A risk zone determination method

Abstract

This application discloses a method for determining risk areas. This method can obtain deformation sequences corresponding to multiple deformation sampling points within a risk-concerned area; analyze the deformation parameters corresponding to the deformation data in the deformation sequences across multiple deformation assessment dimensions; fuse the deformation parameters corresponding to the deformation sampling points under each deformation assessment dimension to obtain and, based on the deformation anomaly index corresponding to the deformation sampling points, determine candidate anomaly sampling points from the multiple deformation sampling points; determine adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point from the candidate candidate anomaly sampling points to determine candidate risk areas from the risk-concerned area; based on the deformation sequences corresponding to each deformation sampling point in the candidate risk areas, determine and filter the candidate risk areas according to the regional deformation trend corresponding to the candidate risk areas to obtain the target risk area. Therefore, this application improves the accuracy and reliability of risk area identification.

Description

Technical Field

[0001] This application relates to the fields of spatial computing and digital twin technology, specifically to a method for determining risk areas. Background Technology

[0002] Risk areas refer to specific spatial ranges exhibiting abnormal deformation characteristics or potential structural instability risks during tasks such as geological monitoring, infrastructure maintenance, and disaster early warning. Accurate delineation and real-time monitoring of risk areas are core prerequisites for ensuring the safety of major projects, preventing natural disasters, and achieving precise digital governance of the geological environment. They play an irreplaceable role in decision-making support in fields such as transportation, energy security, and smart city construction.

[0003] Current risk area identification typically relies on manual on-site surveys or global visual interpretation based on macroscopic remote sensing imagery. However, this approach focuses on static monitoring of surface morphology, primarily depending on the assessment of single time sections or isolated deformation dimensions. It struggles to effectively capture the nonlinear evolutionary characteristics of deformation over time and its spatial co-evolutionary attributes. This results in significant technical deficiencies in the final risk area delineation when dealing with complex and dynamically evolving scenarios, such as severe identification lags and positioning errors. Consequently, it fails to meet the high-performance requirements for early, accurate, and refined management of risk events. Summary of the Invention

[0004] This application provides a method for determining risk areas, which can capture the deformation process of the area, thus facilitating the accurate and timely discovery of risk areas and improving the accuracy and reliability of risk area identification.

[0005] This application provides a method for determining risk areas, the method including:

[0006] Obtain deformation sequences corresponding to multiple deformation sampling points within the risk concern area. The deformation sequences include deformation data of the corresponding deformation sampling points at multiple consecutive collection time points.

[0007] For each deformation sampling point, the deformation data in the corresponding deformation sequence is analyzed and processed according to multiple deformation evaluation dimensions to obtain the deformation parameters corresponding to each deformation evaluation dimension;

[0008] Based on the distribution of deformation parameters of multiple deformation sampling points under each deformation evaluation dimension, the deformation parameters corresponding to the deformation sampling points under each deformation evaluation dimension are fused to obtain the deformation anomaly index corresponding to the deformation sampling points.

[0009] Based on the deformation anomaly index corresponding to each deformation sampling point, candidate anomaly sampling points are determined from multiple deformation sampling points;

[0010] Based on the target candidate anomaly sampling point, the adjacent candidate anomaly sampling point corresponding to the target candidate anomaly sampling point is determined from the candidate candidate anomaly sampling points. Based on the target candidate anomaly sampling point and the adjacent candidate candidate anomaly sampling point, the candidate risk area is determined from the risk concern area. The target candidate anomaly sampling point is any candidate anomaly sampling point, and the candidate candidate anomaly sampling point is the candidate anomaly sampling point other than the target candidate anomaly sampling point.

[0011] Based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region, the regional deformation trend corresponding to the candidate risk region is determined, and the candidate risk region is screened according to the regional deformation trend to obtain the target risk region.

[0012] This application also provides a risk area determination system, the system comprising:

[0013] The acquisition unit is used to acquire the deformation sequence corresponding to multiple deformation sampling points within the risk concern area. The deformation sequence includes the deformation data of the corresponding deformation sampling points at multiple consecutive collection time points.

[0014] The analysis unit is used to analyze and process the deformation data in the corresponding deformation sequence for each deformation sampling point in multiple deformation evaluation dimensions, and obtain the deformation parameters corresponding to each deformation evaluation dimension.

[0015] The fusion unit is used to fuse the deformation parameters corresponding to the deformation sampling points under each deformation evaluation dimension based on the distribution of deformation parameters of multiple deformation sampling points under each deformation evaluation dimension, so as to obtain the deformation anomaly index corresponding to the deformation sampling points.

[0016] The anomaly determination unit is used to determine candidate anomaly sampling points from multiple deformation sampling points based on the deformation anomaly index corresponding to each deformation sampling point.

[0017] The region determination unit is used to determine the adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point from the candidate candidate anomaly sampling points based on the target candidate anomaly sampling point, and to determine the candidate risk region from the risk concern region based on the target candidate anomaly sampling point and the adjacent candidate candidate anomaly sampling points. The target candidate anomaly sampling point is any candidate anomaly sampling point, and the candidate candidate anomaly sampling points are candidate anomaly sampling points other than the target candidate anomaly sampling point.

[0018] The region filtering unit is used to determine the regional deformation trend corresponding to the candidate risk region based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region, and to filter the candidate risk region according to the regional deformation trend to obtain the target risk region.

[0019] This application also provides an electronic device, including a processor and a memory, wherein the memory stores multiple instructions; the processor loads instructions from the memory to execute the steps in any of the risk area determination methods provided in this application.

[0020] This application also provides a computer-readable storage medium storing multiple instructions adapted for loading by a processor to execute steps in any of the risk area determination methods provided in this application.

[0021] This application also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps in any of the risk area determination methods provided in this application.

[0022] In this application, by obtaining deformation sequences corresponding to multiple deformation sampling points within the risk-concern area, dynamic deformation data of each deformation sampling point over time can be acquired. This provides a continuous temporal data foundation for subsequent exploration of deformation evolution patterns, effectively compensating for the limitations of static monitoring in terms of missing data in the time dimension. Based on this, for each deformation sampling point, the deformation data in the corresponding deformation sequence is analyzed and processed across multiple deformation assessment dimensions to obtain deformation parameters corresponding to each dimension. Quantitative analysis of deformation data under different dimensions helps to capture deformation evolution attributes that are difficult to characterize with a single deformation assessment dimension, thereby reducing the lag in risk discovery. Subsequently, based on the distribution of deformation parameters of multiple deformation sampling points under each deformation dimension, the deformation parameters corresponding to the deformation sampling points under each deformation dimension are fused. This allows the deformation parameters of the deformation sampling points to be comprehensively evaluated within the overall distribution of deformation parameters, achieving accurate measurement of deformation anomaly indicators and effectively mitigating the positioning bias problem caused by isolated discrimination. Furthermore, based on the deformation anomaly indicators corresponding to each deformation sampling point, candidate anomaly sampling points are determined from multiple deformation sampling points. This allows for the initial identification of targets with anomaly tendencies from a large number of deformation sampling points, effectively eliminating redundant deformation sampling points that do not exhibit anomalous deformation. Further, any candidate anomaly sampling point is designated as a target candidate anomaly sampling point, and all other candidate anomaly sampling points are designated as prospective candidate anomaly sampling points. Then, adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point are determined from the prospective candidate anomaly sampling points. Based on the target candidate anomaly sampling point and adjacent candidate anomaly sampling points, candidate risk areas are determined from the risk concern area. This fully leverages the spatial adjacency relationships between candidate anomaly sampling points, achieving the aggregation of anomaly features from points to regions. Consequently, it can accurately identify spatially correlated deformation clusters, enhancing spatial coordination in the identification process. Finally, based on the deformation sequences corresponding to each deformation sampling point in the candidate risk areas, the regional deformation trends corresponding to the candidate risk areas are determined. The candidate risk areas are then filtered according to these trends to obtain the target risk areas. This achieves a secondary screening based on the overall regional evolution law. By eliminating areas whose deformation trends do not conform to expectations, the consistency between the risk identification results and the physical evolution logic is ensured, effectively avoiding identification accuracy deviations caused by local abnormal fluctuations. Therefore, the deformation process of a region can be captured, which is beneficial for accurately and promptly identifying risk areas and improving the accuracy and reliability of risk area identification. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1a This is a flowchart illustrating the risk area determination method provided in an embodiment of this application;

[0025] Figure 1b This is a schematic diagram of a key area of ​​concern provided in an embodiment of this application;

[0026] Figure 1c This is another schematic diagram of the key area of ​​concern provided in the embodiments of this application;

[0027] Figure 1d This is a schematic diagram of the temporal deformation features provided in the embodiments of this application;

[0028] Figure 1e This is a temporal anomaly intensity distribution map provided in the embodiments of this application;

[0029] Figure 1f This is a spatial distribution diagram of candidate units provided in the embodiments of this application;

[0030] Figure 1g This is a spatial distribution map of candidate risk areas provided in the embodiments of this application;

[0031] Figure 1h This is the final risk zone and its risk intensity ranking diagram provided in the embodiments of this application. Detailed Implementation

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0033] This application provides a method for determining risk areas.

[0034] Specifically, the risk area determination device can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet, smart Bluetooth device, laptop, or personal computer (PC); the server can be a single server or a server cluster consisting of multiple servers.

[0035] In some embodiments, the risk area determination device may also be integrated into multiple electronic devices, such as multiple servers, with multiple servers implementing the risk area determination method of this application.

[0036] In some embodiments, the server may also be implemented as a terminal.

[0037] The following sections provide detailed descriptions of each example. It should be noted that the sequence numbers of the following embodiments are not intended to limit the preferred order of the embodiments.

[0038] In this embodiment, a method for determining risk areas is provided, such as... Figure 1a As shown, the specific process of this risk area determination method can be as follows:

[0039] S101. Obtain the deformation sequence corresponding to multiple deformation sampling points within the risk concern area. The deformation sequence includes the deformation data corresponding to the deformation sampling points at multiple consecutive collection time points.

[0040] In this context, the risk-concern area refers to a specific geographical area where monitoring is necessary and morphological changes may occur. In this application, the risk-concern area may refer to the geographical space that requires refined deformation calculations after screening and excluding non-critical interference targets. Examples include the 200-meter radius around a foundation pit, the dam body area, and the stress point area of ​​a highway slope.

[0041] Deformation sampling points are reference coordinate points that can maintain stable physical or signal characteristics over a long period of time and are used to accurately capture and quantify changes in spatial position (i.e., deformation). For example, deformation sampling points can be coherent point targets in radar monitoring (such as corner reflectors, stable rock surfaces), sensor array nodes, anchor points in artificial monitoring networks, etc.

[0042] The acquisition time point refers to the independent observation moment at which deformation data for each sampling point is acquired. In this application, the acquisition time point specifically refers to the precise synchronization instant that triggers the deformation data acquisition task. Specific examples include: the sensing transit time of a synthetic aperture radar (SAR) satellite, the timing trigger instant of a ground-based monitoring sensor array (e.g., 12:00:00), and the closing moment when a radar scanning device completes a single target scan. These time points ensure the determinism and consistency of each deformation data point in the deformation sequence during time correlation analysis.

[0043] Multiple consecutive acquisition time points refer to a series of historical observation moments spanning from the past to the present. These moments can be arranged based on a fixed frequency (such as periodic sampling) or a non-fixed frequency (such as event-triggered sampling). The core principle is to fully record the entire process of deformation from its occurrence and development to its current state through continuity along the timeline, forming a time series dimension that characterizes the motion patterns of an object. In the embodiments of this application, multiple consecutive acquisition time points specifically refer to a series of key time nodes used to support the construction of the "deformation sequence." Specific application examples include: daily midnight fixed-point recording times for the past 30 days (fixed frequency), 12 consecutive satellite radar interferometric imaging transit times (orbital period frequency), and automated recording points generated by ground-based sensors every hour, etc.

[0044] Deformation data refers to the displacement vector or scalar of a deformation sampling point relative to a reference position (such as the initial observation position or the previous adjacent observation time) at a specific moment. As the smallest unit of data information, it directly reflects the change in the physical properties of the deformation sampling point at a specific time node. For example, deformation data is specifically represented by various physical measures, including but not limited to settlement values ​​in the vertical direction, displacement in the horizontal coordinate system, and cumulative deformation values ​​superimposed over multiple periods.

[0045] A deformation sequence consists of deformation data from the same deformation sampling point at multiple consecutive acquisition time points. For example, a deformation sequence can be a time series of settlement values, a time series of horizontal displacement, or a time series of cumulative deformation.

[0046] In some embodiments, the deformation sequence includes the geographical location of the deformation sampling point and the deformation data of the deformation sampling point at multiple consecutive sampling time points.

[0047] In some embodiments, through collaborative filtering of image and map data, precise pre-stripping of the monitoring area is achieved, eliminating a large amount of redundant background that does not pose a deformation risk, thus improving the effectiveness of the data from the source. Specifically, before obtaining the deformation sequence corresponding to multiple deformation sampling points within the risk-concern area, the process also includes:

[0048] Obtain the regional image corresponding to the initial risk concern area covered by multiple deformation sampling points, and the map data corresponding to the initial risk concern area;

[0049] Based on preset non-critical objects of interest, the region image is processed to obtain the object region corresponding to the preset non-critical objects of interest.

[0050] Identify key regions of interest from the region image, excluding the object region;

[0051] Local map regions in the graph data that match the key areas of concern are identified as areas of risk concern.

[0052] The initial risk concern area refers to the logical starting point of the monitoring task in the spatial dimension, that is, the raw digital monitoring scope without any screening or processing, which includes all the basic geographic data of the area to be analyzed. For example, the initial risk concern area is the maximum scanning sector set by the slope radar, or the complete rectangular area initially covered by satellite remote sensing imagery. It not only includes the slope surface to be monitored, but also encompasses all the raw data such as the surrounding mountain background, rivers at the foot of the mountain, vegetation, and even distant buildings.

[0053] Regional images refer to optical or radar remote sensing imagery data that corresponds to the initial risk concern area and reflects the physical characteristics of the land surface. For example, visible light remote sensing images taken by high-resolution optical satellites, or radar intensity images generated in real time by slope-based radar, visually present the topographic texture and reflectivity of the monitored area.

[0054] Map data refers to vector data or structured geographic information with geospatial labels corresponding to the initial risk concern area. For example, land use classification maps, digital elevation models (DEMs), or electronic map vector data containing the outlines of roads, waterways, and buildings for the area.

[0055] Pre-defined non-critical objects of interest refer to background objects in the regional image that do not pose a risk of deformation or are unrelated to the analysis target (such as slope instability). Examples include high vegetation cover areas at the top and bottom of the slope (the swaying of trees and grass is considered noise), loose gravel accumulation areas at the bottom of the slope (non-structural subjects), and the sky or distant mountain background at the edge of the monitoring range.

[0056] An object region refers to a specific pixel range or spatial set occupied by a pre-defined non-critical object of interest in a region of image. For example, a pixel mask of vegetation swaying area automatically delineated on radar imagery by an image recognition algorithm, or the set of image coordinates of background noise points identified as "pre-defined non-critical objects of interest".

[0057] The critical area of ​​interest (CIIE) refers to the set of pixels remaining after removing the object region from the regional image, which are the ones that actually need to be analyzed. For example, after removing vegetation and background interference, the remaining areas include exposed slope surfaces, retaining wall surfaces, and manually installed corner reflectors. These areas are the core range for subsequent extraction of deformation sampling points.

[0058] A local map region refers to the actual spatial coordinate range of a key area of ​​interest mapped onto a geographic coordinate system, used to guide subsequent data matching and processing. For example, ... Figure 1b or Figure 1cAs shown, the image pixel coordinates of the selected "key areas of interest" are converted into the actual latitude and longitude or engineering projection coordinate range. The system will then only retrieve or process vector map data that falls within this coordinate range, thus achieving "precise pre-stripping" of the data.

[0059] Understandably, when the initial area of ​​concern for risk is the monitored slope, the first step is to acquire continuous radar data of the slope. Deformation data at each monitoring time point (i.e., the time point of data acquisition). The deformation data at each monitoring time point includes the spatial location of the monitoring unit (the location of the monitoring unit is the deformation sampling point) and the corresponding deformation data for that monitoring time. The monitoring unit can specifically be a coherent point target in radar monitoring, a sensor array node, etc. Assume there are a total of [number missing] monitoring slopes. The monitoring unit, the first The spatial location vector of each monitoring unit is denoted as: .

[0060] No. Each monitoring unit in continuous The deformation sequence at each monitoring time point (i.e., multiple data collection points) is denoted as: .

[0061] The total time series data of the monitored slope can then be represented as:

[0062] Based on this, the area of ​​interest for the monitored slope is delineated, and areas irrelevant to the risk analysis of the monitored slope are excluded, such as areas with high vegetation cover, accumulation zones, and marginal background. The delineated risk focus area is defined as follows: The set of monitoring units within the risk concern area is denoted as The number of monitoring units it contains is denoted as This step obtains time-series data from all monitoring units within the risk-aware area, providing input for subsequent anomaly identification.

[0063] The deformation sequence specifically represents the raw deformation monitoring data corresponding to the slope radar. The table header is: x, y, h, t1-t60, where x, y, and h are the coordinates and elevations (in meters) of the monitoring unit. t1-t60 represent the deformation at different times, and the n value in the aforementioned case is 60.

[0064] S102. For each deformation sampling point, perform analysis and processing on the deformation data in the corresponding deformation sequence using multiple deformation evaluation dimensions to obtain the deformation parameters corresponding to each deformation evaluation dimension.

[0065] Among them, the deformation assessment dimension is used to evaluate different computational angles or analytical perspectives of deformation characteristics. It defines which aspect (such as trend, velocity, acceleration, etc.) to interpret the time series data of the deformation sequence.

[0066] For example, deformation assessment dimensions can include trend-based, dynamic, and statistical types. Trend-based deformation assessment dimensions can specifically include recent mean deformation characteristics (assessing the average subsidence / uplift level over a period of time) and cumulative deformation characteristics. Dynamic deformation assessment dimensions can specifically include deformation growth characteristics (assessing whether deformation is accelerating) and deformation slope change dimensions (assessing the rate of change of deformation rate, i.e., acceleration). Statistical deformation assessment dimensions can specifically include the standard deviation of deformation fluctuations (assessing stability).

[0067] Deformation parameters are specific quantitative values ​​or indicators obtained by mathematical calculations or model fitting of deformation sequences at deformation sampling points under a specific deformation evaluation dimension.

[0068] In some embodiments, the deformation parameters corresponding to the deformation assessment dimension include the recent mean deformation parameter, the deformation growth parameter, and the recent slope parameter.

[0069] In some embodiments, by dividing the deformation sequence into two sub-sequences, a dynamically changing, dedicated benchmark is constructed, effectively eliminating environmental background interference while accurately capturing the inflection point of deformation acceleration. This segmented processing strategy not only reduces the impact of noise through local smoothing but also significantly reduces redundant calculations of the entire dataset, achieving real-time, accurate early warning with strong anti-interference capabilities and high computational efficiency. Specifically, the deformation data in the corresponding deformation sequence is analyzed and processed according to multiple deformation evaluation dimensions to obtain the deformation parameters corresponding to each deformation evaluation dimension, including:

[0070] Based on a preset time interval, the corresponding deformation sequence is divided to obtain a first sub-deformation sequence and a second sub-deformation sequence. The acquisition time of the first sub-deformation sequence is earlier than that of the second sub-deformation sequence.

[0071] The mean value of the deformation data in the second sub-deformation sequence is calculated to obtain the recent mean deformation parameter corresponding to the second sub-deformation sequence.

[0072] The mean value is calculated based on the deformation data in the first sub-deformation sequence to obtain the historical deformation mean parameter corresponding to the first sub-deformation sequence.

[0073] The deformation growth parameter is obtained by determining the data offset of the recent mean deformation parameter relative to the historical mean deformation parameter.

[0074] Linear trend fitting is performed based on the deviation between each deformation data in the second sub-deformation sequence and the recent mean deformation parameter to obtain the recent slope parameter reflecting the deformation enhancement trend.

[0075] The preset time interval is used to divide the deformation sequence corresponding to the deformation sampling point into time window lengths or time nodes of different time period subsequences.

[0076] The first sub-deformation sequence is a set of historical deformation data that precedes a preset time interval.

[0077] For example, if the preset time interval is set to "the last 7 days", the first sub-deformation sequence contains all historical monitoring data up to 7 days ago, representing the baseline state of the monitoring point in the past.

[0078] The second sub-deformation sequence is a set of recent deformation data that are later than or equal to a preset time interval.

[0079] For example, if the preset time interval is set to "the last 7 days", the second sub-deformation sequence contains real-time monitoring data from 7 days ago to the current moment, representing the latest dynamic changes of the monitoring point.

[0080] The recent deformation mean parameter is a quantized value obtained by calculating the arithmetic mean of the deformation data in the second sub-deformation sequence. It is used to characterize the average deformation level of the sampling point within the recent observation window divided by a preset time interval.

[0081] For example, by calculating the data of the second sub-deformation sequence (i.e., within the "last 7 days"), the average displacement rate of the deformation sampling point in the recent period is found to be -1.2 mm / day, which is the recent average deformation parameter.

[0082] The historical deformation mean parameter is a quantified value obtained by calculating the arithmetic mean of the deformation data in the first sub-deformation sequence. It is used to characterize the average deformation level of the sampling point within the historical observation window divided by a preset time interval.

[0083] The data offset is the difference between the recent mean deformation parameter of the second sub-deformation sequence and the historical mean deformation parameter of the first sub-deformation sequence, and is used to quantify the magnitude of change in the deformation baseline.

[0084] For example, if the historical average deformation parameter of the first sub-deformation sequence is -0.8 mm / day and the recent average deformation parameter of the second sub-deformation sequence is -1.2 mm / day, then the data offset is -0.4 mm / day.

[0085] The deformation growth parameter is a parameter determined based on the data offset and is used to describe the degree to which deformation activity increases or decreases in magnitude relative to a historical baseline.

[0086] For example, the calculated data offset of -0.4 mm / day is used as a deformation growth parameter. A negative value indicates that the deformation increases significantly in the settlement direction, meaning that the settlement is intensifying.

[0087] The deviation between each deformation data point in the second sub-deformation sequence and the recent deformation mean parameter is the residual between each specific deformation data point in the second sub-deformation sequence and the mean of the sequence (the recent deformation mean parameter), which is used to reflect the fluctuation of the data points around the mean.

[0088] For example, if the recent average deformation parameter is -1.2 mm / day, and the deformation data for a certain day is -1.5 mm / day, then the deviation at that collection time point is -0.3 mm.

[0089] Linear trend fitting involves using algorithms such as least squares to find the optimal straight line to describe the overall trend of the aforementioned "deviation" over time, thereby quantifying the rate of fluctuation.

[0090] For example, the daily data deviation values ​​in the second sub-deformation sequence are plotted as scatter points on a coordinate system, and a straight line passing through the central trend of these points is calculated. The slope of this straight line is the fitting result.

[0091] The deformation enhancement trend is the dynamic characteristic of deformation activity evolving over time, reflected by the results of linear trend fitting. In particular, it refers to the trend of the degree of deformation deviating from the recent deformation mean parameter intensifying or weakening.

[0092] For example, if the fitted straight line shows an upward sloping trend (positive slope), it indicates that the deformation fluctuations at the deformation sampling points are getting larger and larger, and the degree of deviation from the mean is deepening, which means that the deformation is "accelerating and deteriorating".

[0093] The recent slope parameter is the slope value of the straight line obtained by fitting the linear trend. It is a specific quantitative indicator reflecting the trend of increased deformation and characterizes the acceleration or rate change of deformation.

[0094] For example, calculations show that the trend slope of the above deviation data is +0.08 mm / day². This +0.08 is the recent slope parameter, indicating that the deformation at this monitoring point is accelerating at a rate of 0.08 mm per day.

[0095] In some embodiments, covariant analysis of deformation deviation and time discrete characteristics is performed using the least squares method, which accurately quantifies the linear evolution rate of deformation data and effectively suppresses the interference of short-term random noise, thereby extracting a recent slope parameter that can objectively reflect the trend of deformation acceleration or deceleration. Specifically, based on the deviation between each deformation data in the second sub-deformation sequence and the recent mean deformation parameter, a linear trend fitting is performed to obtain a recent slope parameter reflecting the trend of deformation enhancement, including:

[0096] The average value of the time points corresponding to the collection time points of each deformation data in the second sub-deformation sequence is calculated to obtain the time average value.

[0097] For each deformation data in the second sub-deformation sequence, the difference between the deformation data and the recent mean deformation parameter, and the difference between the collection time point corresponding to the deformation data and the time mean, are multiplied to obtain the covariant term corresponding to the deformation data.

[0098] Summing the covariant terms yields the cumulative covariant characteristic of deformation;

[0099] The time discrete features are obtained by summing the squared differences between the acquisition time points and the time mean of each deformation data in the second sub-deformation sequence.

[0100] The ratio of the cumulative covariant feature of deformation to the discrete feature of time is determined to obtain the recent slope parameter.

[0101] The time mean is the arithmetic mean of all the time points collected in the second sub-deformation sequence, which serves as the central reference point of the time axis.

[0102] For example, if the second sub-deformation sequence contains three time points: Monday 8:00, Tuesday 8:00, and Wednesday 8:00 (denoted as t=1, 2, 3 respectively), then the time mean is .

[0103] The difference between the deformation data in the second sub-deformation sequence and the recent deformation mean parameter is the degree to which each deformation data point deviates from its sequence mean, i.e., the decentralized deformation fluctuation value.

[0104] For example, if the recent average deformation parameter is -1.0 mm, and the data measured on Monday is -1.5 mm, then the difference is... .

[0105] The difference between the collection time point corresponding to the deformation data and the time mean is the degree to which the collection time point corresponding to the deformation data deviates from the time center point, and is used to measure the relative position of the time series.

[0106] For example, based on the example above where the average time is 2, the time difference corresponding to Monday (t=1) is . .

[0107] The covariant term is the product of the deformation fluctuation value and the corresponding time deviation value, reflecting the cooperative relationship between deformation change and time change.

[0108] For example, if the deformation difference at a certain moment is -0.5 mm, and the corresponding time difference is -1, then the covariant term is: .

[0109] The cumulative covariance feature of deformation is the summation of the covariance terms of all data points in the second sub-deformation sequence, which characterizes the overall co-variance trend strength between deformation and time over the entire time period.

[0110] For example, if the covariant terms for the three monitoring points are 0.5, 0.3, and 0.2, respectively, then the cumulative covariant characteristic of deformation is: .

[0111] The difference between the acquisition time point and the time mean of each deformation data in the second sub-deformation sequence is the same as "the difference between the acquisition time point and the time mean of the deformation data", that is, the offset of each moment relative to the time center.

[0112] The summation of squares is performed by squaring all the time differences mentioned above and then summing them up. This is used to eliminate positive and negative cancellation and amplify the degree of dispersion.

[0113] For example, if the time differences are -1, 0, and +1, the result of the sum of squares is: .

[0114] The time discreteness characteristic reflects the degree of dispersion of the acquisition time points on the time axis, and serves as the normalized denominator for slope calculation.

[0115] The ratio of the cumulative covariant feature of deformation to the discrete feature of time is the numerator of the cumulative covariant feature and the denominator of the discrete feature of time.

[0116] For example, if the cumulative covariance characteristic of deformation is 1.0 and the discrete characteristic of time is 2, then the recent slope parameter is: A positive value indicates that the deformation is increasing.

[0117] Understandably, in areas of risk concern Within this framework, the temporal deformation characteristics of each monitoring unit are constructed. The current analysis period corresponding to the deformation sequence is divided into two parts: a pre-deformation window (i.e., a time window composed of the acquisition time points corresponding to the first sub-deformation sequence) and a near-deformation window (i.e., a time window composed of the acquisition time points corresponding to the second sub-deformation sequence). Let the set of times corresponding to the pre-deformation window be denoted as . The set of times corresponding to the near time window is ,and .

[0118] Let the deformation data of the previous time window be The deformation data for the near time window is Then the first The average value of each monitoring unit within the previous time window is The mean value within the recent time window (i.e., the mean parameter of recent deformation) is Further define the monitoring unit The growth characteristics (i.e., deformation growth parameters) are: This feature is used to characterize the overall degree of elevation of the near time window relative to the previous time window.

[0119] To characterize whether the monitoring unit exhibits a sustained enhancement trend within the near-term time window, a linear fit is performed on the time-series curve within the near-term time window. Let the time mean of the near-term time window be... Then the first The slope of change of each monitoring unit within the recent time window (i.e., the recent slope parameter) can be expressed as: .

[0120] Therefore, the temporal deformation characteristics of the monitoring unit include recent deformation mean parameters. Deformation growth parameters Recent slope parameters .

[0121] like Figure 1d As shown, the time-series deformation features of the current risk concern area are calculated in the deformation sequence. The table headers are: x, y, h, recent mean feature (i.e., recent deformation mean parameter) (recent_mean), unit growth feature (i.e., deformation growth parameter) (growth), and recent slope feature (i.e., recent slope parameter) (recent_slope).

[0122] In this calculation, the previous time window is 40 ( ), take 20 in the near time window ( The 60 time periods were divided into 1-40 and 41-60 for calculation.

[0123] S103. Based on the distribution of deformation parameters of multiple deformation sampling points under each deformation evaluation dimension, the deformation parameters corresponding to the deformation sampling points under each deformation evaluation dimension are fused to obtain the deformation anomaly index corresponding to the deformation sampling points.

[0124] The distribution of deformation parameters among multiple deformation sampling points under various deformation assessment dimensions refers to the overall statistical regularity or spatial arrangement characteristics of the deformation parameters of all deformation sampling points within the risk concern area under the same deformation assessment dimension. It reflects the "group state" of all deformation sampling points under that deformation assessment dimension and serves as the macroscopic background for determining whether deformation sampling points are abnormal.

[0125] Fusion processing refers to the process of integrating the deformation parameters of the same deformation sampling point under multiple different deformation evaluation dimensions into a single value that can comprehensively represent the overall deformation state of the deformation sampling point through mathematical methods such as weighted calculation, normalization mapping, or construction of multi-dimensional feature vectors.

[0126] The deformation anomaly index refers to the final quantitative value obtained after fusion processing. It is used to intuitively characterize the degree of abnormality or potential risk level of the deformation activity of a deformation sampling point relative to the risk concern area. The higher the index value, the more the deformation state of that point deviates from the normal range, and the greater the risk of instability.

[0127] In some embodiments, by introducing a central benchmark value reflecting the overall deformation level and discrete scores reflecting the degree of fluctuation, the deformation parameters are normalized and weighted, effectively eliminating dimensional differences and background interference between different deformation assessment dimensions, and realizing standardized quantification and multi-dimensional accurate comprehensive assessment of deformation anomaly intensity. Specifically, based on the distribution of deformation parameters at multiple deformation sampling points under each deformation assessment dimension, the deformation parameters corresponding to each deformation assessment dimension at the deformation sampling points are fused to obtain the deformation anomaly index corresponding to the deformation sampling points, including:

[0128] For each deformation assessment dimension, based on the deformation parameters corresponding to all deformation sampling points in the risk concern area under the deformation assessment dimension, a central benchmark value and discrete scores are determined. The central benchmark value reflects the overall deformation level of multiple deformation sampling points under the deformation assessment dimension, and the discrete scores reflect the degree of fluctuation of the deformation parameters corresponding to each deformation sampling point under the deformation assessment dimension relative to the overall deformation level.

[0129] For each deformation sampling point, determine the degree of deviation of its corresponding deformation parameter from the corresponding central reference value under each deformation evaluation dimension;

[0130] Based on the discrete scores, the corresponding deviation degree is normalized to obtain the sub-anomaly intensity of the deformation sampling point under the corresponding deformation evaluation dimension;

[0131] Based on the preset weights corresponding to each deformation assessment dimension, the sub-anomaly intensities of deformation sampling points in each deformation assessment dimension are weighted and fused to obtain the deformation anomaly index corresponding to the deformation sampling points.

[0132] Among them, the overall deformation level refers to the overall central tendency of deformation parameters of all deformation sampling points in the risk concern area under a certain deformation assessment dimension, which represents the general state of the area under the current dimension.

[0133] The central baseline value is used to represent a specific statistical value (such as the mean or median) of the overall deformation level, serving as a reference value for judging whether a single deformation sampling point is abnormal.

[0134] The degree of fluctuation refers to the discrete or dispersed state of the deformation parameters at each deformation sampling point around the central reference value.

[0135] Discrete scores quantify the degree of fluctuation with specific numerical values ​​(such as standard deviation or interquartile range), reflecting the normal fluctuation range of the data.

[0136] Deviation refers to the absolute difference between the deformation parameter of a single deformation sampling point and the central reference value.

[0137] Normalization involves dividing the degree of deviation by the discrete score to eliminate the influence of dimensions and transform it into a dimensionless relative intensity.

[0138] Sub-anomaly intensity refers to the quantitative characterization value obtained by normalizing the deformation sampling point under a single deformation assessment dimension. Its value is positively correlated with the abnormal deviation of the deformation sampling point under the deformation assessment dimension. The larger the value, the more significant the abnormality of the deformation sampling point under the deformation assessment dimension.

[0139] The preset weights are set based on business experience or model training, and are proportional coefficients that reflect the importance of each deformation assessment dimension.

[0140] In some embodiments, a preset threshold is used to truncate and filter out minute fluctuations, and a stability parameter is used to smooth the discrete scores. This effectively avoids the anomaly of the denominator approaching zero due to extremely small background fluctuations, significantly improving the robustness and computational stability of identifying weak deformation anomalies in a stable environment. Specifically, the deviation of the deformation parameters corresponding to each deformation evaluation dimension from the corresponding central reference value is determined, including:

[0141] Determine the original difference between the deformation parameters of the deformation sampling points and the corresponding central reference values;

[0142] If the absolute value of the original difference is greater than the preset threshold, the original difference is determined as the degree of deviation.

[0143] If the absolute value of the original difference is not greater than the preset threshold, the degree of deviation will be set to zero.

[0144] Based on the discrete scores, the corresponding deviation is normalized to obtain the sub-anomaly intensity of the deformation sampling point under the corresponding deformation assessment dimension, including:

[0145] The discrete scores are summed with the preset stability parameters to obtain the corrected discrete scores;

[0146] By determining the ratio of the degree of deviation to the corrected discrete score, the sub-anomaly intensity of the deformation sampling point under the corresponding deformation evaluation dimension is obtained.

[0147] The original difference refers to the direct calculation result (retaining the positive or negative sign) obtained by subtracting the deformation parameter of the deformation sampling point from its corresponding central reference value under the same deformation assessment dimension. It reflects the initial deviation of the sampling point from the group center in a specific dimension.

[0148] The preset threshold is a critical value used to filter background noise or minor disturbances. When the difference is less than this value, the system considers it a normal environmental fluctuation and does not count it as an anomaly.

[0149] Deviation refers to the original difference between the deformation parameter of a single deformation sampling point and the central reference value.

[0150] The preset stability parameter is a very small positive constant (smoothing term) used to prevent mathematical anomalies such as zero denominator in normalization calculations when the data is extremely stable (discrete scores approach 0).

[0151] The corrected discrete score refers to the result of adding the preset stability parameter to the original discrete score, which is used as the safe denominator for normalization calculation.

[0152] The ratio of the degree of deviation to the corrected discrete score is the formula for calculating the final sub-anomaly intensity. This ratio transforms the truncated degree of deviation into a dimensionless relative intensity.

[0153] Understandably, to unify the dimensions of deformation parameters corresponding to different deformation assessment dimensions and reduce the impact of extreme values ​​on the overall statistical results, robust standardization is performed on the characteristics of all monitoring units within the risk-concern area. Let the value sequence of a certain deformation assessment dimension across all monitoring units be denoted as... Then the deformation assessment dimension is in the th... The positive robust standardized result on each monitoring unit is defined as... ,in, Represents a sequence The median (i.e., the central reference value). Represents a sequence The median absolute deviation (i.e., discrete score). , A small constant (i.e. a preset stability parameter) is used to improve numerical stability. This represents the corrected discrete score. The above definition retains only positive deviations above the overall background median, while deviations below the background median are recorded as 0. By performing positive robust standardization on the recent mean deformation parameter, deformation growth parameter, and recent slope parameter, we obtain: , , Based on this, a monitoring unit is constructed. Temporal anomaly intensity (i.e., the deformation anomaly index corresponding to the deformation sampling point): ,in The preset weights for each deformation assessment dimension satisfy the following conditions. The specific weight values ​​can be adjusted according to different risk concern areas. Calculating the temporal anomaly intensity for all monitoring units within the risk concern area yields the spatial distribution of the temporal anomaly intensity, such as... Figure 1e As shown, the sub-anomaly intensity sequence is obtained. .

[0154] In some embodiments, Take 1e -6 , .

[0155] S104. Based on the deformation anomaly index corresponding to each deformation sampling point, candidate anomaly sampling points are determined from multiple deformation sampling points.

[0156] Among them, candidate anomaly sampling points refer to deformation sampling points whose deformation anomaly indicators reach or exceed the preset warning threshold. The deformation state of these sampling points is judged to be significantly deviating from the normal range in the comprehensive evaluation and has a high potential risk. Therefore, they are selected as the basis for subsequent spatial clustering analysis (to determine risk areas).

[0157] It is understandable that, after obtaining the sub-anomaly intensity sequence of all monitoring units... Then, high-quantile screening is performed to extract candidate anomalous units (i.e., candidate anomalous sampling points) from all monitoring units within the region of interest. Let... Sub-anomaly intensity sequence The The monitoring unit that satisfies the following formula is defined as a candidate anomaly unit: quantile value (i.e., preset warning threshold). .

[0158] in, To preset a high quantile level, it is usually set to 90 or above. The quantile value should be able to distinguish between background and anomalies, while avoiding the influence of noise and fragmentation. When selecting the value, consider the actual situation on site. If the risk area is fragmented, lower Q; if the risk area appears in large areas and does not match the actual risk situation, increase Q. The optimal range for Q value is 85-95.

[0159] Therefore, the set of candidate anomaly units can be represented as .

[0160] This step yields a discrete set of high-anomaly monitoring units, which has not yet formed the final risk zone.

[0161] In some embodiments, the 90th percentile is used as a preset warning threshold to filter candidate abnormal units, such as... Figure 1f As shown, the results of this calculation are stored in a table.

[0162] S105. Based on the target candidate anomaly sampling point, determine the adjacent candidate anomaly sampling point corresponding to the target candidate anomaly sampling point from the candidate candidate anomaly sampling points, and based on the target candidate anomaly sampling point and the adjacent candidate candidate anomaly sampling point, determine the candidate risk area from the risk concern area. The target candidate anomaly sampling point is any candidate anomaly sampling point, and the candidate candidate anomaly sampling point is the candidate anomaly sampling point other than the target candidate anomaly sampling point.

[0163] Among them, adjacent candidate anomaly sampling points refer to other candidate anomaly sampling points that meet the preset proximity conditions with the candidate anomaly sampling point in spatial location. These adjacent candidate anomaly sampling points constitute the direct neighborhood set for spatial correlation analysis of the candidate anomaly sampling point. Other candidate anomaly sampling points are candidate anomaly sampling points other than the current candidate anomaly sampling point.

[0164] A candidate risk area is a geographical region covered by multiple spatially related and clustered candidate anomaly sampling points. It represents a contiguous monitoring area with high potential risk, rather than the risk of isolated candidate anomaly sampling points.

[0165] For example, if the risk concern area is a slope, and there are 5 candidate anomaly points in the upper left corner of the area where the slope is located, which are closely adjacent in space and form a cluster, the system will designate the contiguous area covered by them as the "upper left corner landslide hazard area", which is the candidate risk area.

[0166] In some embodiments, spatial proximity screening of candidate anomalous sampling points using a preset distance threshold can quickly eliminate spatially isolated and scattered anomalous sampling points, retaining only anomalous points with spatial clustering characteristics as seeds for subsequent analysis, effectively reducing the false alarm rate caused by single-point measurement errors or local minor perturbations. Specifically, determining the adjacent candidate anomalous sampling points corresponding to the target candidate anomalous sampling point from the candidate anomalous sampling points includes:

[0167] Obtain the spatial distance between the target candidate anomaly sampling point and the candidate anomaly sampling point;

[0168] Candidate anomaly sampling points whose spatial distance is within a preset distance threshold are identified as adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point.

[0169] Among them, the spatial distance between candidate anomaly sampling points and other candidate anomaly sampling points refers to the physical straight-line distance (or surface distance based on actual terrain) between the currently analyzed candidate anomaly sampling point and any other candidate anomaly sampling point within the risk concern area.

[0170] The preset distance threshold is a pre-defined maximum distance standard used to determine whether two sampling points are spatially "adjacent". For example, the preset distance threshold is set to 50 meters. As long as the spatial distance between a candidate anomaly sampling point and other candidate anomaly sampling points is less than 50 meters, they are considered to be very close.

[0171] Other candidate anomaly sampling points are all remaining candidate anomaly sampling points within the risk concern area, excluding the one currently being analyzed.

[0172] In some embodiments, by introducing a candidate core sampling point determination and neighborhood association mechanism, spatially clustered anomalies can be clustered and divided, integrating discrete anomaly signals into anomaly clusters with clear boundaries. This allows for the accurate identification of contiguous candidate risk areas, providing an intuitive spatial basis for subsequently delineating specific risk control areas. Specifically, based on target candidate anomaly sampling points and adjacent candidate anomaly sampling points, candidate risk areas are determined from the risk concern areas, including:

[0173] When the number of adjacent candidate anomaly sampling points is greater than the preset number threshold, the corresponding target candidate anomaly sampling point is determined as the candidate core sampling point.

[0174] If the spatial distance between any two candidate core sampling points is not greater than a preset distance threshold, then the two candidate core sampling points are determined to be in a neighborhood association state.

[0175] Based on the neighborhood association state, each candidate core sampling point is clustered to obtain at least one target anomaly cluster. Candidate core sampling points belonging to the same target anomaly cluster have direct or indirect neighborhood association relationships.

[0176] The areas covered by each target anomaly cluster in the risk focus area are identified as candidate risk areas.

[0177] The number of adjacent candidate anomaly sampling points refers to the total number of other candidate anomaly sampling points within a preset distance threshold range of the candidate anomaly sampling point. For example, if a 50-meter circle is drawn with candidate anomaly sampling point A as the center, and a total of 4 other candidate anomaly sampling points are found within the circle, the number is 4.

[0178] The preset threshold is the minimum number of neighboring points required to determine whether a candidate outlier is sufficient to be considered a "core" (similar to MinPts in density clustering). For example, the preset threshold can be set to 3. A point is considered a "core" only if it has at least 3 neighboring candidate outliers.

[0179] Candidate core sampling points refer to candidate anomaly sampling points whose number of adjacent candidate anomaly sampling points exceeds a preset threshold. They are high-density anomaly points that constitute the skeleton of the risk area. For example, if candidate anomaly sampling point A has 4 neighbors (4>3), then candidate anomaly sampling point A is a candidate core sampling point; while candidate anomaly sampling point B has only 1 adjacent candidate anomaly sampling point, so it is not a core point.

[0180] The spatial distance between any two candidate core sampling points is used as a distance criterion to determine whether the two candidate core sampling points belong to the same target anomaly cluster. For example, the straight-line distance between candidate core sampling point A and candidate core sampling point B is 40 meters.

[0181] The preset distance threshold is defined as above and is used to determine whether two candidate core sampling points are "connected". For example, the preset distance threshold is still 50 meters. Because 40 meters < 50 meters, candidate core sampling point A and candidate core sampling point B are considered connected.

[0182] The neighborhood association state refers to the logical connectivity established between any two candidate core sampling points when the spatial distance between them meets a preset proximity condition (i.e., not greater than a preset distance threshold). This state indicates that the two sampling points are closely adjacent in spatial distribution and together constitute a local component of the same potential risk entity.

[0183] For example, if the straight-line distance between candidate core sampling point A and candidate core sampling point B is 30 meters, and the preset distance threshold is 50 meters, since 30 meters is less than 50 meters, the system determines that A and B are in a neighboring association state. This means that A and B are considered to be spatially connected and belong to the same contiguous risk area.

[0184] A target anomaly cluster is an independent set (cluster) formed by aggregating all directly or indirectly connected candidate core sampling points through neighborhood association relationships. For example, A is connected to B, and B is connected to C. Although A and C are far apart, they belong to the same group through B. {A, B, C} constitutes a target anomaly cluster.

[0185] Direct or indirect neighborhood associations are defined as follows: "direct" means the distance between two core points is within a threshold; "indirect" means that although two core points are far apart, they can be connected through one or more intermediate core points (density achievable). For example, if A and B are adjacent (direct association), and B and C are adjacent (direct association), then A and C are indirectly associated through B.

[0186] It is understandable that for the set of candidate abnormal units (i.e., the set of candidate anomaly sampling points) introduces spatial constraints. Let the spatial distance threshold (i.e., the preset distance threshold) be... Then the first Each candidate anomaly unit (i.e., candidate anomaly sampling point) is located at a spatial distance threshold. The set of neighboring candidate anomaly cells (i.e., the set of adjacent candidate anomaly sampling points) is defined as... The number of its corresponding neighboring candidate anomaly units (i.e., the number of adjacent candidate anomaly sampling points) is denoted as Let the minimum neighbor support number threshold (i.e., the preset number threshold) be . Then only retain those that satisfy the condition. The candidate anomaly units are obtained, resulting in a set of anomaly units after spatial support constraints (i.e., a set consisting of candidate core sampling points). Subsequently, in the assembly The spatial adjacency relation is defined above. If any two anomalous units (i.e., any two candidate core sampling points) satisfy the following conditions: If they are adjacent, then they are considered to be spatially connected. Based on this adjacency relationship (i.e., the neighborhood association state), it can be... It is decomposed into several connected partitions (i.e., at least one target anomaly cluster). ,in Indicates the first One candidate risk area.

[0187] like Figure 1g As shown, the spatial distance threshold selected here is =5 meters, minimum neighbor support threshold is This yields candidate risk regions, which are... Figure 1g Candidate regions in the data.

[0188] S106. Based on the deformation sequence corresponding to each deformation sampling point in the candidate risk area, determine the regional deformation trend corresponding to the candidate risk area, and filter the candidate risk areas according to the regional deformation trend to obtain the target risk area.

[0189] Among them, the regional deformation trend refers to the overall deformation evolution law of all deformation sampling points within the candidate risk area over time. It reflects whether the candidate risk area is in a state of "accelerated deformation", "uniform deformation" or "tending to stability", and is usually represented by a quantitative parameter (such as the slope of change).

[0190] Target risk areas refer to regions identified after screening for deformation trends that exhibit significant dynamic risks (such as continued growth or acceleration of deformation). These candidate risk areas are not only spatially clustered but also temporally unstable, making them key targets for immediate prevention and control.

[0191] For example, the system identifies three candidate risk regions. In one candidate risk region, the deformation has stopped (stabilized), while in the other two candidate risk regions, the deformation is still rapidly increasing. These two candidate risk regions that are still deteriorating are then identified as the target risk regions.

[0192] In some embodiments, by extracting the median deformation data of each deformation sampling point within the candidate risk region to construct time-series features and using covariance analysis to calculate trend characterization parameters, the interference of local outliers or noise data within the candidate risk region can be effectively suppressed, accurately capturing the overall deformation evolution direction and rate of the candidate risk region, and significantly improving the robustness of regional risk trend identification. Specifically, based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region, the regional deformation trend corresponding to the candidate risk region is determined, including:

[0193] For each data collection time point, the median deformation data corresponding to the data collection time point is obtained from the deformation data corresponding to all deformation sampling points within the candidate risk area.

[0194] Based on the median deformation data corresponding to the collection time points, the corresponding regional deformation time series features are constructed.

[0195] A temporal correlation analysis was conducted on the temporal characteristics of regional deformation to obtain trend parameters representing the evolution of deformation over time.

[0196] The trend characterization parameter is determined as the regional deformation trend corresponding to the candidate risk area.

[0197] The median deformation data corresponding to the collection time point is the value in the middle of the deformation data of all deformation sampling points in the candidate risk area, arranged from smallest to largest at that collection time point. For example, if there are 101 deformation sampling points in a certain area, on the 5th day, after sorting the data of these 101 deformation sampling points, the data of the 51st deformation sampling point is 3.2 mm, which is the median deformation data at that collection time point.

[0198] Regional deformation time series features are formed by concatenating the median deformation data corresponding to each collection time point in chronological order, creating a time series curve representing the overall change of the candidate risk area. For example, an array consisting of median deformation data from day 1 to day 30.

[0199] The trend characterization parameter is used to quantify the rate and direction of deformation change over time (in subsequent algorithms, this is essentially the slope of linear regression). For example, a trend characterization parameter of 0.2 indicates that the average daily deformation increase in this area is 0.2 mm.

[0200] In some embodiments, a method based on the least squares principle for calculating the covariance and variance ratio is used to transform complex temporal evolution patterns into a single quantitative slope parameter. This method is not only computationally efficient but also objectively reflects the average rate of change in regional deformation, providing a precise mathematical basis for subsequent risk level ranking. Specifically, a temporal correlation analysis is performed on the temporal characteristics of regional deformation to obtain trend parameters characterizing the evolution of deformation over time, including:

[0201] The mean value of each median deformation data point corresponding to the collection time point is calculated to obtain the reference time mean value, and the mean value of each median deformation data point is calculated to obtain the reference deformation mean value.

[0202] For each data collection point, the median deformation data is multiplied by the difference between the median deformation data and the reference deformation mean, and by the difference between the corresponding data collection point and the reference time mean, to obtain the corresponding trend covariance term.

[0203] The trend covariance terms corresponding to each data collection time point are summed to obtain the cumulative covariance characteristics of regional deformation.

[0204] The regional time discrete characteristics are obtained by summing the squared differences between each collection time point and the mean reference time.

[0205] The ratio of the cumulative covariant characteristics of regional deformation to the discrete temporal characteristics of regional deformation is determined as the trend characterization parameter representing the evolution of deformation over time.

[0206] The reference time mean is the arithmetic mean of all data collection points (e.g., day 1, day 2, ... day N), representing the center of the time axis. For example, if monitoring was conducted for 10 days, with time points from 1 to 10, the reference time mean would be 5.5.

[0207] The reference mean deformation is the arithmetic mean of the median deformation data across all data collection points, representing the center of the deformation level. For example, the median deformation over these 10 days averages to 4.0 mm.

[0208] The trend covariance term is the difference between the median deformation data and the reference deformation mean, as well as the difference between the corresponding collection time point and the reference time mean, reflecting whether deformation and time change in the same direction.

[0209] The cumulative covariance characteristic of regional deformation is the value obtained by summing up the trend covariance terms at all data collection time points.

[0210] The regional time discrete feature is obtained by squaring the difference between each collection time point and the reference time mean, and then summing them up.

[0211] The ratio of the cumulative covariant feature of regional deformation to the discrete feature of regional time is the sum of the cumulative covariant features divided by the sum of the discrete features of time. The final result is the slope of the trend.

[0212] In some embodiments, the candidate risk areas formed in step 105 need to be further screened for spatial scale and temporal trend.

[0213] (1) Spatial scale selection, specifically:

[0214] Let the first Candidate risk areas The number of monitoring units (i.e., deformation sampling points) included is To eliminate scattered noise and excessively small connected regions, the minimum size threshold of the risk region is set as follows: Only retain those that meet the requirements. Candidate risk areas.

[0215] (2) Time series trend filtering, specifically:

[0216] For each candidate risk area after spatial scale screening The median time series curve (i.e., the time series characteristics of regional deformation) of its candidate risk regions is constructed as follows: A linear fit is performed on the median time series curve over the entire analysis period, and its trend coefficient (i.e., trend characterization parameter) is defined as follows: ,in, , Only retain those that meet the requirements. Candidate risk areas, defined as those with a positive trend coefficient indicating an overall positive growth trend, are included in the subsequent risk ranking. After spatial scale and trend screening, a set of effective risk areas (i.e., a set consisting of target risk areas) is obtained. .

[0217] In some embodiments, risk areas are sorted and graded for early warning based on regional deformation trends, thereby optimizing the allocation of risk prevention and control resources and ensuring that monitoring personnel can prioritize areas with the fastest deterioration and highest risk levels, thus greatly improving the timeliness and pertinence of geological disaster emergency response. Specifically, after screening candidate risk areas based on regional deformation trends to obtain target risk areas, the process also includes:

[0218] Based on the regional deformation trend corresponding to each target risk area, the risk levels of each target risk area are ranked to obtain the corresponding risk ranking results.

[0219] Based on the risk ranking results, the warning level corresponding to each target risk area is determined;

[0220] Based on the warning level, corresponding deformation warning information is sent to preset associated terminals. The deformation warning information includes at least the geographical location of the corresponding target risk area and the risk ranking.

[0221] The risk ranking result is a list that arranges all target risk areas in descending order of their trend characteristic parameter (deformation deterioration rate). For example, target risk area A (slope 0.5) > target risk area B (slope 0.2) > target risk area C (slope 0.05).

[0222] Warning levels are categorized by the degree of urgency based on risk ranking (e.g., red, orange, yellow warnings). For example, target risk area A is designated as a red level 1 warning, while target risk area B is designated as an orange level 2 warning.

[0223] Pre-defined associated terminals are communication terminals that are pre-bound for relevant personnel or equipment to receive deformation early warning information. Examples include the project manager's mobile phone responsible for the slope safety, the monitoring center's large screen, or the automatic alarm broadcast.

[0224] Deformation warning information is the specific alarm content sent to the preset associated terminals.

[0225] For example, a deformation warning message could be a text message or app push notification containing the message: "[Red Alert] Deformation in slope area A of XX is accelerating, with a current rate of 0.5 mm / day. Please investigate immediately."

[0226] The geographical location of the target risk area is its specific coordinates or extent on a map. For example, 30.5°N, 104.2°E, or "the left slope of the section from K12+500 to K12+600".

[0227] Risk ranking directly informs the recipient of the region's ranking among all risk points in the deformation warning information. For example, the deformation warning information may state "Current risk ranking: No. 1".

[0228] It is understandable that the set of effective risk zones... Each target risk area is determined according to its trend coefficient (i.e., the regional deformation trend). Sort the data from largest to smallest. Let the risk zone after sorting be... The first can be further defined The risk order for the target risk areas is as follows: ,in Given the number of target risk areas, the risk area with the larger the trend coefficient is, the higher its risk priority.

[0229] Finally, the set This is the final set of risk zones, and its corresponding and These respectively characterize the trend strength of the risk zone and the risk ranking results.

[0230] like Figure 1h As shown, the minimum risk zone size is After minimum spatial scale screening and trend screening, the target risk area is obtained. The target risk area is... Figure 1h The risk zone.

[0231] In summary, this application can capture the deformation process of a region, which is beneficial for accurately and timely identifying risk areas and improving the accuracy and reliability of risk area identification.

[0232] To better implement the above methods, this application also provides a risk area determination system, which can be integrated into an electronic device, such as a terminal or server. The terminal can be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, or personal computer; the server can be a single server or a server cluster composed of multiple servers.

[0233] For example, in this embodiment, a risk area determination system specifically integrated into an electronic device will be used as an example to describe the method of this application embodiment in detail.

[0234] For example, the risk area determination system may include an acquisition unit 201, an analysis unit 202, a fusion unit 203, an anomaly determination unit 204, an area determination unit 205, and an area filtering unit 206, as follows:

[0235] (a) Obtaining Unit 201.

[0236] The acquisition unit 201 is used to acquire the deformation sequence corresponding to multiple deformation sampling points within the risk concern area. The deformation sequence includes the deformation data of the corresponding deformation sampling points at multiple consecutive collection time points.

[0237] (II) Analysis Unit 202.

[0238] Analysis unit 202 is used to perform analysis and processing on the deformation data in the corresponding deformation sequence for each deformation sampling point, and to obtain the deformation parameters corresponding to each deformation evaluation dimension.

[0239] (III) Fusion Unit 203.

[0240] The fusion unit 203 is used to fuse the deformation parameters corresponding to the deformation sampling points under each deformation evaluation dimension based on the distribution of deformation parameters of multiple deformation sampling points under each deformation evaluation dimension, so as to obtain the deformation anomaly index corresponding to the deformation sampling points.

[0241] (iv) Anomaly determination unit 204.

[0242] The anomaly determination unit 204 is used to determine candidate anomaly sampling points from multiple deformation sampling points based on the deformation anomaly index corresponding to each deformation sampling point.

[0243] (v) Regional determination unit 205.

[0244] The region determination unit 205 is used to determine the adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point from the candidate candidate anomaly sampling points based on the target candidate anomaly sampling point, and to determine the candidate risk region from the risk concern region based on the target candidate anomaly sampling point and the adjacent candidate candidate anomaly sampling points. The target candidate anomaly sampling point is any candidate anomaly sampling point, and the candidate candidate anomaly sampling points are candidate anomaly sampling points other than the target candidate anomaly sampling point.

[0245] (vi) Regional Filtering Unit 206.

[0246] The region filtering unit 206 is used to determine the regional deformation trend corresponding to the candidate risk region based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region, and to filter the candidate risk region according to the regional deformation trend to obtain the target risk region.

[0247] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0248] Therefore, the embodiments of this application can capture the deformation process of the region, which is conducive to accurately and timely discovering risk areas and improving the accuracy and reliability of risk area identification.

[0249] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0250] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the steps in any of the risk area determination methods provided in embodiments of this application.

[0251] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0252] Since the instructions stored in the storage medium can execute the steps in any of the risk area determination methods provided in the embodiments of this application, the beneficial effects that any of the risk area determination methods provided in the embodiments of this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0253] According to one aspect of this application, a computer program product or computer program is provided, comprising a computer program / instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer program / instructions from the computer-readable storage medium and executes the computer program / instructions, causing the electronic device to perform the method provided in the risk area determination aspect of the above embodiments.

[0254] The above provides a detailed description of a risk area determination method provided by the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for determining risk areas, characterized in that, The method includes: Obtain deformation sequences corresponding to multiple deformation sampling points within a risk concern area, wherein the deformation sequences include deformation data corresponding to the corresponding deformation sampling points at multiple consecutive collection time points; For each deformation sampling point, the deformation data in the corresponding deformation sequence is analyzed and processed according to multiple deformation evaluation dimensions to obtain the deformation parameters corresponding to each deformation evaluation dimension; Based on the distribution of deformation parameters of the multiple deformation sampling points under each deformation evaluation dimension, the deformation parameters corresponding to the deformation sampling points under each deformation evaluation dimension are fused to obtain the deformation anomaly index corresponding to the deformation sampling points. Based on the deformation anomaly index corresponding to each deformation sampling point, candidate anomaly sampling points are determined from the plurality of deformation sampling points; Based on the target candidate anomaly sampling point, the adjacent candidate anomaly sampling point corresponding to the target candidate anomaly sampling point is determined from the candidate candidate anomaly sampling points. Based on the target candidate anomaly sampling point and the adjacent candidate anomaly sampling points, the candidate risk area is determined from the risk concern area. The target candidate anomaly sampling point is any candidate anomaly sampling point, and the candidate candidate anomaly sampling point is the candidate anomaly sampling point other than the target candidate anomaly sampling point. Based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region, the regional deformation trend corresponding to the candidate risk region is determined, and the candidate risk region is screened according to the regional deformation trend to obtain the target risk region.

2. The method as described in claim 1, characterized in that, The analysis and processing of deformation data in the corresponding deformation sequence across multiple deformation evaluation dimensions yields deformation parameters corresponding to each deformation evaluation dimension, including: Based on a preset time interval, the corresponding deformation sequence is divided to obtain a first sub-deformation sequence and a second sub-deformation sequence, wherein the acquisition time of the first sub-deformation sequence is earlier than that of the second sub-deformation sequence. The mean value is calculated based on the deformation data in the second sub-deformation sequence to obtain the recent mean deformation parameter corresponding to the second sub-deformation sequence; The mean value is calculated based on the deformation data in the first sub-deformation sequence to obtain the historical deformation mean parameter corresponding to the first sub-deformation sequence. The deformation growth parameter is obtained by determining the data offset of the recent mean deformation parameter relative to the historical mean deformation parameter; Linear trend fitting is performed based on the deviation between each deformation data in the second sub-deformation sequence and the recent mean deformation parameter to obtain the recent slope parameter reflecting the deformation enhancement trend.

3. The method as described in claim 2, characterized in that, The method of performing linear trend fitting based on the deviation between each deformation data in the second sub-deformation sequence and the recent mean deformation parameter to obtain a recent slope parameter reflecting the deformation enhancement trend includes: The average value of the time points corresponding to the collection time points of each deformation data in the second sub-deformation sequence is calculated to obtain the time average value. For each deformation data in the second sub-deformation sequence, the difference between the deformation data and the recent mean deformation parameter, and the difference between the collection time point corresponding to the deformation data and the time mean, are multiplied to obtain the covariance term corresponding to the deformation data. Summing the covariant terms yields the cumulative covariant characteristic of deformation; The time discrete features are obtained by squaring and summing the differences between the acquisition time points of each deformation data in the second sub-deformation sequence and the time mean. The ratio of the cumulative deformation covariance feature to the discrete time feature is determined to obtain the recent slope parameter.

4. The method as described in claim 1, characterized in that, The deformation parameters of the multiple deformation sampling points under each deformation evaluation dimension are fused based on their distribution to obtain the deformation anomaly index corresponding to each deformation sampling point, including: For each deformation assessment dimension, based on the deformation parameters corresponding to all deformation sampling points within the risk concern area under the deformation assessment dimension, a central benchmark value and a discrete score are determined. The central benchmark value reflects the overall deformation level of the multiple deformation sampling points under the deformation assessment dimension, and the discrete score reflects the degree of fluctuation of the deformation parameters corresponding to each deformation sampling point under the deformation assessment dimension relative to the overall deformation level. For each deformation sampling point, determine the degree of deviation of its corresponding deformation parameter from the corresponding central reference value under each deformation evaluation dimension; Based on the discrete scores, the corresponding deviation degree is normalized to obtain the sub-anomaly intensity of the deformation sampling point under the corresponding deformation evaluation dimension. Based on the preset weights corresponding to each deformation assessment dimension, the sub-anomaly intensities of the deformation sampling points in each deformation assessment dimension are weighted and fused to obtain the deformation anomaly index corresponding to the deformation sampling points.

5. The method as described in claim 4, characterized in that, Determining the deviation of the deformation parameters corresponding to each deformation assessment dimension from the corresponding central reference value includes: Determine the original difference between the deformation parameters of the deformation sampling points and the corresponding central reference values; If the absolute value of the original difference is greater than a preset threshold, then the original difference is determined as the degree of deviation; If the absolute value of the original difference is not greater than the preset threshold, then the degree of deviation is determined to be zero; The step of normalizing the corresponding deviation based on the discrete scores to obtain the sub-anomaly intensity of the deformation sampling point under the corresponding deformation evaluation dimension includes: The discrete scores are summed with the preset stability parameters to obtain the corrected discrete scores; The ratio of the deviation degree to the corrected discrete score is determined to obtain the sub-anomaly intensity of the deformation sampling point in the corresponding deformation evaluation dimension.

6. The method as described in claim 1, characterized in that, The step of determining the adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point from the candidate anomaly sampling points includes: Obtain the spatial distance between the target candidate anomaly sampling point and the candidate anomaly sampling point; The candidate anomaly sampling points whose spatial distance is within a preset distance threshold are determined as the adjacent candidate anomaly sampling points corresponding to the target candidate anomaly sampling point.

7. The method as described in claim 1, characterized in that, The step of determining candidate risk regions from the risk concern region based on the target candidate anomaly sampling points and the adjacent candidate anomaly sampling points includes: When the number of adjacent candidate abnormal sampling points is greater than a preset number threshold, the corresponding target candidate abnormal sampling point is determined as a candidate core sampling point. If the spatial distance between any two candidate core sampling points is not greater than the preset distance threshold, then the two candidate core sampling points are determined to be in a neighborhood association state. Based on the neighborhood association state, each candidate core sampling point is clustered to obtain at least one target anomaly cluster. Candidate core sampling points belonging to the same target anomaly cluster have direct or indirect neighborhood association relationships. The areas covered by each target anomaly cluster in the risk concern area are identified as candidate risk areas.

8. The method as described in claim 1, characterized in that, The step of determining the regional deformation trend corresponding to the candidate risk region based on the deformation sequence corresponding to each deformation sampling point in the candidate risk region includes: For each collection time point, the median deformation data corresponding to the collection time point is obtained from the deformation data corresponding to all deformation sampling points within the candidate risk area. Based on the median deformation data corresponding to the acquisition time points, construct the corresponding regional deformation time series features; A temporal correlation analysis was performed on the temporal characteristics of the deformation in the region to obtain trend parameters that characterize the evolution of deformation over time. The trend characterization parameter is determined as the regional deformation trend corresponding to the candidate risk area.

9. The method as described in claim 8, characterized in that, The temporal correlation analysis of the deformation time series characteristics of the region yields trend parameters characterizing the evolution of deformation over time, including: The mean value of the collection time points corresponding to each median deformation data is calculated to obtain the reference time mean value, and the mean value of each of the median deformation data is calculated to obtain the reference deformation mean value. For each data collection time point, the difference between the median deformation data and the reference deformation mean, and the difference between the corresponding data collection time point and the reference time mean, are multiplied to obtain the corresponding trend covariance term. The trend covariance terms corresponding to each data collection time point are summed to obtain the cumulative covariance characteristics of regional deformation. The regional time discrete characteristics are obtained by summing the squared differences between each collection time point and the mean of the reference time. The ratio of the cumulative covariant feature of the regional deformation to the discrete temporal feature of the region is determined as a trend characterization parameter representing the evolution of deformation over time.

10. The method as described in claim 1, characterized in that, After filtering the candidate risk regions based on the regional deformation trend to obtain the target risk region, the method further includes: Based on the regional deformation trend corresponding to each target risk area, the risk levels of each target risk area are ranked to obtain the corresponding risk ranking results; based on the risk ranking results, the warning level corresponding to each target risk area is determined. Based on the warning level, corresponding deformation warning information is sent to preset associated terminals. The deformation warning information includes at least the geographical location of the corresponding target risk area and the risk ranking.

Images