A slope instability early warning method and system based on big data analysis
The slope instability early warning method based on big data analysis constructs multi-condition clustering modeling and dynamic parameter optimization, which solves the problems of single slope early warning model, poor adaptability to conditions, one-sided risk assessment and lack of dynamic optimization capability in the existing technology, and achieves high-precision and low-false-alarm adaptive early warning effect.
Patent Information
- Application Number
- CN202611217431.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-12
- Publication Date
- 2026-11-06
AI Technical Summary
Existing slope instability early warning technologies suffer from problems such as poor model versatility, limited monitoring dimensions, coarse data processing, neglect of multi-factor coupling and linkage mechanisms, fixed and unadaptive model parameters, and crude risk classification and handling mechanisms. These issues result in high false alarm and false alarm rates, failing to meet the requirements for high-precision, low-false-alarm, and adaptive early warning under complex working conditions.
By constructing a closed-loop early warning process that includes multi-condition clustering modeling, dynamic parameter optimization, two-factor coupled calculation, and hierarchical disposal, and by using big data analysis methods, historical big data samples are collected for cluster analysis to generate a dedicated early warning model. Soil and location data are collected in real time, and standardized and time-series noise reduction is performed. The synergistic influence coefficient of soil and location movement is explored, and the weights are iteratively optimized to achieve multi-level risk assessment and disposal.
It achieves high-precision, low-false-alarm, adaptive early warning for slope instability under complex working conditions, reduces the false alarm and missed alarm rates, and improves the adaptability and accuracy of the early warning.
Smart Images
Figure CN123052908A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of slope instability monitoring technology, and in particular to a slope instability early warning method and system based on big data analysis. Background Technology
[0002] Slope instability, landslides, and collapses are the most frequent and widespread geological hazards in geotechnical engineering. They are characterized by high concealment, complex triggering factors, sudden destructive events, and a high risk of chain reactions, easily causing road blockages, engineering damage, casualties, and significant economic losses. Currently, slope safety early warning technology is gradually transitioning from manual inspections and single-point threshold alarms to intelligent monitoring and early warning systems. However, existing intelligent early warning technologies still face several core technological bottlenecks, making it difficult to meet the demands for high-precision, low-false-alarm, and adaptive early warning systems under complex working conditions.
[0003] Existing early warning technologies suffer from the following shortcomings: First, their models have extremely poor versatility, mostly employing a single, fixed early warning model with fixed weight parameters. They fail to differentiate between slope instability mechanisms and cannot adapt to the two completely different failure modes: long-term creep and slippage, and sudden instability due to heavy rain. This leads to missed warnings for some slopes and frequent false alarms for others. Second, their monitoring dimensions are limited and data processing is crude. They only perform simple threshold judgments on raw monitoring data without noise reduction, anomaly removal, or standardized quantification of sensor data. The raw data suffers from significant noise interference, resulting in insufficient accuracy of the basic early warning data. Third, they neglect multi-factor coupling and linkage mechanisms, including the relationship between slope soil stability decay and slope displacement. Deformation is not an independent action; the two are coupled in a synchronous linkage, delayed response, and mutual promotion manner. Traditional technologies have not explored the temporal coupling correlation characteristics, and risk assessment is only a superposition of single data, which cannot identify hidden delayed risks. Fourth, the model parameters are fixed and lack adaptability. Once the model weights are set, they are fixed for a long time and cannot be dynamically optimized according to the stability of monitoring data, the frequency of false alarms, and the evolution of operating conditions. After long-term operation, the adaptability continues to decline, and the false alarm and missed alarm rates increase significantly. Fifth, the risk classification and disposal mechanism is crude and lacks a refined multi-level threshold matching system. It is impossible to match differentiated disposal strategies according to the risk level, resulting in problems of over-prevention or delayed prevention and control.
[0004] Based on the above, the existing technology needs further improvement. Summary of the Invention
[0005] The purpose of this disclosure is to provide a slope instability early warning method and system based on big data analysis. The slope instability early warning method solves the problems of existing technologies such as single slope early warning model, poor adaptability to working conditions, one-sided risk assessment, and lack of dynamic optimization capability by constructing a closed-loop early warning process of multi-condition clustering modeling, dynamic parameter optimization, two-factor coupled calculation and hierarchical treatment.
[0006] To address the aforementioned technical problems, this disclosure provides a slope instability early warning method based on big data analysis, comprising the following steps: Step 1: Collect a large number of historical big data samples of slopes covering different working conditions. By performing cluster analysis on the historical big data samples, obtain multiple slope type clusters and corresponding early warning models for each slope type cluster. Step 2: Based on the historical data of the target slope, determine the slope type cluster to which the target slope belongs, select the exclusive early warning model corresponding to the slope type cluster to obtain the operating model, and collect the soil data and location movement data of the target slope in real time. Input the soil data and location movement data into the operating model to calculate the soil stability value and the location movement hazard value. Step 3: Construct a two-factor time series data sequence using soil stability and location movement hazard values. Analyze the two-factor time series data sequence through big data correlation analysis to obtain the synergistic influence coefficient of soil-location movement interaction. Step 4: Based on the initial weights of the running model, and combined with the false alarm contribution of soil data and location movement data, iteratively modify the weights to obtain the factor importance ratios corresponding to soil stability and location movement hazard. Step 5: Using the synergistic influence coefficient and the proportion of factor importance, the soil stability value and the location movement hazard value are fused and calculated to obtain the comprehensive early warning value of the target slope. Step 6: Compare the comprehensive early warning values with the early warning thresholds of the operating model, determine the instability risk status of the target slope based on the comparison results, and execute the corresponding level of early warning handling operations.
[0007] In some embodiments, in step 1, the K-means big data clustering algorithm is used to perform cluster analysis on historical big data samples to obtain multiple slope type clusters, including: long-term creep type slope clusters and rainstorm sudden type slope clusters.
[0008] In some embodiments, the initial weight configuration of the dedicated early warning model corresponding to the long-term creep slope cluster is that the weight of the location movement hazard is greater than the weight of the soil stability; the initial weight configuration of the dedicated early warning model corresponding to the rainstorm sudden slope cluster is that the weight of the soil stability is greater than the weight of the location movement hazard.
[0009] In some embodiments, in step 2, Based on the data characteristics of the slope type cluster corresponding to the operating model, the real-time collected soil data is standardized and preprocessed to remove abnormal interference data. Combined with the soil mechanics association rules built into the operating model, the soil stability value that characterizes the overall stability performance of the slope soil is quantified. Simultaneously, the real-time acquired location movement data is subjected to time-series noise reduction processing. Combined with the slope displacement deformation response law adapted to the operating model, the location movement hazard level, which characterizes the level of slope slip risk, is obtained by quantitative calculation based on two dimensions: cumulative displacement and displacement change rate.
[0010] In some embodiments, in step 3 Using continuous sampling times as the time axis, the soil stability and location movement risk values at different times are time-aligned and arranged to construct a synchronous and linked two-factor time-series data sequence. A big data correlation mining algorithm was used to perform temporal correlation analysis on the two-factor time series data sequence to explore the synchronous correlation, hysteresis response and mutual promotion coupling between soil stability performance decay and slope position movement and deformation. Based on the laws of synchronous correlation, delayed response, and mutual promotion and coupling, the synergistic influence coefficient is quantified. The synergistic influence coefficient is used to characterize the interaction strength between soil state and slope displacement. The larger the coefficient value, the higher the linkage risk of slope instability induced by the coupling of two factors.
[0011] In some embodiments, in step 4, During real-time monitoring, the frequency of abnormal fluctuations and the number of false alarms triggered by soil data and location movement data were statistically analyzed. The false alarm contribution of soil data and the false alarm contribution of location movement data were calculated separately. The initial weights of the operating model were adjusted based on the false alarm contribution of soil data and the false alarm contribution of location movement data to obtain the proportion of factor importance corresponding to soil stability and location movement hazard.
[0012] In some embodiments, in step 5 F = K x W 1 x X + W 2 ×Y in, F To integrate early warning values, K is Synergistic influence coefficient W 1 and W 2 The proportion of factor importance and satisfying W 1 + W 2 =1，X This represents the soil stability value. Y This represents the numerical value indicating the degree of danger associated with location relocation.
[0013] In some embodiments, in step 6, the warning threshold is divided into multiple levels. The comprehensive warning value is compared with the multi-level warning threshold one by one to determine the instability risk status of the target slope. The corresponding warning and handling operations include routine monitoring, risk warning, emergency prevention and control, and emergency response.
[0014] On the other hand, this disclosure also provides a slope instability early warning system based on big data analysis, used to execute the above-mentioned slope instability early warning method. The slope instability early warning system includes: a big data clustering modeling module, a model matching module, a data acquisition module, a time series correlation analysis module, a fusion calculation module, and a hierarchical early warning execution module, wherein: Big Data Clustering Modeling Module: Used to collect historical big data samples of slopes under multiple working conditions, cluster and divide slope type clusters, and generate special early warning models for various slopes; Model matching module: Matches the corresponding dedicated early warning model based on the historical data of the target slope to obtain the running model; Data acquisition module: used to collect real-time soil data and location movement data of the target slope, and preprocess the data to obtain soil stability value and location movement hazard value; Temporal correlation analysis module: used to construct two-factor time-series data sequences and analyze the synergistic influence coefficient of soil-location interaction; Fusion calculation module: Iteratively optimizes the initial weights based on the contribution of data false alarms, and calculates the comprehensive early warning value by weighting. Tiered early warning execution module: Determines the level of instability risk by comparing with multi-level early warning thresholds, and executes corresponding operations for routine monitoring, risk warning, emergency prevention and control, and disaster relief.
[0015] By adopting the above technical solution, the present invention has at least the following beneficial effects: This disclosure proposes a slope instability early warning method and system based on big data analysis. The slope instability early warning method solves the problems of existing technologies such as single slope early warning model, poor adaptability to working conditions, one-sided risk assessment, and lack of dynamic optimization capability by constructing a closed-loop early warning process of multi-condition clustering modeling, dynamic parameter optimization, two-factor coupled calculation and hierarchical treatment. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a flowchart of a slope instability early warning method based on big data analysis according to an embodiment of the present disclosure. Detailed Implementation
[0018] The embodiments of this disclosure will be further described in detail below with reference to the accompanying drawings and examples. The detailed description of the embodiments and the accompanying drawings are used to illustrate the principles of this disclosure by way of example, but should not be used to limit the scope of this disclosure. This disclosure can be implemented in many different forms and is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.
[0019] These embodiments are provided to make the disclosure thorough and complete, and to fully express the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specifically stated, the relative arrangement of components and steps, material composition, numerical expressions, and values set forth in these embodiments should be interpreted as exemplary only and not as limiting.
[0020] It should be noted that, in the description of this disclosure, unless otherwise stated, "a plurality of" means two or more; the terms "upper," "lower," "left," "right," "inner," and "outer," etc., indicating orientation or positional relationship, are only for the convenience of describing this disclosure and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this disclosure. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0021] Furthermore, the terms "first," "second," and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different parts. "Vertical" is not strictly vertical, but within the permissible margin of error. "Parallel" is not strictly parallel, but within the permissible margin of error. Terms such as "including" or "contains" mean that the element preceding the word encompasses the element listed after the word, and do not exclude the possibility of encompassing other elements as well.
[0022] It should also be noted that, in the description of this disclosure, unless otherwise expressly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this disclosure depending on the specific circumstances. When a particular device is described as being located between a first device and a second device, an intermediary device may or may not be present between the particular device and the first or second device.
[0023] All terms used in this disclosure have the same meaning as understood by one of ordinary skill in the art to which this disclosure pertains, unless otherwise specifically defined. It should also be understood that terms defined in general dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant art, and not as idealized or highly formalized, unless expressly defined herein.
[0024] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the specification.
[0025] As mentioned in the background section above, existing technologies suffer from problems such as simplistic slope early warning models, poor adaptability to working conditions, one-sided risk assessment, and lack of dynamic optimization capabilities. To address these issues, this application proposes a slope instability early warning method and system based on big data analysis. This method solves the problems of simplistic slope early warning models, poor adaptability to working conditions, one-sided risk assessment, and lack of dynamic optimization capabilities in existing technologies by constructing a closed-loop early warning process that includes multi-condition clustering modeling, dynamic parameter optimization, two-factor coupled calculation, and hierarchical handling.
[0026] To address the aforementioned technical problems, this invention provides a slope instability early warning method based on big data analysis, comprising the following steps: Step 1: Collect a large number of historical big data samples of slopes covering different working conditions. Through cluster analysis of the historical big data samples, obtain multiple slope type clusters and corresponding exclusive early warning models for each slope type cluster; Step 2: Based on the historical data of the target slope, determine the slope type cluster to which the target slope belongs, select the exclusive early warning model corresponding to that slope type cluster to obtain the operating model, and collect soil data and location movement data of the target slope in real time. Input the soil data and location movement data into the operating model to calculate the soil stability value and the location movement hazard value; Step 3: Utilize the soil stability value and location movement... Step 4: Construct a two-factor time-series data sequence based on the hazard level values. Analyze the two-factor time-series data sequence using big data correlation analysis to obtain the synergistic influence coefficient of the interaction between soil and location movement. Step 5: Based on the initial weights of the operating model, iteratively modify the false alarm contribution of soil data and location movement data to obtain the factor importance ratios corresponding to soil stability and location movement hazard levels. Step 6: Use the synergistic influence coefficient and factor importance ratios to fuse the soil stability values and location movement hazard values to obtain the comprehensive early warning value for the target slope. Step 7: Compare the comprehensive early warning value with the early warning threshold of the operating model, determine the instability risk status of the target slope based on the comparison results, and execute the corresponding level of early warning response.
[0027] Specifically, this technical solution constructs a six-step early warning architecture: "Big Data Clustering Modeling and Classification—Specific Model Matching and Two-Factor Data Quantization—Time-Series Coupled Correlation Analysis—False Alarm Contribution Weight Iterative Optimization—Two-Factor Coupled Fusion Risk Calculation—Multi-Level Risk Grading Early Warning and Handling." This architecture enables intelligent and accurate early warning for slopes under multiple working conditions throughout the entire process. Figure 1 As shown, the specific process is as follows: Big Data Clustering Modeling: Collect massive historical monitoring big data samples of slopes from different regions, geological conditions, working conditions, and instability types. The sample data covers indicators such as slope soil parameters, displacement time series data, meteorological data, instability records, and historical early warning data. Unsupervised clustering analysis is performed on all historical big data samples. Based on the slope instability mechanism, failure characteristics, and data change patterns, multiple slope type clusters are automatically obtained. At the same time, for each type of slope cluster, sample training and fitting are used to generate a dedicated early warning model adapted to the instability characteristics of that type of slope.
[0028] Model matching and two-factor data quantification: Historical working condition data, geological parameters, and environmental data of the target slope to be monitored are retrieved. The instability mechanism and working condition attributes of the target slope are determined by feature matching algorithm, and the slope type cluster is matched. The exclusive early warning model corresponding to the type cluster is retrieved as the real-time running model. On-site monitoring equipment is started to collect soil monitoring data and location movement monitoring data of the target slope in real time and continuously. The raw monitoring data is standardized preprocessed, anomaly removed, time-series noise reduced, and multi-dimensional quantization calculated to transform the raw sensor data into standardized soil stability value and location movement hazard value.
[0029] Temporal correlation analysis and synergy coefficient calculation: Taking the fixed sampling time of the equipment as a continuous time axis, the soil stability value and the location movement risk value corresponding to different sampling times are time-series aligned and time-synchronized to construct a two-factor synchronous linkage time-series data sequence; a big data correlation mining algorithm is used to perform in-depth analysis of the time-series sequence to mine the temporal linkage pattern and coupling response characteristics of the two types of data, and quantify the synergistic influence coefficient of the interaction between soil state and slope displacement to characterize the risk intensity of instability induced by the coupling of the two factors.
[0030] Iterative optimization of false alarm contribution weight: The initial weight ratio preset by the current running model is retrieved, and the frequency of abnormal fluctuations, the number of false alarms triggered, and the number of invalid alarms of soil data and location movement data are statistically analyzed in real time during the monitoring period. The independent false alarm contribution of the two types of monitoring factors is calculated separately, and the initial weight parameters are finely adjusted in reverse according to the magnitude of the false alarm contribution. The weight ratio of factors with poor stability and high false alarm rate is reduced, while the weight ratio of factors with accurate monitoring and reliable risk characterization is increased. After multiple rounds of real-time iterative correction, the dynamic factor importance ratio adapted to the real-time working conditions of the target slope is obtained.
[0031] Dual-factor coupling and fusion calculation: The importance ratio of dynamically optimized factors and the synergistic influence coefficient obtained from time-series analysis are retrieved. Through a preset coupling and fusion calculation formula, the soil stability value and the location movement hazard value are weighted, fused and coupled for correction calculation. This offsets the error defects of independent single-factor calculation, superimposes the dual-factor linkage instability effect, and finally quantifies a comprehensive early warning value that can truly reflect the overall slope instability risk.
[0032] Multi-level risk assessment and graded response: Retrieve multi-level differentiated early warning thresholds generated by the pre-training of the operation model, compare the real-time calculated comprehensive early warning value with each level threshold, match the corresponding risk level range, and determine the current instability risk status of the target slope; according to different risk levels, automatically execute the corresponding standardized early warning response operations.
[0033] Compared with existing technologies, the slope instability early warning method based on big data analysis proposed in this application solves the problems of traditional slope early warning models being singular, having poor adaptability to working conditions, having one-sided risk assessment, and lacking dynamic optimization capabilities by constructing a closed-loop early warning process that includes multi-condition clustering modeling, dynamic parameter optimization, dual-factor coupled calculation, and hierarchical treatment.
[0034] In the above method, in step 1, the K-means big data clustering algorithm is used to cluster historical big data samples to obtain various slope type clusters, including: long-term creep slope clusters and rainstorm-induced slope clusters. Specifically, the K-means algorithm has the advantages of processing massive time-series data, fast convergence speed, high clustering accuracy, and strong group identification, making it suitable for slope classification. Based on the instability mechanism, data fluctuation characteristics, and environmental induction patterns of the sample data, two types of slope type clusters are finally obtained: long-term creep slope clusters and rainstorm-induced slope clusters.
[0035] Furthermore, the cluster analysis process is as follows: Data preprocessing: The collected historical big data samples of slopes are cleaned, missing values, outliers, and invalid samples are removed, and multi-dimensional dimensional data are standardized and normalized to eliminate the interference of dimensional differences on the clustering results; Cluster center initialization: Preferably, the number of cluster categories is preset to K=2, and the cluster centers of two types of slopes (long-term creep slope cluster and rainstorm sudden slope cluster) are randomly initialized; Distance iteration calculation: The distance between each sample and the two cluster centers is calculated using Euclidean distance, and the sample is assigned to the nearest cluster; Cluster center update: After each round of sample division, the mean of each cluster is recalculated, and the cluster centers are updated; Iterative convergence judgment: The iterative calculation is repeated until the cluster centers no longer change or the maximum number of iterations is reached, and the clustering is completed; Category labeling: Based on the working condition characteristics of the clustered samples, samples with continuous slow displacement and no sudden causes are labeled as long-term creep slope clusters, and samples induced by rainstorms and with rapid soil deterioration are labeled as rainstorm sudden slope clusters.
[0036] Among them, the long-term creep-type slope cluster: the instability induction is mainly due to long-term soil fatigue and slow slope slip accumulation, with no obvious sudden causes. Displacement data shows a continuous small-scale increasing trend, and soil performance deteriorates slowly. Instability is characterized by long-term, cumulative, and predictable characteristics. The rainstorm-induced slope cluster: the instability induction is mainly due to short-term heavy rainfall infiltration. Rainwater causes a sharp increase in soil moisture content, a significant decrease in shear strength, and soil softening failure. Displacement data shows a sudden and rapid increase in a short period of time. Instability is characterized by suddenness, severity, and difficulty in early prediction. By using K-means big data for precise clustering, the problems of traditional early warning models being unable to distinguish slope instability mechanisms and poor adaptability of general models are solved.
[0037] In the above methods, the initial weight configuration of the dedicated early warning model corresponding to the long-term creep type slope cluster is that the weight of the location movement hazard is greater than the weight of the soil stability; the initial weight configuration of the dedicated early warning model corresponding to the rainstorm sudden type slope cluster is that the weight of the soil stability is greater than the weight of the location movement hazard.
[0038] For long-term creep-type slope clusters, the core dominant factor for slope instability is the long-term accumulation of slope displacement and the development of sliding deformation, while the degradation of soil properties is a secondary factor. Therefore, the initial weight configuration of the dedicated early warning model is: weight of location movement hazard > weight of soil stability. Preferably, the initial weight ratio is: weight of location movement hazard 0.65, weight of soil stability 0.35.
[0039] For slope clusters prone to sudden rainstorms, the core instability factors are soil softening and failure of soil stability caused by rainwater infiltration. Displacement deformation is a secondary result of soil failure. Therefore, the initial weight configuration of the dedicated early warning model is: soil stability weight > location movement hazard weight. Preferably, the initial weight ratio is: soil stability weight 0.65, location movement hazard weight 0.35.
[0040] By configuring differentiated initial weights, the technical deficiency of the existing technology that uniform weights cannot be adapted to slopes with multiple working conditions is solved, so that the initial running state of the model is consistent with the real geological laws, thereby reducing the probability of false alarms and false alarms.
[0041] In the above method, in step 2, based on the slope type cluster data characteristics corresponding to the running model, the real-time collected soil data is standardized and preprocessed to remove abnormal interference data. Combined with the soil mechanics association rules built into the running model, the soil stability value characterizing the overall stability performance of the slope soil is quantified. Simultaneously, the real-time collected location movement data is subjected to time-series noise reduction processing. Combined with the slope displacement deformation response law adapted to the running model, the location movement hazard value characterizing the level of slope slip risk is obtained by quantifying the cumulative displacement and the rate of displacement change in two dimensions.
[0042] Specifically, in response to the problems of high noise, numerous abnormal interferences, single data dimension, and inability to directly characterize risks in the raw data of slope field monitoring, this invention preprocesses and quantifies soil data and location movement data respectively, eliminates invalid data interference, and transforms the raw sensor data into standardized risk values in the 0-1 range, thereby achieving data quantification and providing data support for subsequent coupled calculations.
[0043] Furthermore, the numerical calculation process for soil stability is as follows: Data acquisition: Real-time acquisition of raw monitoring data of slope soil using soil moisture sensors, soil stress sensors, and compaction sensors; Preprocessing: Anomaly detection of the raw data using the 3σ criterion, removing abnormal interference data exceeding the normal fluctuation range, and eliminating erroneous data caused by sensor drift and environmental interference; Mechanical rule matching: Retrieving the soil mechanical association rules built into the operating model, and associating the core mechanical indicators of soil shear strength, moisture content, and compaction; Quantitative calculation: Using the soil stability quantification formula, multi-dimensional soil parameters are fused and calculated, and normalized to obtain the soil stability value in the 0-1 range. X The closer the value is to 1, the better the soil stability; the closer it is to 0, the more severe the soil deterioration and the higher the risk of instability.
[0044] Furthermore, the numerical calculation process for the degree of location movement hazard is as follows: Data acquisition: Real-time acquisition of raw time-series data of horizontal and vertical displacement of the slope through slope displacement monitoring equipment; Time-series noise reduction processing: Wavelet noise reduction algorithm is used to smooth and filter the displacement time-series data, eliminating invalid displacement data caused by instantaneous jitter and equipment interference, and retaining the true deformation trend; Index extraction: Simultaneously extracting two indicators: total cumulative displacement and real-time displacement change rate; Fusion quantification: Combining the slope displacement deformation response law adapted to the model, the cumulative displacement and displacement rate are weighted and fused, and normalized to obtain the numerical value of location movement hazard in the 0-1 interval. Y The closer the value is to 1, the higher the risk of slippage; the closer it is to 0, the more stable the slope displacement state.
[0045] In the above method, in step 3, using continuous sampling times as the time axis, the soil stability values and slope displacement hazard values corresponding to different times are time-aligned and arranged to construct a synchronously linked two-factor time-series data sequence. A big data association mining algorithm is used to perform time-series correlation analysis on the two-factor time-series data sequence to uncover the synchronous correlation, hysteresis response, and mutually reinforcing coupling laws between soil stability performance decay and slope displacement deformation. Based on the synchronous correlation, hysteresis response, and mutually reinforcing coupling laws, the synergistic influence coefficient is quantified. The synergistic influence coefficient is used to characterize the interaction strength between soil state and slope displacement. The larger the coefficient value, the higher the linkage risk of slope instability induced by the two-factor coupling.
[0046] Specifically, the decline in the stability performance of slope soil and displacement deformation are not independent changes, but have significant temporal coupling correlation patterns. This invention constructs a two-factor time-series data sequence of soil stability and location movement hazard, and uses a big data association mining algorithm to mine three core coupling frequent association rules: synchronous association, delayed response, and mutual promotion. Based on the strong association rules mined, the synergistic influence coefficient of the two factors is quantified to characterize the linkage risk intensity of slope instability induced by the coupling of the two factors.
[0047] Preferably, the big data association mining algorithm adopts the FP-Growth algorithm, which has the advantages of high computational efficiency, adaptability to massive time series data, high mining accuracy, and suitability for big data analysis of slope time series monitoring.
[0048] Furthermore, the temporal correlation analysis process based on the FP-Growth algorithm is as follows.
[0049] Standardized construction of time series: Using the fixed sampling period of the sensor as the time axis, the soil stability value and the location movement hazard value corresponding to the same sampling time are time-series aligned and time-synchronized, and time misalignment and missing abnormal data are eliminated to construct a continuous, synchronous and linked two-factor standardized time series data sequence. At the same time, the continuous values are discretized into three states: stable, weakly deteriorated and strongly deteriorated.
[0050] Time series dataset scanning and statistics: Traverse all time series samples, count the frequency of individual occurrences and joint occurrences of each soil state and displacement state, filter high-frequency state items that meet the minimum support, and construct the root node and first-level frequent items of the FP tree.
[0051] FP-tree construction and frequent pattern mining: Based on the selected high-frequency items, an FP-Growth tree structure is constructed, and all time-series association frequent patterns that satisfy the minimum support and minimum confidence are recursively mined.
[0052] Three types of temporal coupling patterns were accurately identified: Based on the frequent patterns and temporal lift results obtained from the mining, the synchronous correlation pattern of soil weakening and displacement increase was identified, the lag response pattern of soil deterioration followed by displacement increase was identified, and the mutually promoting coupling pattern of soil weakening aggravating displacement deformation and displacement deformation further damaging the soil structure was identified.
[0053] Quantitative calculation of synergistic impact coefficient: The final synergistic impact coefficient is obtained by weighting and fusing three quantitative indicators: synchronous correlation support, lag response enhancement, and mutual promotion confidence. K Preferably, K The value range is 1.0-1.3. K The larger the value, the stronger the coupling effect between soil condition and slope displacement, and the higher the risk of slope instability induced by the synergy of the two factors; the smaller the K value, the weaker the linkage between the two factors, and the risk of instability is mainly due to the independent action of a single factor.
[0054] In the above method, in step 4, the frequency of abnormal fluctuations and the number of false alarms triggered by soil data and location movement data during real-time monitoring are statistically analyzed. The false alarm contribution of soil data and the false alarm contribution of location movement data are calculated respectively. The initial weights of the operating model are adjusted based on the false alarm contribution of soil data and the false alarm contribution of location movement data to obtain the proportion of factor importance corresponding to soil stability and location movement hazard.
[0055] Specifically, fixed weights cannot adapt to the dynamic situation of slope working condition evolution and sensor status changes. This invention introduces a false alarm contribution iteration mechanism. By statistically analyzing the abnormal fluctuations and false alarm frequencies of two types of monitoring data, the false alarm contribution of each factor is quantified. Based on the false alarm contribution, the initial weights of the model are optimized in reverse iteration to achieve dynamic adaptive updating of weights, allowing the model to adapt to real-time working conditions in the long term.
[0056] Furthermore, the calculation process is as follows: Data statistics: During the real-time monitoring period, the frequency of abnormal fluctuations in soil data and location movement data, the number of false alarms triggered, and the number of invalid alarms are counted independently.
[0057] False alarm contribution calculation: Based on the statistical results, the false alarm contribution of soil data and the false alarm contribution of location movement data are calculated respectively. The false alarm contribution is positively correlated with the frequency of anomalies and the number of false alarms. The higher the frequency, the greater the false alarm contribution of the corresponding factor, and the worse the data reliability. Weight Iteration Adjustment Rules: Retrieve the initial weights of the running model and fine-tune the weight ratio in reverse according to the false alarm contribution of the two types of factors: reduce the weight ratio of factors with high false alarm contribution and poor data stability; increase the weight ratio of factors with low false alarm contribution, accurate monitoring and reliable risk characterization.
[0058] Iterative convergence optimization: Through multiple rounds of real-time monitoring and iterative correction, the interference of unstable factors is gradually weakened and the dominant role of accurate factors is strengthened.
[0059] Output dynamic weights: After iterative convergence, obtain the proportion of factor importance corresponding to the soil stability and location movement hazard of the target slope in real-time working conditions. W 1 、W 2 Always satisfied W 1 + W 2 =1.
[0060] In the above method, in step 5, F = K x W 1 x X + W 2 ×Y , in, F To integrate early warning values, K is Synergistic influence coefficient W 1 and W 2 The proportion of factor importance and satisfying W 1 +W 2 =1，XThis represents the soil stability value. Y This represents the numerical value indicating the degree of danger associated with location relocation.
[0061] In the above method, in step 6, the early warning threshold is divided into multiple levels. The comprehensive early warning value is compared with the multi-level early warning threshold one by one to determine the instability risk status of the target slope. The corresponding early warning and handling operations include routine monitoring, risk warning, emergency prevention and control, and emergency response.
[0062] On the other hand, this disclosure also provides a slope instability early warning system based on big data analysis, used to execute the above-mentioned slope instability early warning method. The slope instability early warning system includes: a big data clustering modeling module, a model matching module, a data acquisition module, a time-series correlation analysis module, a fusion calculation module, and a hierarchical early warning execution module, wherein: the big data clustering modeling module is used to collect historical big data samples of slopes under multiple working conditions, cluster and divide slope type clusters, and generate special early warning models for various slopes; the model matching module is used to match the corresponding special early warning model according to the historical data of the target slope to obtain the running model; the data acquisition module is used to collect real-time soil data and location movement data of the target slope, and preprocess to obtain soil stability value and location movement hazard value; the time-series correlation analysis module is used to construct a two-factor time-series data sequence and analyze to obtain the synergistic influence coefficient of soil and location movement interaction; the fusion calculation module is used to iteratively optimize the initial weights according to the contribution of data false alarms, and calculate the comprehensive early warning value; the hierarchical early warning execution module is used to determine the instability risk level by comparing with multi-level early warning thresholds and execute corresponding operations for routine monitoring, risk early warning, emergency prevention and control, and emergency response.
[0063] The technical solution will be further explained below with reference to specific embodiments.
[0064] Example 1: The monitoring object was a deep-cut slope of a mountain highway. The slope was geologically stable, with no potential hazards induced by concentrated heavy rain. Its main characteristic of change was slow creep displacement over a long period, with no risk of sudden abrupt changes. Model matching determined it to be a long-term creep-type slope cluster. Based on the initial weighting rules, displacement risk was the dominant factor, and the initial weights were set as follows: soil stability weight was 0.35, and location movement hazard weight was 0.65. Data was collected in real time by on-site sensors. After anomaly removal, temporal noise reduction, and standardized quantization, the soil stability value was obtained. X =0.72, a numerical value indicating the degree of danger of location movement. Y =0.85. A continuous time-series data sequence was constructed, revealing a significant displacement hysteresis response and mutually reinforcing coupling pattern in the slope. The synergistic influence coefficient was quantified. K=1.12. Statistical monitoring period data: Soil data shows strong stability with no abnormal false alarms, and the contribution of false alarms is extremely low; displacement data shows slight abnormal fluctuations, and the contribution of false alarms is slightly higher; after multiple rounds of iterative optimization, the dynamic weights finally converged: W 1 =0.38, W 2 =0.62, which satisfies W 1 +W 2 =1. Substitute into the formula: F = K x W 1 x X + W 2 ×Y =0.834. In this slope model's multi-level threshold system, 0.8-0.9 represents the medium-risk warning range. The calculated value... F =0.834 falls within this range. The system automatically determines that there is a risk of creep accumulation and instability on the slope, triggers a risk warning, and executes measures such as intensified monitoring, manual slope inspection, and key displacement observation to intervene in the potential danger in advance.
[0065] Example 2: The monitoring object was a slope of slag heap in a mining area. The soil was loose and had poor impermeability. Heavy rainfall during the rainy season easily caused the soil to soften and become unstable, exhibiting sudden instability. Model matching determined it to be a cluster of slopes prone to sudden rainstorm events. Based on the initial weighting rules, soil stability was the dominant factor, and the initial weights were set as follows: soil stability weight was 0.65, and the risk of location movement weight was 0.35. Under heavy rain conditions, slope data was collected in real time, and after preprocessing and quantification, the soil stability value was obtained. X =0.36 (Severe soil deterioration), numerical value indicating the degree of risk of location movement. Y =0.40 (no obvious displacement has been observed yet). Under heavy rainfall conditions, soil weakening has a strong effect on displacement induction; the risk of dual-factor coupling is high; and the synergistic influence coefficient is quantified. K =1.25. Soil data under heavy rainfall conditions exhibits high sensitivity, accurate risk characterization, and no invalid false alarms; displacement data shows stable fluctuations, but risk characterization lags. After iterative optimization, the dynamic weights are updated as follows: W 1 =0.68, W 2 =0.32. Substitute into the core formula: F = K x W 1 × x X + W 2 ×Y=0.434. In the multi-level threshold of this slope model, 0.4~0.6 is the emergency control range. The calculated value F=0.434 falls into the emergency control range, and the system immediately triggers an emergency control early warning and automatically starts emergency response measures such as drainage, slope reinforcement, evacuation of on-site personnel, and equipment transfer.
[0066] In summary, compared with the prior art, this disclosure provides a slope instability early warning method and system based on big data analysis. The slope instability early warning method solves the problems of existing technologies such as single slope early warning model, poor adaptability to working conditions, one-sided risk assessment, and lack of dynamic optimization capability by constructing a closed-loop early warning process of multi-condition clustering modeling, dynamic parameter optimization, two-factor coupled calculation and hierarchical treatment.
[0067] The embodiments of this disclosure have now been described in detail. To avoid obscuring the concept of this disclosure, some details known in the art have not been described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein based on the above description.
[0068] While specific embodiments of this disclosure have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of this disclosure. Those skilled in the art should understand that modifications can be made to the above embodiments or equivalent substitutions can be made to some technical features without departing from the scope and spirit of this disclosure. In particular, as long as there is no structural conflict, the technical features mentioned in the various embodiments can be combined in any manner.
Claims
1. A slope instability early warning method based on big data analysis, characterized in that, Includes the following steps: Step 1: Collect a large amount of historical big data samples of slopes covering different working conditions. By performing cluster analysis on the historical big data samples, obtain multiple slope type clusters and corresponding early warning models for each slope type cluster. Step 2: Based on the historical data of the target slope, determine the slope type cluster to which the target slope belongs, select the exclusive early warning model corresponding to the slope type cluster to obtain the running model, and collect the soil data and location movement data of the target slope in real time. Input the soil data and location movement data into the running model to calculate the soil stability value and the location movement hazard value. Step 3: Construct a two-factor time series data sequence using the soil stability value and the location movement hazard value, and analyze the two-factor time series data sequence through big data correlation analysis to obtain the synergistic influence coefficient of soil and location movement interaction. Step 4: Based on the initial weights of the running model, and combined with the false alarm contribution of the soil data and the location movement data, iteratively modify the weights to obtain the factor importance ratios corresponding to soil stability and location movement hazard. Step 5: Using the synergistic influence coefficient and the factor importance ratio, perform a fusion calculation on the soil stability value and the location movement hazard value to obtain the comprehensive early warning value of the target slope; Step 6: Compare the comprehensive early warning value with the early warning threshold of the operating model, determine the instability risk status of the target slope based on the comparison result, and perform the corresponding level of early warning handling operation.
2. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 1, the K-means big data clustering algorithm is used to perform cluster analysis on the historical big data samples to obtain various slope type clusters, including: long-term creep type slope clusters and rainstorm sudden type slope clusters.
3. The slope instability early warning method based on big data analysis according to claim 2, characterized in that, The initial weight configuration of the dedicated early warning model corresponding to the long-term creep-type slope cluster is that the weight of the location movement hazard is greater than the weight of the soil stability; the initial weight configuration of the dedicated early warning model corresponding to the rainstorm-prone slope cluster is that the weight of the soil stability is greater than the weight of the location movement hazard.
4. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 2, Based on the slope type cluster data characteristics corresponding to the operating model, the real-time collected soil data is standardized and preprocessed to remove abnormal interference data. Combined with the soil mechanics association rules built into the operating model, the soil stability value that characterizes the overall stability performance of the slope soil is quantified. Simultaneously, the real-time acquired location movement data is subjected to time-series noise reduction processing. Combined with the slope displacement deformation response law adapted to the operating model, the location movement hazard level, which characterizes the level of slope slip risk, is obtained by quantitative calculation based on two dimensions: cumulative displacement and displacement change rate.
5. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 3, Using continuous sampling times as the time axis, the soil stability values and location movement risk values corresponding to different times are time-aligned and arranged to construct the synchronous and linked two-factor time-series data sequence; A big data correlation mining algorithm was used to perform temporal correlation analysis on the two-factor time series data sequence to explore the synchronous correlation, hysteresis response and mutual promotion coupling between soil stability performance decay and slope position movement and deformation. Based on the synchronous correlation law, the hysteresis response law, and the mutually reinforcing coupling law, a synergistic influence coefficient is quantified. The synergistic influence coefficient is used to characterize the interaction strength between soil state and slope displacement. The larger the coefficient value, the higher the linkage risk of slope instability induced by the coupling of the two factors.
6. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 4, During real-time monitoring, the frequency of abnormal fluctuations and the number of false alarms triggered by soil data and location movement data are statistically analyzed. The false alarm contribution of soil data and the false alarm contribution of location movement data are calculated respectively. The initial weights of the operating model are adjusted based on the false alarm contribution of soil data and the false alarm contribution of location movement data to obtain the factor importance ratios corresponding to soil stability and location movement hazard.
7. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 5, F=K×W 1 ×X+W 2 ×Y in, F To integrate early warning values, K is Synergistic influence coefficient W 1 and W 2 The proportion of factor importance and satisfying W 1 +W 2 =1， X This represents the soil stability value. Y This represents the numerical value indicating the degree of danger associated with location relocation.
8. The slope instability early warning method based on big data analysis according to claim 1, characterized in that, In step 6, the early warning threshold is divided into multiple levels. The comprehensive early warning value is compared with the multi-level early warning threshold one by one to determine the instability risk status of the target slope. The corresponding early warning and handling operations include routine monitoring, risk warning, emergency prevention and control, and emergency response.
9. A slope instability early warning system based on big data analysis, characterized in that, The slope instability early warning system, used to execute the slope instability early warning method according to any one of claims 1-8, comprises: a big data clustering modeling module, a model matching module, a data acquisition module, a time-series correlation analysis module, a fusion calculation module, and a hierarchical early warning execution module, wherein: Big Data Clustering Modeling Module: Used to collect historical big data samples of slopes under multiple working conditions, cluster and divide slope type clusters, and generate special early warning models for various slopes; Model matching module: Matches the corresponding dedicated early warning model based on the historical data of the target slope to obtain the running model; Data acquisition module: used to collect real-time soil data and location movement data of the target slope, and preprocess the data to obtain soil stability value and location movement hazard value; Temporal correlation analysis module: used to construct two-factor time-series data sequences and analyze the synergistic influence coefficient of soil-location interaction; Fusion calculation module: Iteratively optimizes the initial weights based on the contribution of data false alarms, and calculates the comprehensive early warning value by weighting. Tiered early warning execution module: Determines the level of instability risk by comparing with multi-level early warning thresholds, and executes corresponding operations for routine monitoring, risk warning, emergency prevention and control, and disaster relief.