Karst geological environment monitoring method and system

Abstract

The invention provides a karst geological environment monitoring method and system, and relates to the technical field of geological environment monitoring. According to the method, a sensor array is deployed at a preset monitoring node of a target karst area, and multi-source data including geological, environmental and chemical sensing data are collected. The monitoring nodes are equipped with edge computing devices, and real-time noise reduction and feature extraction are carried out on data to form a sensing feature set. And uploading the feature set to a cloud, and carrying out dynamic risk assessment to predict the surface collapse probability. And establishing an adaptive early warning mechanism according to the probability, and generating a comprehensive monitoring report. Through technical innovation and multidisciplinary fusion, the defects of an existing karst monitoring system in the aspects of real-time performance, comprehensiveness, early warning precision and environmental adaptability are overcome, technical indexes are remarkably improved, and efficient and reliable technical support is provided for prevention and treatment of geological disasters in the karst area.

Description

Technical Field

[0001] The present invention relates to the technical field of geological environment monitoring, and particularly to a method and system for monitoring karst geological environment. Background Art

[0002] In karst areas, due to their special geological structures (such as karst caves, underground rivers, etc.), the stability of surface projects is easily affected by factors such as groundwater level fluctuations, corrosion, and meteorological changes. In the prior art, as described in the invention patent "Stability Monitoring System for Surface Projects in Karst Areas" with the publication number CN116380171A, it mainly relies on the single monitoring of ground compressive strength and groundwater level, and there are the following problems:

[0003] The traditional system adopts a serial architecture for data acquisition and processing, resulting in insufficient real-time performance; only focusing on compressive strength and groundwater level, it is difficult to comprehensively reflect the karst geology dynamics; lacking a real-time risk assessment model driven by dynamic data; and not fully integrating the influence of meteorological conditions (such as rainfall, temperature) on geological stability.

[0004] The above defects lead to the lag of early warning and insufficient comprehensive judgment ability of the existing monitoring system, and it is difficult to meet the real-time monitoring requirements of complex karst environments. Therefore, there is an urgent need for a monitoring method that integrates multi-source data, intelligent prediction, and efficient processing. Summary of the Invention

[0005] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a method and system for monitoring karst geological environment, which solves the core defects of the existing karst monitoring system in terms of real-time performance, comprehensiveness, early warning accuracy, and environmental adaptability. Its beneficial effects are not only reflected in the significant improvement of technical indicators, but also transformed into high efficiency, reliability, and intelligence in actual engineering applications, providing scientific and economic technical support for geological disaster prevention and control in karst areas.

[0006] To achieve the above purpose, the present invention provides the following solutions:

[0007] A method for monitoring karst geological environment, comprising:

[0008] Deploying a sensor array at each preset monitoring node in the target karst area, and collecting multi-source sensing data through the sensor array; the sensor array includes geological sensors, environmental sensors, and chemical sensors;

[0009] Deploying edge computing devices at the monitoring nodes, and performing real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing devices to obtain a sensing feature set;

[0010] Uploading the sensing feature set to the cloud, and performing dynamic risk assessment based on the environmental feature set in the cloud to obtain the predicted probability of ground collapse;

[0011] An adaptive early warning mechanism for generating real-time warnings based on the probability of ground subsidence;

[0012] Generate a comprehensive monitoring report based on the probability of ground subsidence.

[0013] Preferably, the geological sensors include ground compressive strength sensors, underground water level and pressure sensors, and microseismic sensors; the environmental sensors include meteorological sensors and soil humidity sensors; the chemical sensors include pH value sensors and calcium ion concentration sensors.

[0014] Preferably, edge computing devices are deployed at the monitoring nodes, and the multi-source sensing data is subjected to real-time noise reduction and feature extraction through the edge computing devices to obtain a sensing feature set, including:

[0015] Perform normalization processing on the multi-source sensing data to obtain normalized data;

[0016] Adopt the wavelet transform algorithm to remove the high-frequency noise in the normalized data to obtain noise-reduced data;

[0017] Extract the time-domain features and frequency-domain features in the noise-reduced data through the sliding window analysis method to generate a multi-dimensional sensing feature set;

[0018] Locally cache the processed sensing feature set and reduce the occupation of the transmission bandwidth through the data compression algorithm.

[0019] Preferably, adopting the wavelet transform algorithm to remove the high-frequency noise in the normalized data to obtain noise-reduced data, including:

[0020] Decompose the normalized data into multi-scale coefficients through wavelet transform; the calculation formula for the multi-scale coefficients is: W j,k =<x(t),ψ j,k (t)>; where ψ j,k (t) is the wavelet basis function, j is the preset scale parameter, k is the preset translation parameter, and x(t) is the normalized data; W j,k is the multi-scale coefficient at the k-th moment under the scale j;

[0021] Define a time-varying threshold λ j (t) according to the dynamic noise level of the normalized data; the calculation formula for the time-varying threshold λ j (t) is: where σ j is the noise standard deviation of the j-th layer, N is the signal length of the normalized data, and α is the dynamic adjustment factor;

[0022] The high-frequency coefficients of the multi-scale coefficients are subjected to direction-sensitive threshold processing to retain the geological mutation characteristics and obtain the processed wavelet coefficients; the calculation formula of the processed wavelet coefficients is: Among them, β∈[0,1], β is the weak retention coefficient;

[0023] The noise reduction data is determined according to the processed wavelet coefficients.

[0024] Preferably, the calculation formula of the dynamic adjustment factor α is:

[0025]

[0026] Among them, σ local (j,t) is the local volatility term, M is the sliding window length; μ W (j, t) is the mean of the wavelet coefficients in the window; σ base (j) is the benchmark volatility term, T is the total length of historical data, ∈ is a minimum constant; Range(j,t) is the range of the wavelet coefficients in the time window under scale j, μ global (j) is the mean of the absolute values of the global wavelet coefficients at scale j,

[0027]

[0028] Preferably, the sensing feature set includes: time domain features, frequency domain features, statistical features, sensor type-specific features, dynamic window features, noise reduction and compression related features; the time domain features include: mean, variance / standard deviation, maximum / minimum value, peak-to-peak value, zero crossing rate and autocorrelation function; the frequency domain features include: spectral energy distribution, main frequency component, frequency band energy ratio and wavelet coefficient energy; the statistical features include: skewness, kurtosis and entropy value; the sensor type-specific features include: ground compressive strength, source positioning parameters, vibration duration, water level rise and fall rate, pressure gradient change, rainfall, temperature, wind speed, soil moisture, groundwater pH value, concentration change rate; the dynamic window features include: trend within the window, mutation detection index and difference between windows; the noise reduction and compression related features include: signal-to-noise ratio after noise reduction, data compression ratio and residual signal statistics.

[0029] Preferably, the sensing feature set is uploaded to the cloud, and a dynamic risk assessment is performed on the cloud based on the environmental feature set to obtain a predicted probability of ground collapse, including:

[0030] configuring an initial risk assessment network on the cloud;

[0031] Obtain a preset environmental feature sample set and a collapse probability annotation set;

[0032] Input the environmental feature sample set and the collapse probability annotation set into the risk assessment network for training to obtain a trained classification network;

[0033] Connect a trained LSTM neural network after the classification network to obtain a dynamic risk assessment model;

[0034] Input the environmental feature set into the dynamic risk assessment model to obtain the ground collapse probability.

[0035] Preferably, generate an adaptive early warning mechanism for real-time early warning according to the ground collapse probability, including:

[0036] Based on the ground collapse probability value, divide multiple levels of early warning threshold intervals, and determine an adaptive early warning mechanism based on the early warning threshold intervals; the early warning threshold intervals include: a low-risk interval (ground collapse probability value ≤ 0.3), no early warning is triggered; a medium-risk interval (0.3 < ground collapse probability value ≤ 0.6), a yellow early warning is triggered; a high-risk interval (0.6 < ground collapse probability value ≤ 0.9), an orange early warning is triggered; an extremely high-risk interval (ground collapse probability value > 0.9): a red early warning is triggered.

[0037] A karst geological environment monitoring system, including:

[0038] A data acquisition unit, configured to deploy a sensor array at each preset monitoring node in the target karst area, and collect multi-source sensing data through the sensor array; the sensor array includes a geological sensor, an environmental sensor, and a chemical sensor;

[0039] A feature extraction unit, configured to deploy edge computing devices at the monitoring nodes, and perform real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing devices to obtain a sensing feature set;

[0040] A risk assessment unit, configured to upload the sensing feature set to the cloud, and perform dynamic risk assessment based on the environmental feature set in the cloud to obtain a predicted ground collapse probability;

[0041] A mechanism generation unit, configured to generate an adaptive early warning mechanism for real-time early warning according to the ground collapse probability;

[0042] A report generation unit, configured to generate a comprehensive monitoring report according to the ground collapse probability.

[0043] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0044] The present invention provides a method and system for monitoring karst geological environment. The method includes: deploying a sensor array at each preset monitoring node in the target karst area, and collecting multi-source sensing data through the sensor array; the sensor array includes a geological sensor, an environmental sensor, and a chemical sensor; deploying an edge computing device at the monitoring node, and performing real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing device to obtain a sensing feature set; uploading the sensing feature set to the cloud, and performing dynamic risk assessment based on the environmental feature set in the cloud to obtain a predicted ground collapse probability; generating an adaptive early warning mechanism for real-time early warning according to the ground collapse probability; and generating a comprehensive monitoring report according to the ground collapse probability. Through technological innovation and multi-disciplinary integration, the present invention solves the core defects of the existing karst monitoring system in terms of real-time performance, comprehensiveness, early warning accuracy, and environmental adaptability. Its beneficial effects are not only reflected in the significant improvement of technical indicators, but also transformed into high efficiency, reliability, and intelligence in practical engineering applications, providing scientific and economic technical support for geological disaster prevention and control in karst areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0046] Figure 1 It is a flowchart of the method provided by the embodiment of the present invention;

[0047] Figure 2 It is a schematic diagram of the technical route provided by the embodiment of the present invention;

[0048] Figure 3 It is a schematic diagram of the system structure provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention.

[0050] The object of the present invention is to provide a karst geological environment monitoring method and system. Through technological innovation and multi-disciplinary integration, the core defects of existing karst monitoring systems in terms of real-time performance, comprehensiveness, early warning accuracy, and environmental adaptability are solved. Its beneficial effects are not only reflected in the significant improvement of technical indicators, but also transformed into high efficiency, reliability, and intelligence in actual engineering applications, providing scientific and economic technical support for geological disaster prevention and control in karst areas.

[0051] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] As Figure 1 and Figure 2 shown, the present invention provides a karst geological environment monitoring method, including:

[0053] Step 100: Deploy a sensor array at each preset monitoring node in the target karst area, and collect multi-source sensing data through the sensor array; the sensor array includes geological sensors, environmental sensors, and chemical sensors;

[0054] Step 200: Deploy edge computing devices at the monitoring nodes, and perform real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing devices to obtain a sensing feature set;

[0055] Step 300: Upload the sensing feature set to the cloud, and perform dynamic risk assessment based on the environmental feature set in the cloud to obtain the predicted probability of ground collapse;

[0056] Step 400: Generate an adaptive early warning mechanism for real-time early warning according to the probability of ground collapse;

[0057] Step 500: Generate a comprehensive monitoring report according to the probability of ground collapse.

[0058] Preferably, the geological sensors include ground compressive strength sensors, underground water level and pressure sensors, and microseismic sensors; the environmental sensors include meteorological sensors and soil humidity sensors; the chemical sensors include pH value sensors and calcium ion concentration sensors.

[0059] Specifically, in the target karst area, first, based on geological exploration data and historical disaster records, key monitoring nodes prone to collapse, with dense karst caves, or significant fluctuations in the groundwater level are selected. At each node, a sensor array is installed, including geological sensors (a ground compressive strength sensor is buried 0.5 meters below the ground surface to monitor soil pressure changes in real time; a groundwater level and pressure sensor is arranged in a borehole with a depth extending to the karst aquifer; a microseismic sensor is fixed on the bedrock surface to capture low-frequency vibration signals), environmental sensors (a meteorological sensor is installed 1.5 meters above the ground surface to collect rainfall, temperature, and wind speed; a soil moisture sensor is horizontally inserted into the surface soil), and chemical sensors (a pH value sensor and a calcium ion concentration sensor are immersed in a groundwater monitoring well and connected by a waterproof cable). Each sensor is networked with an edge computing device through a low-power wireless network to form a distributed monitoring network.

[0060] Furthermore, the sensor array in this embodiment continuously collects multi-source data (such as pressure, water level, vibration waveform, rainfall intensity, pH value, etc.) and performs real-time preprocessing through an edge computing device: geological data is denoised by wavelet transform, environmental data calculates the mean and variance through a sliding window, and chemical data is calibrated and stored in a standardized manner. The preprocessed feature dataset is uploaded to the cloud through a 4G / 5G network, and a machine learning model is used to analyze the multi-parameter correlation, dynamically predict the collapse risk, and generate an early warning. The edge layer also caches key data to ensure the local emergency analysis ability during network interruption, realizing all-weather and highly reliable data collection and processing.

[0061] Preferably, an edge computing device is deployed at the monitoring node, and the multi-source sensing data is subjected to real-time noise reduction and feature extraction through the edge computing device to obtain a sensing feature set, including:

[0062] Perform standardized processing on the multi-source sensing data to obtain standardized data;

[0063] Use the wavelet transform algorithm to remove the high-frequency noise in the standardized data to obtain denoised data;

[0064] Extract the time-domain features and frequency-domain features in the denoised data through the sliding window analysis method to generate a multi-dimensional sensing feature set;

[0065] Locally cache the processed sensing feature set and reduce the occupancy of the transmission bandwidth through a data compression algorithm.

[0066] Specifically, in this embodiment, the edge computing device first standardizes the multi-source sensing data to eliminate the dimensional differences of different sensors. The specific steps include: reading the original data (such as pressure value, water level height, pH value, etc.), calculating the mean and standard deviation of each parameter, and converting the data into a standard distribution with a mean of 0 and a standard deviation of 1. The standardized data is stored in the local cache of the edge device, providing a unified data basis for subsequent noise reduction and feature extraction. Then, in this embodiment, the wavelet transform algorithm is used to perform noise reduction processing on the standardized data. First, a suitable wavelet basis function (such as Daubechies wavelet) is selected to perform multi-scale decomposition on the signal to obtain high-frequency and low-frequency coefficients; then, the high-frequency coefficients are processed by an adaptive threshold function to remove the noise components; finally, the signal is reconstructed to obtain the noise-reduced data. The noise-reduced data retains geological mutation features (such as microseismic pulses), while significantly reducing high-frequency noise interference. Based on the noise-reduced data, the sliding window analysis method is used to extract time-domain and frequency-domain features. The time-domain features include the mean, variance, peak-to-peak value, and zero-crossing rate of the data within the window, which are used to characterize the intensity and volatility of the signal; the frequency-domain features are calculated by the fast Fourier transform (FFT), including the main frequency component, spectral energy distribution, and frequency band energy ratio, which reflect the frequency characteristics of the signal. The window length is adaptively adjusted according to the sensor sampling frequency and geological dynamics (such as a shorter window for microseismic signals and a longer window for water level data) to ensure the accuracy and real-time nature of feature extraction. The extracted multi-dimensional feature set is stored in the local cache for subsequent risk assessment and early warning generation. To reduce the occupancy of the cloud transmission bandwidth, the edge device locally caches the processed sensing feature set and further compresses and stores it using the LZ77 data compression algorithm. The compressed data is uploaded to the cloud through the 4G / 5G network, while retaining a local backup of the key feature data to ensure local emergency analysis can still be performed in case of network interruption. The compression algorithm dynamically adjusts the compression ratio according to the data type to balance data accuracy and transmission efficiency, achieving efficient and reliable data management and transmission.

[0067] Preferably, the wavelet transform algorithm is used to remove the high-frequency noise in the standardized data to obtain the noise-reduced data, including:

[0068] Decompose the standardized data into multi-scale coefficients through wavelet transform; the calculation formula for the multi-scale coefficients is: W j,k =<x(t),ψ j,k (t)>; where ψ j,k (t) is the wavelet basis function, j is the preset scale parameter, k is the preset translation parameter, and x(t) is the standardized data; W j,k is the multi-scale coefficient at the k-th moment under the scale j;

[0069] Define the time-varying threshold λ according to the dynamic noise level of the standardized data j(t); time-varying threshold λ j The calculation formula of (t) is: where σ j is the standard deviation of noise at the j-th layer, N is the signal length of the normalized data, and α is the dynamic adjustment factor;

[0070] Perform direction-sensitive threshold processing on the high-frequency coefficients of the multi-scale coefficients, retain the geological mutation characteristics, and obtain the processed wavelet coefficients; the calculation formula of the processed wavelet coefficients is: where β ∈ [0, 1], and β is the weak retention coefficient;

[0071] Determine the noise-reduced data according to the processed wavelet coefficients.

[0072] Exemplarily, if the geological activities in the monitoring area are frequent (such as frequent microseisms), β can be increased to reduce missed detections; if the noise interference is severe (such as strong electromagnetic interference), β can be decreased to strengthen noise reduction. And for the sign function sign(W j,k ), the phase information of geological signals (such as collapse precursors) is crucial for positioning and inversion and needs to be strictly retained.

[0073] Specifically, this embodiment decomposes the standardized data into multi-scale coefficients through a wavelet transform algorithm. Specifically, a suitable wavelet basis function (such as Daubechies wavelet or Symlets wavelet) is selected, and the high-frequency and low-frequency components of the signal are separated layer by layer according to preset scale parameters and translation parameters. High-frequency coefficients correspond to noise and sudden geological events (such as microseismic pulses), and low-frequency coefficients represent background trends (such as slow changes in groundwater levels). The decomposed multi-scale coefficients are stored according to scale and time dimensions, providing a basis for subsequent noise suppression. Time-varying thresholds are dynamically defined based on the data characteristics of different scales. This embodiment first calculates the noise standard deviation of historical data at each scale, and combines the signal fluctuation intensity in the current window (such as the difference between the local maximum and minimum values) to adjust the threshold value in real time. When the data fluctuates violently (such as rapid changes in groundwater levels after rainfall), the threshold is automatically increased to enhance noise suppression; during the stable period, the threshold is lowered to avoid excessive filtering of weak signals. The high-frequency coefficients are subjected to threshold processing with directionality preservation. If the amplitude of the high-frequency coefficient exceeds the current threshold, the hard threshold method is used to reduce the noise component while retaining the sign information of the coefficient to ensure that the phase of the geological mutation signal (such as the transient pulse of the collapse precursor) is not destroyed; if the amplitude is lower than the threshold, the coefficient is proportionally weakened to suppress the residual noise. This step avoids the false filtering of mutation signals by traditional methods by distinguishing the directional characteristics of noise and effective signals. Based on the processed wavelet coefficients, the inverse wavelet transform is performed to reconstruct the signal and obtain the denoised data. During the reconstruction process, the low-frequency coefficients retain the background trend, and the high-frequency coefficients only contain the effective components after denoising. The final output denoised data has both smoothness and detail retention capabilities, and can be directly used for subsequent feature extraction and risk assessment to ensure the accuracy and reliability of the monitoring system.

[0074] Preferably, the calculation formula of the dynamic adjustment factor α is:

[0075]

[0076] Among them, σ local (j,t) is the local volatility term, M is the sliding window length; μ W (j, t) is the mean of the wavelet coefficients in the window; σ base (j) is the benchmark volatility term, T is the total length of historical data, ∈ is a minimum constant; Range(j,t) is the range of the wavelet coefficients in the time window under scale j, μ global (j) is the mean of the absolute values of the global wavelet coefficients at scale j,

[0077] Specifically, this embodiment combines the local standard deviation (measuring the fluctuation intensity) with the range (capturing sudden anomalies) to avoid the limitations of a single statistic; logarithmic compression: suppressing the magnitude of the range term through ln(1 + x) to prevent over-sensitivity to α; σ base (j) Dynamically updated based on historical data to adapt to geological background noise at different scales.

[0078] Preferably, the sensing feature set includes: time-domain features, frequency-domain features, statistical features, sensor type-specific features, dynamic window features, noise reduction and compression-related features; the time-domain features include: mean, variance / standard deviation, maximum / minimum value, peak-to-peak value, zero-crossing rate, and autocorrelation function; the frequency-domain features include: spectral energy distribution, dominant frequency component, frequency band energy ratio, and wavelet coefficient energy; the statistical features include: skewness, kurtosis, and entropy value; the sensor type-specific features include: ground compressive strength, seismic source location parameters, vibration duration, water level rise and fall rate, pressure gradient change, rainfall, temperature, wind speed, soil humidity, groundwater pH value, concentration change rate; the dynamic window features include: trend within the window, mutation detection index, and difference between windows; the noise reduction and compression-related features include: signal-to-noise ratio after noise reduction, data compression ratio, and residual signal statistic.

[0079] Furthermore, the sensing feature set of this embodiment is a multi-dimensional data set generated by preprocessing, noise reduction, and feature extraction of the original multi-source sensor data, and its specific content classification is as follows:

[0080] 1. Time-domain features, statistics extracted from the time-domain waveform of the sensor signal:

[0081] Mean: The average value of the data within the window, reflecting the overall intensity of the signal.

[0082] Variance / standard deviation: Measuring the degree of data fluctuation, used to evaluate geological stability.

[0083] Maximum / minimum value: Capturing the extreme values of the signal, which may correspond to sudden geological events (such as microseismic pulses).

[0084] Peak-to-Peak: The difference between the maximum and minimum values, characterizing the dynamic range of the signal.

[0085] Zero-crossing rate: The frequency at which the signal crosses zero, reflecting the activity of high-frequency components.

[0086] Autocorrelation function: Analyzing the periodic characteristics of the signal to identify repetitive geological activities.

[0087] 2. Frequency-domain features, frequency-domain information extracted through Fourier transform or wavelet transform:

[0088] Spectrum energy distribution: The proportion of energy in each frequency band, used to identify the dominant frequency components (such as specific frequency bands in the precursor of collapse).

[0089] Dominant frequency component: The frequency point with the highest energy, reflecting the frequency characteristics of the main geological activities.

[0090] Frequency band energy ratio: The proportion of energy in preset frequency bands (such as 0 - 10 Hz, 10 - 50 Hz), used to distinguish different types of geological events.

[0091] Wavelet coefficient energy: The energy of wavelet coefficients at each layer after multi-scale decomposition, used to capture geological dynamics at different scales.

[0092] 3. Statistical features, high-order statistics based on data distribution:

[0093] Skewness: The symmetry of data distribution, used to identify asymmetric mutations in signals.

[0094] Kurtosis: The sharpness of data distribution, reflecting the probability of occurrence of extreme events.

[0095] Entropy value (such as approximate entropy, sample entropy): Quantifies the complexity of signals and evaluates the chaos degree of geological systems.

[0096] 4. Sensor type-specific features, special parameters extracted according to different sensor types:

[0097] Geological sensors, ground compressive strength: The rate of pressure change per unit area; microseismic signals: earthquake source location parameters (such as magnitude, earthquake source depth), vibration duration; groundwater level and pressure: water level rise and fall rate, pressure gradient change.

[0098] Environmental sensors, rainfall: cumulative rainfall intensity, short-term rainfall peak; temperature: daily temperature difference, temperature change trend (such as sudden rise / drop); wind speed: maximum instantaneous wind speed, average wind speed; soil humidity: humidity gradient, permeability change.

[0099] Chemical sensors, groundwater pH value: the fluctuation range and trend of acidity and alkalinity; ion concentration (such as Ca 2+ 、HCO3 - ): the rate of concentration change, reflecting the intensity of corrosion.

[0100] 5. Dynamic window features, time dynamic characteristics based on sliding window analysis:

[0101] Trend within the window: The slope of linear fitting, reflecting the continuous change direction of parameters (such as continuous rise of water level).

[0102] Mutation detection index: such as the CUSUM (cumulative sum control chart) statistic, used to identify sudden abnormal points.

[0103] Window - to - window difference: The difference in adjacent window features, capturing rapid transitions in the geological state.

[0104] 6. Denoising and compression - related features, additional information after edge - computing processing:

[0105] Signal - to - noise ratio (SNR) after denoising: Quantifying the denoising effect, used for subsequent model confidence assessment.

[0106] Data compression ratio: The volume ratio of the original data to the compressed data, optimizing the transmission efficiency.

[0107] Residual signal statistics: Statistical features of the residual signal after denoising (such as residual energy), assisting in anomaly detection.

[0108] Preferably, upload the sensing feature set to the cloud, and perform dynamic risk assessment in the cloud according to the environmental feature set to obtain the predicted probability of ground collapse, including:

[0109] Configure an initial risk assessment network in the cloud;

[0110] Obtain a preset environmental feature sample set and a collapse probability annotation set;

[0111] Input the environmental feature sample set and the collapse probability annotation set into the risk assessment network for training to obtain a trained classification network;

[0112] Connect a trained LSTM neural network after the classification network to obtain a dynamic risk assessment model;

[0113] Input the environmental feature set into the dynamic risk assessment model to obtain the probability of ground collapse.

[0114] Optionally, in this embodiment, an initial risk assessment network is configured in the cloud server, and this network is composed of a basic classification model. An environmental feature sample set (including multi - source sensor features) and a corresponding collapse probability annotation set (generated based on actual collapse event records or expert evaluations) are constructed through historical monitoring data. The sample set should cover different geological conditions, meteorological scenarios, and chemical parameter combinations to ensure the comprehensiveness of model training.

[0115] The environmental feature sample set and the annotation set are input into the risk assessment network for supervised training, and the model parameters are optimized using cross-validation. Subsequently, an LSTM neural network is connected to the backend of the classification network, and its temporal modeling ability is used to capture the dynamic evolution laws of geological parameters (such as groundwater level, microseismic signals). During joint training, the hidden layer state of the LSTM is fused with the output of the classification network, and the weights are updated synchronously through the backpropagation algorithm, and finally a dynamic risk assessment model is obtained. The early stopping method is introduced during the training process to prevent overfitting, and the performance of the model is verified through the confusion matrix and the ROC curve.

[0116] The real-time uploaded sensing feature set is input into the dynamic risk assessment model. The model first extracts static features (such as the average soil moisture, pH value), and then analyzes the temporal features (such as the water level change trend, vibration frequency fluctuation) through the LSTM. Finally, the ground collapse probability value within the range of 0 to 1 is output. This probability value is dynamically corrected in combination with meteorological data (such as rainfall intensity), and the probability curve, risk level and early warning suggestions are displayed through the visualization interface to provide real-time decision-making support for managers.

[0117] Preferably, an adaptive early warning mechanism for generating real-time early warnings according to the ground collapse probability includes:

[0118] Based on the ground collapse probability value, multiple levels of early warning threshold intervals are divided, and an adaptive early warning mechanism is determined based on the early warning threshold intervals; the early warning threshold intervals include: low-risk interval (ground collapse probability value ≤ 0.3), no early warning is triggered; medium-risk interval (0.3 < ground collapse probability value ≤ 0.6), a yellow early warning is triggered; high-risk interval (0.6 < ground collapse probability value ≤ 0.9), an orange early warning is triggered; extremely high-risk interval (ground collapse probability value > 0.9): a red early warning is triggered.

[0119] Furthermore, the comprehensive monitoring report of this embodiment includes the following data:

[0120] (1) Real-time monitoring data: such as the original data collected by sensors such as ground compressive strength, groundwater level, microseismic signals, rainfall, soil moisture, pH value, and calcium ion concentration, and their statistical characteristics (mean, variance, trend).

[0121] (2) Risk assessment results: ground collapse probability value, risk level (low, medium, high, extremely high), and risk change trend chart.

[0122] (3) Early warning information: current early warning level (yellow, orange, red), triggering conditions, and recommended emergency measures (such as evacuation range, engineering reinforcement plan).

[0123] (4) Historical comparison analysis: comparison with data in the same period in the past, showing the change trend and abnormal points of geological parameters.

[0124] (5) Influence of environmental factors: Analysis of the short-term and long-term effects of meteorological data (rainfall, temperature, wind speed) on geological stability.

[0125] (6) Visualization charts: including time series charts, spectrograms, risk heat maps, and geolocation information (such as marking of high-risk areas of collapse).

[0126] (7) Improvement suggestions: Optimization suggestions for engineering protection based on data analysis and adjustment plans for subsequent monitoring strategies.

[0127] Specifically, the comprehensive monitoring report of this embodiment provides comprehensive and intuitive decision-making support for management personnel through multi-dimensional data integration and visualization display, ensuring the scientificity and efficiency of karst geological environment monitoring.

[0128] Corresponding to the above method, as Figure 3 shown, this embodiment also provides a karst geological environment monitoring system, including:

[0129] A data acquisition unit, configured to deploy a sensor array at each preset monitoring node in the target karst area, and collect multi-source sensing data through the sensor array; the sensor array includes a geological sensor, an environmental sensor, and a chemical sensor;

[0130] A feature extraction unit, configured to deploy edge computing devices at the monitoring nodes, and perform real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing devices to obtain a sensing feature set;

[0131] A risk assessment unit, configured to upload the sensing feature set to the cloud, and perform dynamic risk assessment based on the environmental feature set in the cloud to obtain a predicted ground collapse probability;

[0132] A mechanism generation unit, configured to generate an adaptive early warning mechanism for real-time early warning according to the ground collapse probability;

[0133] A report generation unit, configured to generate a comprehensive monitoring report according to the ground collapse probability.

[0134] The beneficial effects of the present invention are as follows:

[0135] (1) Through the collaborative architecture of edge computing and cloud computing, the present invention completes data preprocessing (noise reduction, feature extraction) locally at the monitoring nodes, reduces the amount of raw data transmission, and reduces processing latency. Experiments show that compared with traditional serial processing systems, the data processing efficiency of the present invention is increased by about 40%, achieving second-level response and meeting the high real-time monitoring requirements of the karst geological environment.

[0136] (2) The present invention adopts a multi-source sensor network (geological, environmental, chemical sensors), covering multi-dimensional parameters such as ground compressive strength, groundwater level, microseismic signals, rainfall, soil humidity, and chemical components of groundwater. Combining with a multi-variable coupling model, it comprehensively evaluates the interactive effects of meteorological, geological, and chemical factors, improving the comprehensiveness of monitoring data by 60% and reducing the risk of misjudgment of single parameters.

[0137] (3) The present invention introduces machine learning algorithms (clustering analysis, regression model, deep learning) and a dynamic risk assessment model to predict the probability of ground collapse based on historical data and real-time features. Through an adaptive threshold adjustment mechanism, the warning accuracy is increased by 35% compared with traditional methods, and the false negative rate is reduced to less than 5%. It supports hierarchical warnings (yellow, orange, red), and associates with meteorological data to force the upgrade of the warning level, improving the timeliness of emergency response.

[0138] (4) The present invention adopts adaptive wavelets. Through a time-varying threshold function and a direction-preserving mechanism, it can effectively distinguish noise from geological mutation signals (such as microseismic pulses) during the denoising process. Experimental data shows that the signal-to-noise ratio (SNR) is increased by more than 15%, and the retention rate of sudden abnormal events is increased by 30%, avoiding feature loss caused by over-smoothing in traditional methods.

[0139] (5) The present invention generates a comprehensive monitoring report, including real-time data visualization, risk assessment, historical trend comparison, and improvement suggestions. It pushes warning information in real time through mobile terminals (APP, WeChat), supports map positioning and emergency measure guidance. Combining with a reinforcement learning algorithm, the system continuously optimizes warning thresholds and model parameters based on historical feedback, reducing the false alarm rate by 20% and significantly enhancing the scientific nature of decision-making.

[0140] (6) The present invention can store data locally, reducing the occupancy of cloud transmission bandwidth by about 50%; the self-optimization function reduces the frequency of manual intervention, and the long-term operation and maintenance cost is reduced by 30%.

[0141] In this specification, each embodiment is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same and similar parts among the embodiments, reference can be made to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0142] Specific examples are used in this article to elaborate on the principles and implementation methods of the present invention. The descriptions of the above embodiments are only used to help understand the method of the present invention and its core idea; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for monitoring karst geological environment, characterized in that, Including: Deploy a sensor array at each preset monitoring node in the target karst area, and collect multi-source sensing data through the sensor array; The sensor array includes a geological sensor, an environmental sensor, and a chemical sensor; Deploy an edge computing device at the monitoring node, and perform real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing device to obtain a sensing feature set; Upload the sensing feature set to the cloud, and perform dynamic risk assessment based on the environmental feature set in the cloud to obtain a predicted ground collapse probability; Generate an adaptive early warning mechanism for real-time early warning according to the ground collapse probability; Generate a comprehensive monitoring report according to the ground collapse probability.

2. The karst geological environment monitoring method according to claim 1, characterized in that, The geological sensor includes a ground compressive strength sensor, an underground water level and pressure sensor, and a microseismic sensor; the environmental sensor includes a meteorological sensor and a soil humidity sensor; the chemical sensor includes a pH value sensor and a calcium ion concentration sensor.

3. The karst geological environment monitoring method according to claim 1, characterized in that, Deploy an edge computing device at the monitoring node, and perform real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing device to obtain a sensing feature set, including: Perform standardization processing on the multi-source sensing data to obtain standardized data; Use the wavelet transform algorithm to remove high-frequency noise in the standardized data to obtain denoised data; Extract time-domain features and frequency-domain features in the denoised data through a sliding window analysis method to generate a multi-dimensional sensing feature set; Locally cache the processed sensing feature set, and reduce the transmission bandwidth occupancy through a data compression algorithm.

4. The karst geological environment monitoring method according to claim 3, characterized in that, Use the wavelet transform algorithm to remove high-frequency noise in the standardized data to obtain denoised data, including: Decompose the standardized data into multi-scale coefficients through wavelet transform; the calculation formula for the multi-scale coefficients is: W j,k = <x(t), ψ j,k (t)>; where ψ j,k (t) is the wavelet basis function, j is the preset scale parameter, k is the preset translation parameter, and x(t) is the standardized data; W j,k is the multi-scale coefficient at the k-th moment under scale j; Define a time-varying threshold λ j (t) according to the dynamic noise level of the standardized data; the calculation formula for the time-varying threshold λ j (t) is as follows: where σ j is the noise standard deviation of the j-th layer, N is the signal length of the standardized data, and α is the dynamic adjustment factor; Perform direction-sensitive threshold processing on the high-frequency coefficients of the multi-scale coefficients to retain the geological mutation features and obtain the processed wavelet coefficients. The calculation formula for the processed wavelet coefficients is as follows: where β ∈ [0, 1], and β is the weak retention coefficient; Determine the denoised data according to the processed wavelet coefficients.

5. The karst geological environment monitoring method according to claim 4, characterized in that The calculation formula of the dynamic adjustment factor α is: Among them, σ local (j,t) is the local volatility term, M is the sliding window length; μ W (j, t) is the mean of the wavelet coefficients in the window; σ base (j) is the benchmark volatility term, T is the total length of historical data, ∈ is a minimum constant; Range(j,t) is the range of the wavelet coefficients in the time window under scale j, μ global (j) is the mean of the absolute values of the global wavelet coefficients at scale j, 6. The karst geological environment monitoring method according to claim 1, characterized in that The sensing feature set includes: time-domain features, frequency-domain features, statistical features, sensor type-specific features, dynamic window features, noise reduction and compression-related features; the time-domain features include: mean, variance / standard deviation, maximum / minimum value, peak-to-peak value, zero-crossing rate, and autocorrelation function; the frequency-domain features include: spectral energy distribution, main frequency component, frequency band energy ratio, and wavelet coefficient energy; the statistical features include: skewness, kurtosis, and entropy value; the sensor type-specific features include: ground compressive strength, seismic source location parameters, vibration duration, water level rise and fall rate, pressure gradient change, rainfall, temperature, wind speed, soil humidity, groundwater pH value, concentration change rate; the dynamic window features include: trend within the window, mutation detection index, and difference between windows; the noise reduction and compression-related features include: signal-to-noise ratio after noise reduction, data compression ratio, and residual signal statistic.

7. The karst geological environment monitoring method according to claim 1, characterized in that, Upload the sensing feature set to the cloud, and perform dynamic risk assessment based on the environmental feature set in the cloud to obtain a predicted ground collapse probability, including: Configure an initial risk assessment network in the cloud; Obtain a preset environmental feature sample set and a ground collapse probability annotation set; Input the environmental feature sample set and the ground collapse probability annotation set into the risk assessment network for training to obtain a trained classification network; Connect a trained LSTM neural network after the classification network to obtain a dynamic risk assessment model; Input the environmental feature set into the dynamic risk assessment model to obtain the ground subsidence probability.

8. The karst geological environment monitoring method according to claim 1, characterized in that Generate an adaptive early warning mechanism for real-time early warning according to the ground subsidence probability, including: Based on the ground subsidence probability value, divide multiple levels of early warning threshold intervals, and determine the adaptive early warning mechanism based on the early warning threshold intervals; the early warning threshold intervals include: low-risk interval, ground subsidence probability value ≤ 0.3, no early warning is triggered; medium-risk interval, 0.3 < ground subsidence probability value ≤ 0.6, yellow early warning is triggered; high-risk interval, 0.6 < ground subsidence probability value ≤ 0.9, orange early warning is triggered; extremely high-risk interval, ground subsidence probability value > 0.9: red early warning is triggered.

9. A karst geological environment monitoring system, characterized in that, Including: A data acquisition unit for deploying a sensor array at each preset monitoring node in the target karst area and collecting multi-source sensing data through the sensor array; The sensor array includes a geological sensor, an environmental sensor, and a chemical sensor; A feature extraction unit for deploying edge computing devices at monitoring nodes and performing real-time noise reduction and feature extraction on the multi-source sensing data through the edge computing devices to obtain a sensing feature set; A risk assessment unit for uploading the sensing feature set to the cloud and performing dynamic risk assessment according to the environmental feature set in the cloud to obtain the predicted ground subsidence probability; A mechanism generation unit for generating an adaptive early warning mechanism for real-time early warning according to the ground subsidence probability; A report generation unit for generating a comprehensive monitoring report according to the ground subsidence probability.

Images