conditions, driving parameters, and other data, while the output interface is used to transmit simulation calculation results. These code modules, interfaces, and built-in logical operation processes are integrated and encapsulated to form a coupled analysis engine with independent running capabilities. This engine can automatically call the corresponding constitutive relations to perform multi-field coupled calculations based on input data.

[0050] The physical meanings of the displacement-stress covariant characteristics and seepage-microseismic correlation characteristics among the key coupling feature factors are clarified. The displacement-stress covariant characteristics correspond to the synergistic state of stress and deformation of the geological body, serving as the driving parameter for the interaction between the stress field and deformation field in the coupling analysis engine. The seepage-microseismic correlation characteristics correspond to the correlation state between seepage and microseismic activity in the geological body, serving as the driving parameter for the interaction between the seepage field and stress field. Based on the preset data format and parameter type of the coupling analysis engine input interface, the key coupling feature factors are format-converted to ensure that the numerical type and data structure of each feature factor fully match the requirements of the corresponding input interface, avoiding calculation anomalies caused by data incompatibility. The converted displacement-stress covariant characteristics are input into the coupling analysis engine through the stress-deformation field driving parameter input interface, and the seepage-microseismic correlation characteristics are input through the seepage-stress field driving parameter input interface. Simultaneously, fixed boundary conditions such as the topographic boundary and soil / rock distribution boundary of the geological hazard area are configured into the engine through a dedicated boundary condition input interface, completing the accurate integration of all input data.

[0051] The time step is determined by combining the actual rate of change of the dynamic response of the geological body. Referencing the change cycles of indicators such as stress, displacement, and water content in historical monitoring data, the set time step ensures that it can fully capture the smallest unit of change for each indicator. This avoids missing key changes due to an excessively large time step, or causing computational redundancy due to an excessively small time step. The total simulation duration is determined based on the actual needs of geological disaster early warning. The total simulation duration must cover the complete gestation stage of the geological body from its current stable state to the potential for instability. It is set by referencing the gestation cycles of similar historical geological disasters and considering the current geological environment of the potential hazard area, ensuring that the simulation process fully reflects the long-term evolution trend of the geological body under multi-field coupling effects. The set time step and total simulation duration are entered through the time parameter configuration interface of the coupling analysis engine. The engine automatically divides the total simulation duration into several consecutive iteration cycles according to the time step, with each iteration cycle corresponding to a time step, laying the time framework for subsequent iterations.

[0052] The coupling analysis engine is launched. First, it reads all input data, including configured boundary conditions, driving parameters, time steps, and total simulation duration. Then, it calls the corresponding multi-field coupling logic from the constitutive relation set according to preset priorities. Within the first time step, the engine calculates the current stress, seepage, and deformation field state parameters of the geological body based on the initial input data, obtaining the simulation results for the first iteration. It then automatically enters the next time step, using the state parameters output from the previous iteration as initial conditions. Combined with the dynamic changes in the driving parameters, it calls the constitutive relation again to calculate and update the multi-field coupling state parameters of the geological body. This iterative process is repeated, with each time step based on the calculation results of the previous step, continuously updating the state parameters until the calculations for all time steps within the total simulation duration are completed. Throughout the iteration process, the engine automatically records the calculation results for each time step, ensuring the continuity of the simulation process and the integrity of the data.

[0053] After completing all iterations, the coupling analysis engine automatically stores the simulated state parameters updated in each iteration step. These state parameters include stress distribution values ​​for the stress field, water content distribution values ​​for the seepage field, and displacement distribution values ​​for the deformation field. Each state parameter contains corresponding spatial location information and a time step identifier. The simulated state parameters for all iteration steps are sorted according to their time steps, and complete state parameter data for all spatial locations at each time step are extracted, forming a two-dimensional data matrix with time as the vertical axis and spatial location as the horizontal axis. The two-dimensional data matrices corresponding to each time step are concatenated in chronological order to form a complete data sequence containing three dimensions: time, space, and state parameters. This sequence clearly presents the changes in the multi-field coupling state of the geological body at different time points and spatial locations, ultimately yielding a spatiotemporal evolution sequence that reflects the entire dynamic response process of the geological body.

[0054] Based on topographic data and geological exploration information of the geological hazard-prone area, a complete three-dimensional spatial grid model is constructed. The grid division is determined according to the geological complexity and monitoring accuracy requirements of the hazard area, ensuring that the grid unit can accurately cover the entire hazard area, and each grid unit has unique three-dimensional spatial coordinates that completely correspond to the spatial location coordinates of the state parameters in the spatiotemporal evolution sequence. After determining the current simulation time, the state parameter values ​​corresponding to all spatial locations at that time are extracted from the spatiotemporal evolution sequence. The state parameter values ​​include stress distribution values, water content distribution values, displacement distribution values, etc. According to the three-dimensional spatial coordinates of the grid unit, the state parameter values ​​corresponding to the spatial locations in the spatiotemporal evolution sequence are assigned to each grid unit one by one, ensuring that each grid unit can obtain the state parameter values ​​at the current simulation time that accurately match its own spatial location, thus achieving precise binding between the state parameters and the three-dimensional spatial grid.

[0055] The preset instability criteria are determined based on the mechanical properties of soil and rock, historical disaster instability data, and geological disaster mechanics models. They clearly define the critical values ​​for different types of state parameters. For example, when the stress parameter reaches a certain fixed value, the displacement parameter exceeds a specific range of change, or the water content parameter is within a certain range, the soil and rock mass meets the instability triggering conditions. The risk assessment rules clarify the gap between state parameter values ​​and instability critical values, and the correspondence between the changing trends of state parameters and risk probability values. For example, the closer the state parameter value is to the instability critical value and the faster the rate of change, the higher the corresponding risk probability value. For each grid cell with assigned state parameter values, its current state parameter value is compared one by one with the critical values ​​in the instability criteria. The analysis shows whether the changing trend of the state parameters is developing in the direction of instability. Combined with the risk assessment rules, the specific risk probability value of the grid cell in the next simulation period is directly determined, ensuring that each grid cell receives an accurate corresponding risk probability value.

[0056] The risk probability values ​​of all grid cells are smoothed cell-by-cell. During the process, each grid cell is considered as the center, and its neighboring grid cells are selected as reference cells. The risk probability values ​​of the central grid cell and the reference cells are collected, and the arithmetic mean of these seven grid cell risk probability values ​​is calculated. During the calculation, it is ensured that each value is equally included, without bias towards any single cell. The calculated arithmetic mean replaces the original risk probability value of the central grid cell. This method eliminates abrupt changes in risk probability values ​​caused by local anomalies in a single grid cell, resulting in a natural and continuous transition between risk probability values ​​of adjacent grid cells. Following the same method, all grid cells in the 3D spatial grid model are smoothed sequentially, ultimately yielding a risk probability field data for the geological hazard hazard area with a continuous overall distribution and no significant abrupt changes.

[0057] The risk probability is clearly defined based on its magnitude, for example, divided into four levels from low to high: low risk, medium risk, high risk, and extremely high risk. Each level corresponds to a fixed range of risk probability values, and the classification criteria are determined based on the actual needs of geological disaster early warning and historical risk assessment experience. Each risk probability level is assigned a unique color, such as blue for low risk, yellow for medium risk, orange for high risk, and red for extremely high risk. Different levels also have different levels of transparency; the higher the risk level, the lower the transparency, making high-risk areas more clearly highlighted in the model, while low-risk areas remain relatively transparent, not obscuring spatial information from other areas. The risk probability value of each grid cell in the risk probability field data is matched with the defined risk probability level, assigning corresponding color and transparency attributes. Combined with the spatial structure of the 3D spatial grid model, all grid cells with color and transparency attributes are integrated and rendered to form a model that intuitively displays different risk probability levels and their spatial distribution. This model adjusts its color and transparency distribution in real time as the simulation progresses based on updated risk probability field data, ultimately resulting in a 3D dynamic risk field model of the geological disaster hazard area.

[0058] The beneficial effects include: by encapsulating and integrating constitutive relations of stress, seepage, and deformation fields based on experimental data, field verification, and theoretical derivation, a coupled analysis engine with standardized interfaces and complete computational logic is constructed. This ensures that multi-field coupled calculations are scientifically reliable, providing professional and stable core support for dynamic response simulation of geological bodies. Key coupling characteristic factors are precisely adapted as boundary conditions and driving parameters and integrated into the engine, making the simulation process closely match the actual geological coupling relationships in the hazard area, avoiding interference from irrelevant parameters, and significantly improving the simulation's relevance and accuracy. The time step and total simulation duration are scientifically set based on the actual rate of change of the geological body and early warning requirements, which can fully capture the smallest changing unit of the geological body and cover the complete gestation stage of the disaster, balancing simulation detail and comprehensiveness, and avoiding information omissions or computational redundancy. The driving engine runs iteratively according to the time step, using the results of each iteration as the initial conditions for the next round of calculation, realizing dynamic updates of the multi-field coupled state, ensuring that the simulation process is consistent with the actual evolution law of the geological body, and restoring a continuous and realistic dynamic response process. By extracting and integrating the simulated state parameters of each iteration step, a three-dimensional spatiotemporal evolution sequence covering time, space, and state parameters is formed. This clearly presents the state changes of geological bodies at different stages and locations under the coupling of multiple fields, providing comprehensive and systematic data support for dynamic tracking, risk assessment, and early warning decision-making of geological disaster hazards, and effectively improving the foresight and scientific nature of geological disaster early warning.

[0059] By precisely binding the state parameter values ​​at the current simulation moment in the spatiotemporal evolution sequence with three-dimensional spatial grid cells, a one-to-one correspondence between state parameters and spatial locations is achieved, providing a precise spatial data foundation for subsequent risk assessment. Based on scientifically determined instability criteria and risk assessment rules, the instability risk probability value of the grid cells in the next simulation period is predicted, ensuring that the risk probability assessment conforms to the actual instability patterns of the geological body and improving the accuracy and reliability of risk prediction. Smoothing of the risk probability values ​​effectively eliminates probability abrupt changes caused by local anomalies, making the risk probability field data distribution continuous and uniform, enhancing the rationality and usability of the data. Different risk probability levels and spatial distributions are precisely represented by dedicated colors and transparency, and combined with the three-dimensional spatial grid model for integrated rendering to form a three-dimensional dynamic risk field model. This model intuitively and clearly presents the risk distribution of hazard areas and can be updated in real time as the simulation progresses, providing intuitive and visual core support for geological disaster risk assessment and early warning decision-making, helping to accurately identify high-risk areas and improve the pertinence and effectiveness of geological disaster prevention and control.

[0060] S4. Extract the dynamic feature vectors of risk evolution rate and risk spatial clustering degree in the three-dimensional dynamic risk field model; In this embodiment of the invention, extracting the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model includes: Extract the risk probability time series of each spatial unit within a preset time window from the three-dimensional dynamic risk field model; The rate of change of risk probability of each spatial unit in the risk probability time series is used as a quantitative indicator of the risk evolution rate in the geological hazard hazard area. Identify the spatial clustering index of high-risk probability regions in the three-dimensional dynamic risk field model; By combining the quantitative index of risk evolution rate with the index of spatial clustering, a dynamic feature vector of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model is obtained.

[0061] Before extracting the risk probability time series of each spatial unit within a preset time window from the 3D dynamic risk field model, the preset time window is determined based on the typical timescale of geological disaster evolution and actual early warning needs. The start and end simulation times of the time window are clearly defined to ensure that the time window fully covers the critical periods during which significant changes in risk probability occur. All spatial units in the 3D dynamic risk field model are located, each corresponding to a unique 3D spatial coordinate. Following the chronological order from the start to the end simulation time, the risk probability value corresponding to each simulation time within the time window for each spatial unit is extracted one by one. All extracted risk probability values ​​for the same spatial unit are then arranged chronologically to form a risk probability time series specific to each spatial unit and strictly corresponding to time, ensuring that each risk probability value in the series accurately matches the corresponding simulation time and spatial unit.

[0062] For each extracted spatial unit's risk probability time series, the risk probability values ​​corresponding to two adjacent simulated moments in the series are selected one by one. The difference between the risk probability value at the later moment and the risk probability value at the previous moment is calculated. This difference directly reflects the change in risk probability between the two moments. The calculated change in risk probability is divided by the time interval between these two adjacent simulated moments to obtain the average change in risk probability within that time interval. This average change is the rate of change of risk probability of the spatial unit over time within the corresponding time interval. The rate of change for all adjacent time intervals is calculated sequentially according to the order of adjacent moments in the time series. These rates of change are then integrated in chronological order to form a sequence that can completely reflect the rate of change of risk probability of the spatial unit. This sequence is the quantitative indicator of the risk evolution rate of the corresponding spatial unit in the geological hazard hazard area.

[0063] When identifying the spatial clustering index of high-risk probability regions in a 3D dynamic risk field model, the threshold range of high-risk probability is first determined based on a preset risk level classification standard. This threshold range is consistent with the probability ranges corresponding to high-risk and extremely high-risk levels set during the initial construction of the 3D dynamic risk field model. All spatial units in the 3D dynamic risk field model are traversed, and the risk probability value of each spatial unit is extracted. This value is compared with the high-risk probability threshold range, and all spatial units with risk probability values ​​within this threshold range are selected. The 3D spatial coordinates of these high-risk spatial units are then marked. The total number of selected high-risk spatial units is counted, and the 3D spatial volume occupied by these high-risk units is calculated. The connection between adjacent high-risk units is then analyzed, and the proportion of adjacent high-risk unit pairs to the total number of high-risk units is calculated. These three pieces of information—the number of high-risk units, the occupied spatial volume, and the adjacent connection ratio—are integrated to form an index that comprehensively reflects the spatial concentration of high-risk regions. This index is the spatial clustering index of high-risk probability regions.

[0064] Before combining the quantitative indicators of risk evolution rate with the spatial clustering degree indicators, the matching relationship between the two indicators corresponding to the same spatial unit must be clarified to ensure that the quantitative indicator of risk evolution rate of each spatial unit can accurately correspond to the spatial clustering degree indicator of the area where the unit is located. Following a fixed order, the quantitative indicators of risk evolution rate and the corresponding spatial clustering degree indicators of the same spatial unit are arranged in an orderly manner to form a vector containing two dimensions: one dimension is the risk evolution rate information, and the other dimension is the spatial clustering degree information. This combination operation is performed on all spatial units in the three-dimensional dynamic risk field model. The vectors corresponding to all spatial units are integrated and summarized to finally obtain the dynamic feature vector of risk evolution rate and risk spatial clustering degree in the three-dimensional dynamic risk field model, which can comprehensively reflect the rate of risk change and the concentration of high-risk areas in each spatial unit of the entire geological hazard hazard area.

[0065] The beneficial effects are as follows: By determining the preset time window based on the evolution characteristics of geological hazards and early warning needs, the risk probability time series of each spatial unit during key periods is extracted. This ensures that the acquired time series can accurately capture the significant changes in risk probability, providing a targeted data foundation for subsequent dynamic feature quantification. A risk evolution rate quantification index is obtained by calculating the ratio of the change in risk probability at adjacent times to the time interval. This index intuitively and accurately represents the speed of risk change in each spatial unit, clearly reflecting the dynamic evolution of risk. High-risk spatial units are screened based on a unified high-risk threshold. The spatial clustering index is obtained by integrating the number of units, occupied volume, and adjacent connection ratio, comprehensively and accurately depicting the spatial concentration characteristics of high-risk areas. The risk evolution rate quantification index and the spatial clustering index are precisely matched and systematically combined to form a dynamic feature vector that fully covers both the speed of risk change and the spatial clustering state. This provides comprehensive and crucial feature support for subsequent in-depth analysis, trend prediction, and accurate early warning of geological hazard risks, improving the scientific rigor and relevance of risk assessment.

[0066] In this embodiment of the invention, S5, an early warning decision is made on the geological hazard area based on the dynamic feature vector to obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard area; Historical disaster case data, dynamic feature vector samples corresponding to different risk evolution stages, and corresponding disaster intensity and early warning response data were collected from geological hazard-prone areas. Combined with the theoretical basis of risk evolution stages and disaster occurrence probability in geological hazard mechanics models, a complete early warning decision-making rule was formulated. This rule clarifies the correspondence between different numerical ranges and combinations of quantitative indicators of risk evolution rate and risk spatial aggregation degree indicators in dynamic feature vectors and early warning levels. It also clarifies the risk evolution stage corresponding to each early warning level. For example, a slow risk evolution rate and low spatial aggregation degree correspond to the risk nascent stage, matching a low early warning level; a fast risk evolution rate and high spatial aggregation degree correspond to the risk escalation and impending instability stage, matching a high early warning level. All correspondences have been verified by historical data to ensure accurate reflection of the correlation between dynamic feature vectors and risk evolution stages and early warning levels.

[0067] The dynamic feature vectors of the extracted three-dimensional dynamic risk field model are comprehensively compared with the preset early warning decision rules. The value of the risk evolution rate quantification index in the dynamic feature vector is checked one by one to see if it meets the rate range corresponding to a certain early warning level in the rules. At the same time, the value of the risk spatial aggregation degree index is checked to see if it meets the aggregation degree range corresponding to the early warning level. After confirming that both indicators meet the corresponding conditions of the same early warning level, the current risk evolution stage of the geological disaster hazard area is determined. Based on the early warning level corresponding to the risk evolution stage, an early warning level instruction is generated, which includes the early warning level, a description of the risk evolution stage, key areas of concern, and preliminary disposal suggestions. This instruction can accurately match the actual situation of the current risk evolution stage and provide a clear decision-making basis for subsequent geological disaster prevention and control work.

[0068] The beneficial effects include: relying on historical disaster cases, dynamic feature vector samples, disaster response data, and geological disaster mechanics theory, establishing precise correspondence rules between dynamic feature vectors and risk evolution stages and warning levels. These rules are validated and calibrated using historical data, ensuring the scientific rigor and reliability of warning decisions. A comprehensive comparison of dynamic feature vectors with warning decision rules accurately verifies the numerical matching of risk evolution rate and spatial clustering indicators, clearly defining the current risk evolution stage and ensuring the accuracy of warning level determination. Warning level instructions are generated, encompassing warning levels, descriptions of risk evolution stages, key areas of concern, and preliminary response recommendations. This comprehensive and targeted information provides clear and specific action guidelines for prevention and control work. The entire warning decision-making process achieves precise matching between dynamic feature vectors and warning levels, ensuring a high degree of consistency between warning instructions and the current risk evolution stage. This significantly improves the accuracy of geological disaster warnings and the scientific rigor of prevention and control decisions, providing solid support for the timely deployment of targeted prevention and control measures and the reduction of disaster losses.

[0069] S6. Based on the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, optimize the parameters in the dynamic response process in reverse, and use the optimized parameters for the construction and updating of the three-dimensional dynamic risk field model in the next round.

[0070] In this embodiment of the invention, the step of reversely optimizing the parameters in the dynamic response process based on the morphological change characteristics of the warning level instruction and the three-dimensional dynamic risk field model includes: Based on the level of the early warning instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, an optimization target for improving the accuracy of the next simulation cycle is generated. Based on the optimization objective, a set of parameters to be adjusted associated with the optimization objective is identified from the constitutive relation set of the coupling analysis engine; The set of parameters to be adjusted is fine-tuned according to a preset adjustment strategy to generate optimized parameters.

[0071] First, clarify the risk control accuracy requirements corresponding to the warning level. Higher warning levels correspond to higher simulation accuracy requirements, while lower warning levels only require ensuring basic simulation accuracy. Simultaneously, meticulously analyze the morphological change characteristics of the three-dimensional dynamic risk field model, specifically including the continuity of risk probability distribution, the clarity of the spatial outline of high-risk areas, the smoothness of the transition of risk field morphology at different simulation times, and the degree of fit between risk evolution trends and actual monitoring data. Combine the warning level requirements with the morphological change characteristic analysis results. If the warning level is high and the model has problems such as blurred high-risk area outlines and significant deviations between evolution trends and reality, the optimization objective is set to improve the spatial positioning accuracy of high-risk areas and the prediction accuracy of risk evolution trends. If the warning level is low but the model has local risk probability abrupt changes, the optimization objective is set to enhance the continuity of the risk field distribution. Finally, generate optimization objectives that clearly point to the direction of accuracy improvement for the next simulation cycle.

[0072] The functions of all parameters in the constitutive relation set of the coupling analysis engine are analyzed, and the simulation stage corresponding to each parameter is clarified. For example, some parameters control the interaction strength between the stress field and the deformation field, some affect the propagation rate of the seepage field, and some are related to the stress-seepage coupling effect. The generated optimization objectives are matched with the functions of each parameter one by one. If the optimization objective is to improve the spatial positioning accuracy of high-risk areas, parameters related to stress distribution calculation and spatial propagation of the deformation field in the constitutive relation set are selected. If the optimization objective is to enhance the accuracy of risk evolution trend prediction, parameters related to the field update rate within the time step and the feedback strength of coupling effect are identified. All parameters that match the functions of the optimization objective are summarized and organized to form a set of parameters to be adjusted that are directly related to the optimization objective.

[0073] The pre-defined adjustment strategy clarifies the basis for determining the adjustment direction and the adjustment range standard for each parameter in the parameter set to be adjusted. The adjustment direction is determined based on the optimization objective. For example, if the optimization objective is to improve the stress simulation accuracy in high-risk areas, and the calculated stress values ​​in high-risk areas in the current model are lower than the actual monitored values, then the adjustment direction for the corresponding stress-related parameters is to increase. The adjustment range standard specifies the numerical range for each fine-tuning. This range is determined based on historical parameter adjustment experience and the stability requirements of constitutive relations, ensuring that the adjustment does not destroy the rationality of the constitutive relations. According to the pre-defined adjustment strategy, the specific adjustment direction for each parameter to be adjusted is first determined, and then small numerical adjustments are made within the specified adjustment range. After the adjustment is completed, it is checked whether the adjusted parameters maintain logical consistency with other related parameters in the constitutive relation set to avoid parameter conflicts. All parameters that have completed fine-tuning and are logically consistent are integrated to generate the optimized parameters.

[0074] The beneficial effect is that by combining the level requirements of the early warning level instructions with the morphological change characteristics of the three-dimensional dynamic risk field model, the direction for improving the accuracy of the next simulation cycle can be accurately located, generating targeted optimization targets, ensuring that parameter optimization always revolves around the actual performance of the model and the needs of risk prevention and control, and avoiding aimless adjustments.

[0075] By systematically reviewing the functions of each parameter in the constitutive relation set of the coupling analysis engine, the optimization objectives are precisely matched with the parameter functions. This efficiently identifies closely related sets of parameters to be adjusted, eliminates interference from irrelevant parameters, ensures the focus and effectiveness of parameter optimization, and avoids model stability issues caused by blind adjustments. Based on a pre-defined adjustment strategy, the direction of parameter adjustment is determined according to the optimization objectives. Small-scale fine-tuning is performed within a reasonable range, while simultaneously verifying the logical consistency of the adjusted parameters with the constitutive relation and other related parameters. This ensures that the optimized parameters meet the accuracy improvement requirements without compromising the rationality of the constitutive relation and the stability of the model. The optimized parameters are used for the construction and updating of the next round of the 3D dynamic risk field model, achieving closed-loop optimization of model parameters. This continuously improves the model's simulation accuracy, spatial positioning accuracy, and reliability of evolution trend prediction for geological disaster risks, enabling the model to dynamically adapt to the actual changes in geological disaster hazard areas and provide more accurate support for subsequent early warning decisions.

[0076] like Figure 2 The diagram shown is a functional module diagram of a dynamic comprehensive early warning system for geological disasters based on three-dimensional monitoring, provided in an embodiment of the present invention.

[0077] The geological disaster dynamic comprehensive early warning system 100 based on three-dimensional monitoring described in this invention can be installed in an electronic device. Depending on the functions implemented, the geological disaster dynamic comprehensive early warning system 100 based on three-dimensional monitoring may include a multi-source data acquisition and fusion module 101, a key feature factor identification module 102, a dynamic risk field simulation and visualization module 103, a risk evolution feature extraction module 104, an intelligent early warning decision-making module 105, and a model parameter adaptive optimization module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0078] In this embodiment, the functions of each module / unit are as follows: The multi-source data acquisition and fusion module 101 is used to deploy and activate a network of multi-source sensing devices deployed in the geological hazard hazard area, acquire multi-dimensional three-dimensional monitoring data streams of the geological hazard hazard area in real time, and perform time-series alignment and spatial registration on the multi-dimensional three-dimensional monitoring data streams to obtain a multi-source fusion dataset of the geological hazard hazard area. The key feature factor identification module 102 is used to mine the causal relationship of the geological disaster gestation process and quantify the dynamic contribution weight based on the dynamic Bayesian network to obtain the key coupling feature factors in the multi-source fusion dataset. The dynamic risk field simulation and visualization module 103 is used to simulate the dynamic response process of the geological body under the multi-field coupling effect according to the key coupling characteristic factors, and output the three-dimensional dynamic risk field model of the geological hazard area in the form of risk probability cloud map. The risk evolution feature extraction module 104 is used to extract the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model. The intelligent early warning decision module 105 is used to make early warning decisions for the geological hazard area based on the dynamic feature vector, and obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard area. The model parameter adaptive optimization module 106 is used to optimize the parameters in the dynamic response process in reverse according to the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, and use the optimized parameters for the construction and updating of the three-dimensional dynamic risk field model in the next round.

[0079] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0080] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0081] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0082] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0083] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A dynamic comprehensive early warning method for geological disasters based on three-dimensional monitoring, characterized in that, The method includes: By deploying a network of multi-source sensing devices in areas prone to geological disasters, multi-dimensional three-dimensional monitoring data streams of the areas prone to geological disasters are acquired in real time. The multi-dimensional three-dimensional monitoring data streams are then time-series aligned and spatially registered to obtain a multi-source fusion dataset of the areas prone to geological disasters. Based on dynamic Bayesian networks, causal association mining and dynamic contribution weight quantification of the geological disaster gestation process are performed on the multi-source fusion dataset to obtain key coupling feature factors in the multi-source fusion dataset. Based on the key coupling characteristic factors, the dynamic response process of the geological body under multi-field coupling is simulated, and the three-dimensional dynamic risk field model of the geological hazard area is output in the form of a risk probability cloud map. Extract the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model; Based on the dynamic feature vector and the preset early warning decision rules, an early warning decision is made for the geological hazard hazard area to obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard hazard area. The preset early warning decision rules are used to define the correspondence between the different numerical ranges and combinations of the quantitative indicators of risk evolution rate and risk spatial aggregation degree indicators in the dynamic feature vector and the early warning level, and the risk evolution stage corresponding to each early warning level. Based on the warning level instructions and the morphological change characteristics of the three-dimensional dynamic risk field model, the parameters in the dynamic response process are optimized in reverse, and the optimized parameters are used for the construction and updating of the three-dimensional dynamic risk field model in the next round.

2. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 1, characterized in that, The method involves using a network of multi-source sensors deployed in geological hazard-prone areas to acquire multi-dimensional, three-dimensional monitoring data streams of these areas in real time. The data streams are then time-series aligned and spatially registered to obtain a multi-source fusion dataset of the geological hazard-prone areas, including: Deploy and activate sensing devices located on the surface and in boreholes within geological hazard zones to form a multi-source sensing device network in the geological hazard zone. Receive surface displacement data stream, soil moisture content data stream, underground rock and soil stress data stream, and microseismic signal data stream from the multi-source sensing device network, add a timestamp to each data stream, and obtain the multi-source asynchronous data stream of the geological hazard area; Based on the timestamp, the multi-source asynchronous data stream is interpolated and synchronized to obtain a synchronized monitoring data stream with a completely aligned time series. Based on a preset geographic information system coordinate reference, each monitoring data point in the synchronized monitoring data stream is mapped to a unified three-dimensional spatial coordinate system to obtain the registration data stream of the geological hazard area; The registered data stream is restructured according to the time-space dimension to obtain a multi-source fusion dataset of the geological hazard hazard area.

3. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 1, characterized in that, The method, based on a dynamic Bayesian network, mines the causal relationships of geological disaster formation processes and quantifies dynamic contribution weights in the multi-source fusion dataset, obtaining key coupling feature factors in the multi-source fusion dataset, including: The multi-source fusion dataset is subjected to anomaly cleaning and standardization normalization to obtain multi-source monitoring time-series data of the geological hazard hazard area; From the multi-source monitoring time series data, the correlation between the rate of change of surface displacement and the trend of stress change in underground rock and soil is extracted to obtain the displacement-stress covariance characteristics of the geological hazard zone. From the multi-source monitoring time-series data, the correlation between the spatiotemporal variation of soil moisture content and the occurrence sequence of microseismic signal events is extracted to obtain the seepage-microseismic correlation characteristics of the geological hazard area. Based on the prior knowledge of the pre-set geological hazard mechanics model, the contribution of the displacement-stress covariance characteristics and the seepage-microseismic correlation characteristics to the geological hazard incubation stage is evaluated. Based on the contribution degree, features with a contribution degree higher than the set standard are selected from the displacement-stress covariance features and the seepage-microseismic correlation features to obtain the key coupling feature factors of the geological hazard hazard area.

4. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 3, characterized in that, The formula for calculating the contribution of the displacement-stress covariance characteristics to the geological hazard incubation stage is as follows: ； In the formula, The contribution of the displacement-stress covariance characteristic. The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, This is the cross-adjustment coefficient; The formula for calculating the contribution of the seepage-microseismic correlation characteristics to the geological hazard incubation stage is as follows: ； In the formula, The contribution of the seepage-microseismic correlation characteristics, The sensitivity to self-variation is obtained by normalizing the displacement-stress covariance characteristics. The sensitivity to self-change is obtained by normalizing the seepage-microseismic correlation characteristics. The dynamic coupling degree between the displacement-stress covariant characteristics and the seepage-microseismic correlation characteristics. For coupling enhancement coefficient, This is the cross-adjustment coefficient.

5. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 4, characterized in that, The simulation of the dynamic response process of geological bodies under multi-field coupling based on the key coupling characteristic factors includes: The constitutive relation set of the interaction between stress field, seepage field and deformation field is encapsulated into a coupled analysis engine; The key coupling feature factors are configured as boundary conditions and driving parameters into the corresponding input interface of the coupling analysis engine; In the coupling analysis engine, the time step and total simulation duration are set for simulating the dynamic response process; The coupling analysis engine is driven to run iteratively according to the time step within the total simulation duration; After the coupled analysis engine completes its iterative run, it extracts and outputs the simulation state parameters updated by each iteration step, thus obtaining the spatiotemporal evolution sequence of the geological body.

6. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 5, characterized in that, The process of driving the coupling analysis engine to run iteratively according to the time step within the total simulation duration includes: In the coupling analysis engine, the initial state vector of the geological body is initialized based on the key coupling feature factors and preset initial geological parameters; Starting with the initial state vector, an iterative loop is initiated for the total simulation duration; In each iteration, based on the key coupling feature factor corresponding to the current iteration step and the geological body state vector corresponding to the current iteration step, the constitutive relation set is invoked to generate the change in the geological body state under the current multi-field coupling effect. Based on the change, update the current geological body state vector to obtain the geological body state vector corresponding to the next iteration step; The updated geological body state vector is used as the new current geological body state vector for the next iteration cycle; When the number of iterations reaches the number of iteration steps corresponding to the total simulation time, the iteration loop is terminated and the updated state vector of the geological body in all iterations is output.

7. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 6, characterized in that, The method of outputting the three-dimensional dynamic risk field model of the geological hazard hazard area in the form of a risk probability cloud map includes: Assign state parameter values ​​corresponding to the current simulation moment in the spatiotemporal evolution sequence to each grid cell in the three-dimensional spatial grid model of the geological hazard risk area; Based on preset instability criteria and risk assessment rules, the probability value of instability and failure of the grid cell in the next simulation period is predicted; The risk probability values ​​are smoothed to obtain the risk probability field data of the geological hazard hazard area; By using different colors and transparency to characterize the different risk probability levels and spatial distributions in the risk probability field data, a three-dimensional dynamic risk field model of the geological hazard hazard area is obtained.

8. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 7, characterized in that, The extraction of dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model includes: Extract the risk probability time series of each spatial unit within a preset time window from the three-dimensional dynamic risk field model; The rate of change of risk probability of each spatial unit in the risk probability time series is used as a quantitative indicator of the risk evolution rate in the geological hazard hazard area. Identify the spatial clustering index of high-risk probability regions in the three-dimensional dynamic risk field model; By combining the quantitative index of risk evolution rate with the index of spatial clustering, a dynamic feature vector of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model is obtained.

9. The method for dynamic comprehensive early warning of geological disasters based on three-dimensional monitoring as described in claim 8, characterized in that, The step of back-optimizing the parameters in the dynamic response process based on the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model includes: Based on the level of the early warning instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, an optimization target for improving the accuracy of the next simulation cycle is generated. Based on the optimization objective, a set of parameters to be adjusted associated with the optimization objective is identified from the constitutive relation set of the coupling analysis engine; The set of parameters to be adjusted is fine-tuned according to a preset adjustment strategy to generate optimized parameters.

10. A dynamic integrated early warning system for geological disasters based on three-dimensional monitoring, characterized in that, The system for implementing the dynamic comprehensive early warning method for geological disasters based on three-dimensional monitoring as described in claim 1 includes: The multi-source data acquisition and fusion module is used to deploy and activate a network of multi-source sensing devices deployed in geological hazard hazard areas, acquire multi-dimensional three-dimensional monitoring data streams of the geological hazard hazard areas in real time, and perform temporal alignment and spatial registration on the multi-dimensional three-dimensional monitoring data streams to obtain multi-source fusion datasets of the geological hazard hazard areas. The key feature factor identification module is used to mine the causal relationship of the geological disaster gestation process and quantify the dynamic contribution weight based on the dynamic Bayesian network to obtain the key coupling feature factors in the multi-source fusion dataset. The dynamic risk field simulation and visualization module is used to simulate the dynamic response process of geological bodies under multi-field coupling based on the key coupling characteristic factors, and output the three-dimensional dynamic risk field model of the geological hazard area in the form of a risk probability cloud map. The risk evolution feature extraction module is used to extract the dynamic feature vectors of risk evolution rate and risk spatial clustering in the three-dimensional dynamic risk field model. The intelligent early warning decision module is used to make early warning decisions for the geological hazard hazard area based on the dynamic feature vector, and obtain an early warning level instruction that matches the current risk evolution stage in the geological hazard hazard area; The model parameter adaptive optimization module is used to optimize the parameters in the dynamic response process in reverse according to the warning level instruction and the morphological change characteristics of the three-dimensional dynamic risk field model, and use the optimized parameters for the construction and update of the three-dimensional dynamic risk field model in the next round.

Images