Mine slope risk assessment method and system based on multi-source data space-time fusion
By constructing a three-dimensional geological-engineering coupled base model and a spatiotemporal fusion risk field, the problem of data silos in mine slope monitoring was solved, enabling full-field, dynamic perception and intelligent early warning of slope risks, thus improving the accuracy and foresight of early warnings.
Patent Information
- Application Number
- CN202511689927.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-18
- Publication Date
- 2026-02-10
AI Technical Summary
Existing technologies for mine slope monitoring suffer from data silos. The lack of a unified spatiotemporal benchmark for multi-source heterogeneous data leads to inaccurate risk assessments and delayed early warnings, and fails to effectively integrate the dynamic coupling effects of multiple factors.
A three-dimensional geological-engineering coupled base model is constructed, multi-source heterogeneous monitoring data are unified, and a four-dimensional spatiotemporal fusion risk field is generated through spatiotemporal fusion to realize the quantitative perception of risk intensity at any point in the slope body, and intelligent early warning is carried out by combining spatiotemporal convolutional neural network.
It enables full-field, continuous, and dynamic perception of slope risks, improves the accuracy and foresight of early warnings, and can capture instability precursors earlier and more accurately, providing intelligent hierarchical early warning support.
Smart Images

Figure CN121504171A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mine monitoring technology, and in particular to a method and system for mine slope risk assessment based on spatiotemporal fusion of multi-source data. Background Technology
[0002] The stability of open-pit mine slopes is crucial for ensuring safe production and protecting the lives and property of miners. However, mine slopes are massive geotechnical engineering structures dynamically coupled with multiple factors, including complex geological conditions, frequent blasting vibrations, groundwater level fluctuations, and rainfall infiltration. Their instability mechanisms are complex, and traditional monitoring and assessment methods face severe challenges. Accurate and forward-looking risk warnings remain a pressing problem for the industry.
[0003] Currently, the commonly used technical approach in the industry relies on independent analysis and threshold determination of limited, discrete monitoring point data. In practice, this involves first acquiring the displacement of scattered points by deploying GNSS receivers on the slope surface, or using inclinometers to monitor deep displacement inside boreholes; secondly, regular manual inspections and recording of macroscopic deformation indicators; and finally, setting fixed safety thresholds for the above data. When the data from a monitoring point exceeds its threshold, an alarm is triggered. It has some obvious limitations. At the data level, monitoring methods such as GNSS, inclinometers, and manual inspections are independent of each other, forming data silos. Moreover, the acquired heterogeneous data from multiple sources, such as displacement, water level, and vibration, differ greatly in spatiotemporal reference and structure. There is a lack of a unified three-dimensional spatial structure model as a basis for effective fusion, resulting in a one-sided and fragmented monitoring of slope conditions. At the risk assessment level, most existing methods are based on static thresholds of single indicators such as displacement or velocity, failing to fuse multi-source monitoring data to generate a continuous dynamic field that can reflect the risk intensity at any point inside the slope. The early warning model cannot fully consider the dynamic coupling effect of multiple factors, resulting in delayed and inaccurate early warnings. Summary of the Invention
[0004] In view of this, this invention proposes a method and system for mine slope risk assessment based on multi-source data spatiotemporal fusion. By constructing a unified three-dimensional geological-engineering coupled base model as a spatiotemporal benchmark, it effectively integrates multi-source heterogeneous monitoring data such as GNSS, deep displacement, blasting vibration, and groundwater level, fundamentally solving the data silo problem of traditional monitoring methods. At the same time, through the data fusion model, a four-dimensional spatiotemporal fusion risk field that can characterize the risk intensity at any location within the slope at any time is generated. This transforms risk assessment from static threshold judgment based on discrete points and single indicators to a full-field, continuous, and quantitative perception of the spatiotemporal dynamic evolution of risk, greatly improving the accuracy and foresight of early warning. Finally, based on the overall evolution of the risk field, a graded early warning is triggered, realizing a fundamental transformation from a one-sided and lagging single alarm to intelligent and visualized full-field situational early warning.
[0005] The technical solution of this invention is implemented as follows: On the one hand, this invention provides a method for mine slope risk assessment based on spatiotemporal fusion of multi-source data, including: Acquire geological and engineering data, and construct a basic model representing the three-dimensional spatial structure of the slope; Multi-source heterogeneous monitoring data were collected on-site at the mine slope, and the multi-source heterogeneous monitoring data was aligned and fused with the basic model in the spatiotemporal dimension to generate a continuous dynamic parametric field. A data fusion model is set up, and a four-dimensional spatiotemporal fusion risk field is obtained by taking the basic model and dynamic parameter field as input. The four-dimensional spatiotemporal fusion risk field is used to characterize the risk intensity of any point in the slope body at any time. Based on the overall evolution of the spatiotemporal integrated risk field, slope risk early warning is carried out.
[0006] Based on the above technical solutions, preferably, the acquisition of geological and engineering data and the construction of a basic model representing the three-dimensional spatial structure of the slope include: Acquire geological exploration data and generate a three-dimensional geological model that includes rock mass mechanical parameters and fault structure planes; Acquire 3D surface data and generate a 3D surface model containing step geometry information; The three-dimensional geological model is coupled with the three-dimensional surface model to form a three-dimensional geological-engineering coupled base model, which serves as the basic model.
[0007] More preferably, the coupling of the three-dimensional geological model with the three-dimensional surface model includes the following steps: Using the three-dimensional surface model as a geometric constraint boundary, the morphology of rock layer interfaces and fault structure surfaces in the near-surface part of the three-dimensional geological model is corrected and fitted, so that the surface geometry of the geological model is consistent with the three-dimensional surface model. The geometric information of the steps extracted from the 3D surface model is used as engineering attribute data, and then associated and assigned to the corresponding spatial location in the corrected 3D geological model. A three-dimensional geological-engineering coupled base model is generated by performing geometric corrections and attribute associations.
[0008] Based on the above technical solutions, preferably, the step of collecting multi-source heterogeneous monitoring data on the mine slope in the field, aligning and fusing the multi-source heterogeneous monitoring data with the basic model in the spatiotemporal dimension to generate a continuous dynamic parametric field includes: Collect multi-source heterogeneous monitoring data, which includes at least GNSS surface displacement data, deep displacement data, blasting vibration data, and groundwater level data; Multi-source heterogeneous monitoring data are uniformly mapped to the spatiotemporal coordinate system of the basic model to complete spatiotemporal alignment; The spatiotemporally aligned multi-source heterogeneous monitoring data is transformed into a spatially and temporally continuous dynamic parametric field by using a spatiotemporal interpolation algorithm.
[0009] More preferably, the step of uniformly mapping multi-source heterogeneous monitoring data to the spatiotemporal coordinate system of the basic model includes: For blasting vibration data, extract peak vibration velocity and dominant frequency characteristic parameters; Based on the aforementioned characteristic parameters, the propagation and attenuation field of blasting vibration in the slope rock mass is calculated; The propagation and attenuation fields are transformed into gridded dynamic perturbation factors and mapped to the underlying model.
[0010] Based on the above technical solutions, preferably, the setting of the data fusion model, using the basic model and dynamic parameter field as input, yields a four-dimensional spatiotemporal fusion risk field, including: Based on the basic model, a static geological risk field is generated; Based on the dynamic parametric field, a dynamic disturbance field is generated, and displacement time series data is extracted; The static geological risk field, dynamic disturbance field, and displacement time series data are input into the pre-trained data fusion model, and the risk intensity value at any location within the slope at any time is output to form a four-dimensional spatiotemporal fusion risk field.
[0011] Based on the above technical solutions, preferably, the slope risk early warning based on the overall evolution of the spatiotemporal fusion risk field includes: Set multiple warning levels and configure corresponding warning criteria for each warning level; Real-time calculation and monitoring of the evolution of the four-dimensional spatiotemporal fusion risk field; The evolving situation is matched with the aforementioned early warning criteria, and when the criteria for a specific early warning level are met, an early warning signal for that level is triggered.
[0012] More preferably, the early warning criterion is set based on the evolutionary characteristics of the four-dimensional spatiotemporal fusion risk field, and the evolutionary characteristics include at least one of the following: The spatial volume of high-risk areas and its rate of change over time; The rate of change of the risk field strength value over time; The gradient distribution of the risk field in space and the changes in the gradient distribution; Synergy of risk field evolution timelines among different spatial units; The degree of spatial connectivity in high-risk areas.
[0013] Based on the above technical solutions, preferably, the data fusion model is a spatiotemporal convolutional neural network model.
[0014] On the other hand, the present invention provides a mine slope risk assessment system based on spatiotemporal fusion of multi-source data, which incorporates the above-mentioned mine slope risk assessment method based on spatiotemporal fusion of multi-source data.
[0015] The mine slope risk assessment method and system based on multi-source data spatiotemporal fusion of the present invention has the following advantages over the prior art: 1. By constructing a basic model representing the three-dimensional spatial structure of the slope, a unified spatiotemporal bearing base is provided for all heterogeneous monitoring data, enabling previously isolated data such as GNSS displacement, deep displacement, blasting vibration, and groundwater level to be integrated in a unified coordinate system. This fundamentally solves the problem of data fragmentation in existing technologies and forms a holistic and integrated expression of slope engineering and geological conditions. 2. A four-dimensional spatiotemporal fusion risk field is established. This risk field can quantitatively characterize the risk intensity of any point within the slope at any time, upgrading risk assessment from a passive and delayed judgment of a few discrete monitoring points to an active and continuous perception of the spatiotemporal evolution of slope risk. This makes early warning no longer dependent on the static threshold of a single indicator, but based on the dynamic evolution of the entire field risk, thus enabling earlier and more accurate detection of instability precursors. 3. By conducting early warning based on the overall evolution of the risk field through spatiotemporal fusion, it is possible to comprehensively analyze various characteristics of the risk field, such as spatial expansion, intensity growth, gradient changes, and collaborative evolution, thereby intelligently triggering early warning signals of different levels. This overcomes the shortcomings of existing early warning models, which are one-sided and lagging, and realizes a fundamental transformation from simple over-limit alarms to the classification and forward-looking assessment of risk development trends. This provides managers with longer decision response time and more intuitive decision support. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a schematic diagram of the steps in the mine slope risk assessment method based on multi-source data spatiotemporal fusion of the present invention; Figure 2 This diagram illustrates the construction of a three-dimensional geological-engineering coupled base model for the mine slope risk assessment method based on multi-source data spatiotemporal fusion according to the present invention. Figure 3 This is a schematic diagram illustrating the spatiotemporal alignment and fusion of multi-source monitoring data in the mine slope risk assessment method based on spatiotemporal fusion of multi-source data according to the present invention. Figure 4 This is an example diagram showing the visualization output of early warning information for the mine slope risk assessment method based on multi-source data spatiotemporal fusion according to the present invention. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0019] In the description of the embodiments of the present invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of the present invention based on the specific circumstances.
[0020] In the description of the embodiments of the present invention, it should be noted that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of the present invention.
[0021] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.
[0022] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.
[0023] The following disclosure provides numerous different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. Additionally, examples of various specific processes and materials are provided in this invention; however, those skilled in the art will recognize the applicability of other processes and / or the use of other materials.
[0024] like Figure 1-4 As shown, the present invention provides a mine slope risk assessment method based on spatiotemporal fusion of multi-source data. This method aims to solve the problems of difficulty in effectively integrating multi-source heterogeneous monitoring data, lagging and one-sided risk assessment models, and unintuitive early warning information, thereby achieving early, accurate, visualized, and dynamic early warning of slope instability risk.
[0025] Obtain geological and engineering data, and construct a basic model representing the three-dimensional spatial structure of the slope.
[0026] First, it is necessary to construct a basic model that can accurately represent the three-dimensional spatial structure of open-pit mine slopes, which can be called a three-dimensional geological-engineering coupled base model. This model is the unified spatiotemporal base and carrier for all subsequent data fusion, risk calculation and visualization.
[0027] In some specific implementations, it is necessary to first establish a three-dimensional geological model and a three-dimensional surface model, and then couple the three-dimensional geological model and the three-dimensional surface model to realize the construction of the basic model.
[0028] Acquire geological exploration data to generate a three-dimensional geological model containing rock mass mechanical parameters and fault structure planes. This step aims to reconstruct the geological structure inside the slope. Its core input data is geological exploration data, including but not limited to borehole core records, geological profile maps, and geophysical exploration results. These data reveal key geological information such as rock strata distribution, lithological changes, faults, and joints.
[0029] Implicit geological modeling algorithms can be used to generate three-dimensional geological models. Unlike traditional explicit modeling based on manual profiles, implicit modeling represents geological interfaces, such as rock strata interfaces and fault planes, as scalar field functions in three-dimensional space. By interpolating discrete exploration data points, this function is fitted, thereby automatically generating a smooth, continuous, and geologically consistent three-dimensional solid model.
[0030] Specifically, given a set of exploration data points {P_i (x_i, y_i, z_i, a_i)}, where (x_i, y_i, z_i) are spatial coordinates and a_i is the attribute of the point, the implicit modeling algorithm finds a function F(x, y, z) such that the function satisfies the constraints at the known data points. For example, for points belonging to the same rock layer, the function value approaches a constant. The algorithm then predicts the function value in the entire three-dimensional space through spatial interpolation and finally generates the boundaries of each geological body by extracting the isosurface of the function F(x, y, z).
[0031] The final generated three-dimensional geological model not only includes geometric shape, but also serves as an attribute model. Each unit or voxel is associated with rock mass mechanical parameters assigned according to lithology, such as cohesion, internal friction angle, and elastic modulus. At the same time, it clearly characterizes the spatial distribution of structural planes such as faults and weak interlayers.
[0032] This step involves acquiring 3D surface data to generate a 3D surface model containing step geometry information. The aim is to accurately obtain the latest geometry of the slope surface, particularly for step-like engineering structures formed by mining activities. The core input data is 3D surface data, primarily acquired through oblique photogrammetry or 3D laser scanning.
[0033] Specifically, aerial photography can be conducted using drones equipped with oblique photography cameras to acquire multi-view, highly overlapping images. These images can then be processed through photogrammetry to generate high-precision 3D models and dense point clouds of the real scene. Alternatively, high-precision point cloud data of the ground surface can be directly acquired using a 3D laser scanner.
[0034] Subsequently, the acquired raw point cloud data is processed, including preprocessing such as denoising, registration, and thinning. Then, feature extraction algorithms, such as segmentation algorithms based on slope angle changes and region growing algorithms, are used to identify and extract the key geometric information of the steps. The key geometric information includes the step slope angle, step width, platform elevation, etc. Based on the processed point cloud, a detailed three-dimensional surface model is generated through surface reconstruction algorithms. This model not only realistically reflects the surface topography, but its triangular patches or mesh vertices are also associated with the geometric attribute information extracted from the point cloud.
[0035] The 3D geological model is coupled with the 3D surface model to form a 3D geological-engineering coupled base model, which serves as the foundation model. In this step, a unified 3D geological-engineering coupled base model is ultimately generated. This model contains precise geological structures and mechanical properties internally, while its external surface is wrapped with precise surface morphology rich in engineering information, becoming a geological-engineering integrated digital twin that provides a reliable spatial framework for all subsequent analyses.
[0036] Specifically, the first step is to use the three-dimensional surface model as a geometric constraint boundary to correct and fit the morphology of the rock layer interfaces and fault structures in the near-surface part of the three-dimensional geological model, so that the surface geometry of the geological model is consistent with the three-dimensional surface model.
[0037] In this step, a high-precision 3D surface model is used as the geometric constraint boundary to correct the morphology of the 3D geological model in the near-surface portion. Furthermore, the original surface triangulation in the geological model is replaced with a precise triangulation from the 3D surface model. For the exposed portions of rock strata interfaces and fault structures in the geological model near the surface, spatial interpolation algorithms, such as constrained Delaunay triangulation, are used to fit and correct their morphology to the geometry of the 3D surface model, ensuring that the geological interfaces at their exposed locations perfectly match the actual terrain.
[0038] Then, the geometric information of the steps extracted from the 3D surface model is used as engineering attribute data, associated with and assigned to the corresponding spatial location in the corrected 3D geological model.
[0039] In this step, the geometric information of the steps extracted from the 3D surface model is used as engineering attribute data. Through spatial location mapping, this information is associated with and assigned to the corresponding spatial location in the geometrically corrected 3D geological model. For example, all geological model meshes located on the same step platform are assigned the platform's elevation and step width values.
[0040] Finally, a three-dimensional geological-engineering coupled base model is generated by performing geometric corrections and attribute associations.
[0041] To achieve comprehensive perception of slope conditions, multiple sensors need to be deployed on the mine slope in the field, and multi-source heterogeneous monitoring data needs to be collected. This multi-source heterogeneous monitoring data needs to be aligned and fused with the basic model in the spatiotemporal dimensions to generate a continuous dynamic parametric field. This step is crucial for achieving information fusion from data silos. Its core is to integrate the various heterogeneous monitoring data collected on the slope into a unified three-dimensional geological-engineering coupled base model, and to process this discrete point data into spatially and temporally continuous field data using advanced algorithms, providing standardized dynamic input for subsequent risk field calculations.
[0042] Specifically, firstly, multi-source heterogeneous monitoring data is collected, which includes at least GNSS surface displacement data, deep displacement data, blasting vibration data, and groundwater level data.
[0043] Among these methods, GNSS surface displacement data is obtained by deploying GNSS receiver arrays on the slope surface to acquire high-frequency, high-precision three-dimensional coordinate time series of each monitoring point, which are then processed to obtain surface displacement vectors with millimeter-level accuracy. Deep displacement data is obtained through periodic measurements using fixed inclinometers installed in boreholes or portable inclinometers, capturing horizontal displacement at different depths within the slope to reveal the formation and development of potential deep sliding surfaces. Blasting vibration data is obtained by deploying digital vibration meters on the slope and key structural locations near the blasting area to fully record the time history curves of rock mass vibration velocity / acceleration caused by each blasting operation. Groundwater level data is obtained through continuous monitoring of dynamic changes in pore water pressure or elevation within the slope using automatic water level gauges or piezometers buried in monitoring wells.
[0044] Then, the multi-source heterogeneous monitoring data are uniformly mapped to the spatiotemporal coordinate system of the base model, completing spatiotemporal alignment. Due to the inherent differences in spatial location, sampling frequency, data format, and time reference among various sensors, strict spatiotemporal alignment is necessary to unify them within the framework of the base model. To this end, engineering surveying techniques can be used to accurately determine the three-dimensional coordinates of all monitoring points, such as the GNSS antenna phase center, inclinometer nozzle coordinates, and water level gauge probe elevation, in the unified coordinate system used by the three-dimensional geological-engineering coupled base model. The timestamps of all sensors are then unified to Coordinated Universal Time (UTC) or a specified system time through network synchronization or post-correction. For non-uniform time series data, interpolation methods are used to normalize them onto a unified standard time grid.
[0045] Finally, the spatiotemporally aligned multi-source heterogeneous monitoring data is transformed into a spatially and temporally continuous dynamic parameter field through a spatiotemporal interpolation algorithm. The aligned data remains spatially discrete. To obtain continuous state information at any location within the slope, a specific embodiment employs a spatiotemporal interpolation algorithm for field processing. Taking spatiotemporal kriging as an example, it combines spatial interpolation with time series prediction.
[0046] For a point to be estimated parameter values Its estimated value is derived from n surrounding known data points P. i (x i ,y i ,z i ,t i The value of Z(P) i Linear weighted calculation yields:
[0047] Among them, Z(P) i () is a known data point P i (x i ,yi ,z i ,t i The measured value of λ i These are the weighting coefficients obtained by solving the Kriging equations, where n is the number of known points used for interpolation. The weighting coefficient λ... i The Kriging equations, based on a spatiotemporal variogram model, are used to determine the data. This variogram simultaneously characterizes the spatial structure and temporal autocorrelation of the data. Using this method, discrete GNSS displacement, deep displacement, and water level data can be generated into continuous displacement fields U(x,y,z,t) and pore water pressure fields P covering the entire slope, respectively. w (x,y,z,t).
[0048] It should be noted that blasting vibration is a transient and high-energy dynamic disturbance that cannot be directly integrated with the static field. It requires special processing to transform it into a parameter that can be integrated with the static field.
[0049] First, for the blasting vibration data, peak vibration velocity and dominant frequency characteristic parameters are extracted. Spectral analysis is performed on the vibration velocity time-history curve to extract two core characteristic parameters: peak vibration velocity and dominant frequency. Peak vibration velocity is the maximum amplitude in the waveform, directly reflecting the vibration intensity, while dominant frequency represents the frequency component where vibration energy is most concentrated.
[0050] Based on the aforementioned characteristic parameters, the propagation and attenuation field of blasting vibration in the slope rock mass is calculated. Based on the above characteristic parameters, an empirical model is used to calculate the propagation and attenuation of the vibration wave in the rock mass. The peak vibration velocity (PPV) (r) prediction model at a distance r meters from the blast source is as follows:
[0051] Where Q is the maximum charge in a single segment, in kilograms, and k and α are attenuation coefficients related to site conditions.
[0052] Using the above formula, the peak vibration velocity field PPV(x, y, z) generated by each blast in the entire slope rock mass can be simulated.
[0053] Finally, the propagation and attenuation fields are transformed into gridded dynamic perturbation factors and mapped to the underlying model. In this step, to unify dimensions, the PPV field is normalized into a dimensionless dynamic perturbation factor field D_v(x, y, z), defined as:
[0054] Among them, PPV critical To determine the critical safe vibration velocity, the gridded dynamic perturbation factor field D is finally obtained. v (x,y,z).
[0055] The calculated gridded D v The field is mapped to the corresponding spatial location of the basic model, serving as a key dynamic perturbation input parameter field.
[0056] After obtaining the dynamic parameter field, a four-dimensional spatiotemporal fusion risk field is obtained by using a preset data fusion model with the basic model and the dynamic parameter field as input. This four-dimensional spatiotemporal fusion risk field is used to characterize the risk intensity at any point within the slope at any time. This step deeply integrates the previously constructed basic model with the dynamic parameter field, and generates a four-dimensional spatiotemporal fusion risk field through an intelligent algorithm that can quantitatively characterize the risk intensity at any spatial location within the slope at any time point. This risk field R total (x,y,z,t) breaks through the limitations of traditional discrete point monitoring and single index criteria, and realizes spatiotemporal continuous and dynamic quantitative perception of slope stability, providing a direct and comprehensive data foundation for accurate early warning.
[0057] Based on the basic model, a static geological risk field is generated. The static geological risk field R... g (x, y, z) represents the background risk distribution determined by the inherent geological and engineering geometry of the slope, which does not change over time or only undergoes phased changes after the excavation. The construction process is as follows: First, based on a three-dimensional geological-engineering coupled base model, static attribute parameters for each computational unit are extracted, mainly including: rock mass mechanical parameters, degree of structural surface development, and slope geometry. Then, a multi-factor weighted comprehensive evaluation model is established to calculate the static risk intensity value. This model can be expressed as:
[0058] Where f1, f2, and f3 are normalization and quantization functions for different factors, and w1, w2, and w3 are weighting coefficients assigned according to the degree of influence of each factor on slope stability. The final generated R... g The field directly reveals the potential weak areas within the slope due to inherent geological deficiencies, even under conditions of no dynamic disturbance.
[0059] Based on the dynamic parametric field, a dynamic disturbance field is generated, and displacement time-series data is extracted. Dynamic disturbance field R d (x,y,z,t) represents the dynamic risk component caused by external time-varying factors, which changes continuously with time. The dynamic disturbance field is calculated by fusing multiple dynamic parameter fields. The displacement acceleration field and dynamic disturbance factor field D generated above are used as examples. v Taking groundwater level variation field and rainfall infiltration intensity field as inputs, these parameter fields are first normalized to eliminate the influence of dimensions, and then integrated through linear or nonlinear combination models, which can be expressed as:
[0060] in, Let ΔP be the displacement acceleration. w I is the rate of change of pore water pressure. rain Let be the rainfall infiltration intensity, and a, b, c, d be the coupling coefficients reflecting the contribution of each dynamic factor.
[0061] Meanwhile, the original displacement time series data U(t) of key monitoring points are extracted from the fused dynamic parameter field. The original sequence data without interpolation contains detailed time-domain characteristics of slope deformation evolution, which is crucial for the data fusion model to learn the time-series evolution law of risk.
[0062] The static geological risk field, dynamic disturbance field, and displacement time series data are input into the pre-trained data fusion model, and the risk intensity value at any location within the slope at any time is output to form a four-dimensional spatiotemporal fusion risk field.
[0063] In one specific embodiment, a spatiotemporal convolutional neural network model is used as the data fusion model to achieve deep feature extraction and fusion of multi-source information. The input of the data fusion model includes the static geological risk field R. g Dynamic disturbance field R d (t) and displacement time series data U(t), where the static geological risk field is used as the spatial background, the dynamic disturbance field is used as the spatiotemporal disturbance, and the displacement time series data is used as the detailed time series features.
[0064] In terms of specific model structure, it includes a spatial encoding module, a temporal encoding module, and a feature fusion and decoding module. Among them, the spatial encoding module uses three-dimensional convolutional layers to extract R. g and R d To address the complex spatial structure features and spatial distribution patterns in the data, the temporal coding module uses one-dimensional convolutional layers to extract data from U(t) and R. d The temporal dynamic features of deformation evolution are extracted from the time series of (t). The feature fusion and decoding module splices and fuses the extracted spatial and temporal features, and then outputs a risk intensity scalar value of each unit at a specific time t through a fully connected layer or a deconvolution layer.
[0065] After obtaining the four-dimensional spatiotemporal fusion risk field, slope risk early warning can be performed based on the overall evolution of the spatiotemporal fusion risk field. This step can be based on the four-dimensional spatiotemporal fusion risk field R calculated above. totalBy analyzing the overall evolution of (x, y, z, t), intelligent slope risk management is achieved. This early warning method abandons the traditional static and lagging model based on threshold criteria for a single monitoring point. Instead, it achieves forward-looking, multi-level hierarchical early warning through in-depth analysis and intelligent identification of the evolutionary characteristics of the risk field in the spatiotemporal dimensions. Its core lies in the fact that the triggering of the early warning no longer depends on the data exceeding limits at a single isolated point, but is based on the changes in the spatial organization of the risk field, its temporal evolutionary trend, and the synergistic effects between its internal units. This allows for earlier and more accurate detection of precursory signals of slope instability.
[0066] Multiple warning levels are set up, and corresponding warning criteria are configured for each level. To achieve precise and differentiated risk response, this invention establishes a multi-level warning system from low to high, typically divided into four levels: Attention Level, Warning Level, Alert Level, and Alarm Level. Each warning level is configured with corresponding quantified warning criteria. These criteria are not based on the absolute values of a single indicator such as displacement or rate, but are set entirely according to the evolutionary characteristics of the four-dimensional spatiotemporal fusion risk field. Key evolutionary characteristic parameters include: the spatial volume of high-risk areas and their expansion rate; the time change rate of risk field intensity values and their higher-order derivatives; the gradient distribution and evolution of the risk field in space to identify stress concentration areas; the synergy index obtained by calculating the correlation coefficient of the time series curves of risk fields of different spatial units; and the degree of spatial connectivity of high-risk areas assessed based on morphological algorithms or graph theory analysis. By setting different levels of thresholds for these characteristic parameters, a multi-dimensional and dynamic early warning criterion library is formed. On the actual visual monitoring interface, different early warning levels can be alarmed in different forms. For example, by distinguishing by color, the four levels of alarms can be set to four colors: blue, yellow, orange, and red.
[0067] The system calculates and monitors the evolution of the four-dimensional spatiotemporal fusion risk field in real time. It performs real-time calculations and tracking analysis on the newly generated four-dimensional spatiotemporal fusion risk field. This step not only displays a current snapshot of the risk field but also quantitatively tracks its dynamic evolution process. The system automatically calculates and records the time series of various defined evolutionary characteristic parameters. For example, it tracks the volume change curve of high-risk areas in real time, calculates the average risk index and its rate of change across the entire field, and continuously analyzes the collaborative deformation trends among sub-regions within the risk field. This real-time monitoring constitutes a continuous sensory nervous system for the slope system, providing a real-time dynamic data stream for subsequent criterion matching.
[0068] The evolving situation is matched with the aforementioned early warning criteria. When the criteria for a specific early warning level are met, an early warning signal for that level is triggered. In this step, various evolutionary characteristic parameters obtained from real-time monitoring are automatically and rapidly matched and logically judged with pre-configured multi-level early warning criteria. For example, when the system identifies that the average intensity of the risk field in a certain local area exceeds the basic threshold for the first time, and its risk evolution rate changes from stable to positive growth and exceeds the threshold, it is determined that the conditions for a concern-level early warning are met, and the system will automatically trigger an early warning signal for that level. If this area subsequently shows further spatial expansion or a continuous increase in the risk gradient, the warning level is upgraded to an alert level. This process continues. Once the monitored evolutionary situation matches a higher-level criterion, the corresponding orange or red alarm is immediately triggered. The triggered early warning signal, along with the specific risk location, level, and evolutionary trend analysis, is encapsulated and pushed to the visualization terminal and the mobile devices of the responsible personnel to drive the emergency response process.
[0069] The mine slope risk assessment system based on multi-source data spatiotemporal fusion of the present invention incorporates the above-mentioned mine slope risk assessment method based on multi-source data spatiotemporal fusion.
[0070] The system adopts a hierarchical architecture, consisting of a data acquisition module, a data processing and fusion module, a model calculation and risk assessment module, and an application and visualization module, from bottom to top. The data acquisition module comprises a network of physical sensors deployed on the mine slope, responsible for sensing and collecting raw data. The data processing and fusion module, deployed on edge computing nodes or a central server, is responsible for cleaning, aligning, and fusing multi-source heterogeneous data. The model calculation and risk assessment module, as the core of the system, carries core algorithms such as 3D geological modeling, dynamic parametric field calculation, and spatiotemporal fusion risk field generation. The application and visualization module provides users with 3D visualization of the risk field, early warning information push, and decision support through a human-computer interaction interface.
[0071] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for risk assessment of mine slopes based on spatiotemporal fusion of multi-source data, characterized in that, include: Acquire geological and engineering data, and construct a basic model representing the three-dimensional spatial structure of the slope; Multi-source heterogeneous monitoring data were collected on-site at the mine slope, and the multi-source heterogeneous monitoring data was aligned and fused with the basic model in the spatiotemporal dimension to generate a continuous dynamic parametric field. A data fusion model is set up, and a four-dimensional spatiotemporal fusion risk field is obtained by taking the basic model and dynamic parameter field as input. The four-dimensional spatiotemporal fusion risk field is used to characterize the risk intensity of any point in the slope body at any time. Based on the overall evolution of the spatiotemporal integrated risk field, slope risk early warning is carried out.
2. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 1, characterized in that, The acquisition of geological and engineering data, and the construction of a basic model characterizing the three-dimensional spatial structure of the slope, include: Acquire geological exploration data and generate a three-dimensional geological model that includes rock mass mechanical parameters and fault structure planes; Acquire 3D surface data and generate a 3D surface model containing step geometry information; The three-dimensional geological model is coupled with the three-dimensional surface model to form a three-dimensional geological-engineering coupled base model, which serves as the basic model.
3. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 2, characterized in that, The coupling of the three-dimensional geological model with the three-dimensional surface model includes the following steps: Using the three-dimensional surface model as a geometric constraint boundary, the morphology of rock layer interfaces and fault structure surfaces in the near-surface part of the three-dimensional geological model is corrected and fitted, so that the surface geometry of the geological model is consistent with the three-dimensional surface model. The geometric information of the steps extracted from the 3D surface model is used as engineering attribute data, and then associated and assigned to the corresponding spatial location in the corrected 3D geological model. A three-dimensional geological-engineering coupled base model is generated by performing geometric corrections and attribute associations.
4. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 1, characterized in that, The process of collecting multi-source heterogeneous monitoring data on the mine slope, aligning and fusing this data with the basic model in the spatiotemporal dimensions, and generating a continuous dynamic parametric field includes: Collect multi-source heterogeneous monitoring data, which includes at least GNSS surface displacement data, deep displacement data, blasting vibration data, and groundwater level data; Multi-source heterogeneous monitoring data are uniformly mapped to the spatiotemporal coordinate system of the basic model to complete spatiotemporal alignment; The spatiotemporally aligned multi-source heterogeneous monitoring data is transformed into a spatially and temporally continuous dynamic parametric field by using a spatiotemporal interpolation algorithm.
5. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 4, characterized in that, The process of uniformly mapping multi-source heterogeneous monitoring data to the spatiotemporal coordinate system of the basic model includes: For blasting vibration data, extract peak vibration velocity and dominant frequency characteristic parameters; Based on the aforementioned characteristic parameters, the propagation and attenuation field of blasting vibration in the slope rock mass is calculated; The propagation and attenuation fields are transformed into gridded dynamic perturbation factors and mapped to the underlying model.
6. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 1, characterized in that, The data fusion model is set up by taking the basic model and dynamic parameter field as inputs to obtain a four-dimensional spatiotemporal fusion risk field, including: Based on the basic model, a static geological risk field is generated; Based on the dynamic parametric field, a dynamic disturbance field is generated, and displacement time series data is extracted; The static geological risk field, dynamic disturbance field, and displacement time series data are input into the pre-trained data fusion model, and the risk intensity value at any location within the slope at any time is output to form a four-dimensional spatiotemporal fusion risk field.
7. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 1, characterized in that, The slope risk early warning is based on the overall evolution of the spatiotemporal fusion risk field, including: Set multiple warning levels and configure corresponding warning criteria for each warning level; Real-time calculation and monitoring of the evolution of the four-dimensional spatiotemporal fusion risk field; The evolving situation is matched with the aforementioned early warning criteria, and when the criteria for a specific early warning level are met, an early warning signal for that level is triggered.
8. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 7, characterized in that, The early warning criterion is set based on the evolutionary characteristics of the four-dimensional spatiotemporal fusion risk field, and the evolutionary characteristics include at least one of the following: The spatial volume of high-risk areas and its rate of change over time; The rate of change of the risk field strength value over time; The gradient distribution of the risk field in space and the changes in the gradient distribution; Synergy of risk field evolution timelines among different spatial units; The degree of spatial connectivity in high-risk areas.
9. The mine slope risk assessment method based on multi-source data spatiotemporal fusion as described in claim 1, characterized in that, The data fusion model is a spatiotemporal convolutional neural network model.
10. A mine slope risk assessment system based on multi-source data spatiotemporal fusion, characterized in that, The mine slope risk assessment method based on spatiotemporal fusion of multi-source data, as described in any one of claims 1-9, is embedded.
Citation Information
Cited By
Regional geological three-dimensional model construction method based on multi-source data fusion
CN121837532A
Interior and exterior deformation monitoring method integrating Beidou GNSS and inclinometer
CN121977482A
A gently inclined broken ore body stope stability analysis method based on real-time monitoring
CN122222398A
