Dangerous rock mass collapse landslide monitoring and early warning method and system based on deep learning

Abstract

The invention provides a deep learning-based dangerous rock mass collapse landslide monitoring and early warning method and system, and relates to the technical field of geological disaster monitoring and early warning, and the method comprises the steps: obtaining rock mass monitoring data; according to the rock mass monitoring data, carrying out dangerous rock mass collapse landslide risk preliminary identification; when it is preliminarily recognized that the dangerous rock body does not have the collapse landslide risk, rock body monitoring data are obtained again; when the collapse landslide risk of the dangerous rock body is preliminarily identified, acquiring environment monitoring data; according to the rock mass monitoring data and the environment monitoring data, through a deep learning algorithm, carrying out dangerous rock mass collapse landslide risk fine identification; when it is finely recognized that the dangerous rock body does not have the collapse landslide risk, rock body monitoring data are obtained again; and when the collapse landslide risk of the dangerous rock body is finely identified, an early warning signal is sent out. According to the invention, environmental factors are considered, and timely and accurate early warning information can be provided in the face of complex and changeable natural environments.

Description

Technical Field

[0001] The present invention relates to the field of geological disaster monitoring and early warning technology, and in particular to a deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method and system. Background Art

[0002] With the acceleration of urbanization, the risk of geological disasters such as collapse and landslides is increasing in many areas due to factors such as human activities and climate change. Rock mass stability is particularly vulnerable during rainy seasons or in areas with significant groundwater level fluctuations. Rock mass collapse and landslides can not only cause significant loss of life and property, but can also damage the ecological environment and severely impact the safety and well-being of surrounding residents. Therefore, timely detection of potential collapse and landslide risks and early warning are crucial for disaster prevention and mitigation.

[0003] In the existing technology, the traditional method of monitoring the risk of dangerous rock collapse and landslide mainly relies on manual detection, which requires a lot of manpower input and cannot provide real-time monitoring and accurate early warning.

[0004] In addition, traditional methods for monitoring the risk of collapse and landslides of dangerous rock masses mainly focus on the physical characteristics of dangerous rock masses and ignore environmental factors. When faced with a complex and changing natural environment, it is difficult to provide timely and accurate early warning information. Summary of the Invention

[0005] In order to solve the technical problems that the existing dangerous rock mass collapse and landslide risk monitoring methods mainly rely on manual detection, require a large amount of manpower input, cannot be monitored in real time and accurately warned, and focus on the physical characteristics of the dangerous rock mass, ignore environmental factors, and are difficult to provide timely and accurate warning information when facing a complex and changeable natural environment, the present invention provides a dangerous rock mass collapse and landslide monitoring and early warning method and system based on deep learning.

[0006] The technical solutions provided by the embodiments of the present invention are as follows: First aspect: The embodiment of the present invention provides a deep learning-based monitoring and early warning method for dangerous rock mass collapse and landslide, including: S1: Obtain rock mass monitoring data; S2: Preliminary identification of the risk of dangerous rock mass collapse and landslide based on the rock mass monitoring data; S3: When it is preliminarily determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1; S4: When it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide, proceed to the next step; S5: Obtain environmental monitoring data; S6: Based on the rock mass monitoring data and the environmental monitoring data, a deep learning algorithm is used to perform a detailed identification of the risk of dangerous rock mass collapse and landslide; S7: When it is determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1; S8: When the dangerous rock mass is precisely identified as posing a risk of collapse and landslide, an early warning signal is issued.

[0007] Second aspect: An embodiment of the present invention provides a deep learning-based dangerous rock mass collapse and landslide monitoring and early warning system, comprising: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the dangerous rock mass collapse and landslide monitoring and early warning method based on deep learning as described in the first aspect is implemented.

[0008] The third aspect: An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for monitoring and early warning of dangerous rock mass collapse and landslide based on deep learning as described in the first aspect is implemented.

[0009] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, by obtaining rock mass monitoring data and performing preliminary identification of the risk of collapse and landslide of dangerous rock masses based on the rock mass monitoring data, it is no longer dependent on manual detection, reduces manpower input, and can achieve real-time monitoring and accurate early warning. Through environmental monitoring data and based on the rock mass monitoring data and environmental monitoring data, a deep learning algorithm is used to perform fine identification of the risk of collapse and landslide of dangerous rock masses. It no longer focuses solely on the physical characteristics of the dangerous rock mass, but takes environmental factors into consideration. When faced with a complex and changeable natural environment, it can provide timely and accurate early warning information. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flowchart of a deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method provided by an embodiment of the present invention; Figure 2 A structural schematic diagram of a dangerous rock mass collapse and landslide monitoring and early warning system based on deep learning provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0012] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0013] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0014] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0015] In the embodiment of the present invention, sometimes a subscript such as W1 may be written as a non-subscript form such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0016] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0017] Reference Manual Figure 1 , which shows a flow chart of a deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method provided by an embodiment of the present invention.

[0018] The present invention provides a method for monitoring and warning dangerous rock mass collapse and landslides based on deep learning. This method can be implemented by a device for monitoring and warning dangerous rock mass collapse and landslides based on deep learning. The device can be a terminal or a server. The processing flow of the method for monitoring and warning dangerous rock mass collapse and landslides based on deep learning can include the following steps: S1: Obtain rock mass monitoring data.

[0019] In a possible implementation, the rock mass monitoring data includes: a triaxial acceleration signal, a vibration velocity signal, and a plurality of displacement data of the dangerous rock mass.

[0020] Optionally, a three-axis acceleration signal of the dangerous rock mass is obtained through a MEMS sensor.

[0021] It should be noted that MEMS sensor (micro-electromechanical system sensor) is a miniature sensor manufactured based on microelectronics and micromachining technology. By integrating mechanical structure and electronic circuit on silicon-based materials, it can achieve high-sensitivity detection of physical quantities (such as acceleration, pressure, vibration, etc.).

[0022] Optionally, a laser Doppler vibrometer is used to obtain a vibration velocity signal of the dangerous rock mass.

[0023] It's important to note that a laser Doppler vibrometer (LDV) is a non-contact vibration measurement instrument based on the laser Doppler effect and optical heterodyne interferometry. It works by emitting a laser beam at the surface of the object being measured. The reflected light undergoes a Doppler frequency shift due to the object's vibration, and the interference signal is then analyzed to determine the velocity, displacement, and acceleration of the vibration.

[0024] Optionally, a laser rangefinder is used to obtain multiple displacement data of the dangerous rock mass.

[0025] It's important to note that a laser rangefinder uses the propagation properties of a laser beam (e.g., time-of-flight or phase-difference methods) to measure the distance to a target. During operation, the laser emits a pulsed or continuous-wave beam. The distance is calculated by calculating the round-trip time of the beam or the phase difference between the received light and a reference beam, combined with the speed of light.

[0026] The advantage of using MEMS sensors, laser Doppler vibrometers, and laser rangefinders to acquire rock mass monitoring data in this invention lies in their ability to accurately and in real time collect triaxial acceleration signals, vibration velocity signals, and displacement data from dangerous rock masses. MEMS sensors provide highly sensitive acceleration and vibration monitoring, laser Doppler vibrometers accurately measure vibration velocity without contact, and laser rangefinders precisely measure displacement changes. The combination of these sensors provides comprehensive data support for real-time monitoring of rock collapse and landslide risks, improving monitoring accuracy and reliability, and ultimately providing a solid data foundation for subsequent risk identification and early warning.

[0027] S2: Based on rock mass monitoring data, conduct preliminary identification of the risk of dangerous rock mass collapse and landslide.

[0028] It should be noted that the initial identification step (S2) of dangerous rock mass collapse and landslide risks includes two optional implementation plans: one is dynamic monitoring based on triaxial acceleration signal analysis (steps S201A to S206A), and the other is spectrum monitoring based on vibration velocity signal analysis (steps S201B to S211B).

[0029] Specifically, the former is suitable for real-time monitoring of high-frequency vibration characteristics, while the latter focuses on stability analysis of low-frequency spectral characteristics. Both are technically independent and serve the same purpose of initial risk screening. In practical applications, one or the other can be flexibly selected to perform preliminary risk identification based on the deployment conditions of the monitoring equipment, the intensity of environmental interference, and the required data acquisition accuracy.

[0030] The present invention provides two optional preliminary risk identification schemes (dynamic monitoring based on triaxial acceleration signals and spectral monitoring based on vibration velocity signals). The advantage is that the most appropriate technical approach can be flexibly selected based on the actual application scenario. For monitoring scenarios with distinct high-frequency vibration characteristics, dynamic monitoring can provide real-time response, while for environments with more stable low-frequency fluctuations, spectral monitoring helps capture long-term stability changes. This flexibility enables the system to operate effectively under different monitoring conditions, improves identification accuracy and applicability, and optimizes the use of monitoring resources.

[0031] In a possible implementation, S2 specifically includes: S201A: Determine the triaxial vibration acceleration components and triaxial gravity acceleration components of the dangerous rock mass based on the triaxial acceleration signals.

[0032] Optionally, the three-axis acceleration signal is specifically:

[0033] in, A x 、 A y as well as A z The three-axis acceleration signals of the dangerous rock mass are respectively x axis, y Axis and z The acceleration signal on the axis, a x 、 a y as well as a z It represents the three-axis vibration acceleration components of the dangerous rock mass, that is, the dangerous rock mass is x axis, y Axis and z The vibration acceleration component on the axis, g x 、 g y as well as g z The three-axis gravity acceleration components of the dangerous rock mass are x axis, y Axis and zThe component of gravitational acceleration about the axis.

[0034] S202A: Calculate the strong vibration acceleration of dangerous rock mass based on the triaxial vibration acceleration components.

[0035] Alternatively, the strong vibration acceleration of the dangerous rock mass can be calculated according to the following formula:

[0036] in, A s Indicates the strong vibration acceleration of dangerous rock mass.

[0037] S203A: Calculate the inclination angle of the dangerous rock mass based on the three-axis gravity acceleration components.

[0038] Alternatively, the inclination angle of the dangerous rock mass can be calculated according to the following formula:

[0039] in, represents the initial gravitational acceleration vector, g x0 Indicates that the dangerous rock mass is x The initial gravitational acceleration component on the axis, g y0 Indicates that the dangerous rock mass is y The initial gravitational acceleration component on the axis, g z0 Indicates that the dangerous rock mass is z The initial gravitational acceleration component on the axis, Represents the current gravity acceleration vector, Indicates that the dangerous rock mass is x The current gravitational acceleration component on the axis, Indicates that the dangerous rock mass is y The current gravitational acceleration component on the axis, Indicates that the dangerous rock mass is z The current gravitational acceleration component on the axis, i represents the inclination angle of the dangerous rock mass, and arccos( ) represents the inverse cosine function.

[0040] S204A: Calculate the inclination rate of the dangerous rock mass based on the inclination angle.

[0041] Alternatively, the inclination rate of the dangerous rock mass can be calculated according to the following formula:

[0042] in, R Indicates the inclination rate of the dangerous rock mass, Δ i Indicates the change in the inclination angle of the dangerous rock mass, Δ t represents the change in time, i t+Δt Indicates t +Δ t The inclination angle of the dangerous rock mass at the moment, i t Indicates t The inclination angle of the dangerous rock mass at the moment.

[0043] S205A: When the strong vibration acceleration is less than the strong vibration acceleration threshold and the inclination rate is less than the inclination rate threshold, it is preliminarily identified that the dangerous rock mass does not have the risk of collapse and landslide.

[0044] S206A: When the strong vibration acceleration is greater than or equal to the strong vibration acceleration threshold or the inclination rate is greater than or equal to the inclination rate threshold, it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide.

[0045] It should be noted that those skilled in the art can set the magnitude of the strong vibration acceleration threshold and the tilt rate threshold according to actual needs, and the present invention does not limit this.

[0046] This method, by simultaneously analyzing triaxial acceleration signals, vibration acceleration components, and gravity acceleration components, comprehensively monitors the dynamic changes of dangerous rock masses in different directions, enhancing understanding of rock mass behavior. Calculating strong vibration acceleration helps accurately identify the risk of collapse and landslides caused by severe vibration, while monitoring the inclination angle and tilt rate provides real-time feedback on changes in rock mass stability, thereby promptly identifying potential landslide risks.

[0047] In a possible implementation, S2 specifically includes: S201B: Calculate the vibration amplitude of the dangerous rock mass based on the vibration velocity signal.

[0048] Optionally, the vibration amplitude of the dangerous rock mass is calculated according to the following formula:

[0049] in, A Indicates the vibration amplitude of the dangerous rock mass, max indicates the maximum value, v t Indicates t Vibration velocity signal at the moment.

[0050] S202B: Dynamically determine a vibration amplitude threshold value according to the vibration amplitude.

[0051] Optionally, the vibration amplitude threshold is specifically:

[0052] in, k t Indicates tThe vibration amplitude threshold at time , A t-1 Indicates t Vibration amplitude of the dangerous rock mass at time -1.

[0053] In the present invention, the dynamic threshold setting enhances the system's adaptability to rock mass conditions, improves the accuracy of risk identification, and ensures that potential collapse and landslide risks can be identified in a timely manner when the vibration amplitude changes significantly.

[0054] S203B: When the vibration amplitude is less than or equal to the vibration amplitude threshold, it is preliminarily identified that the dangerous rock mass does not have the risk of collapse and landslide.

[0055] S204B: When the vibration amplitude is greater than the vibration amplitude threshold, proceed to the next step.

[0056] S205B: Construct a dynamic model of dangerous rock mass.

[0057] Optionally, the dangerous rock mass dynamic model is specifically:

[0058] in, M Indicates the quality of the dangerous rock mass, L Indicates the distance from the center of gravity of the dangerous rock mass to the center of rotation, represents the angular acceleration of the dangerous rock mass, K represents the bond stiffness of the dangerous rock mass, l represents the bonding length of the dangerous rock mass, r Represents the angular displacement of the dangerous rock mass.

[0059] S206B: Perform fast Fourier transform on the pre-processed vibration velocity signal to generate a spectrum diagram.

[0060] Specifically, in frequency domain analysis, the pre-processed vibration velocity signal is first converted from the time domain to a complex result in the frequency domain using a fast Fourier transform (FFT). Subsequently, the amplitude spectrum of the FFT result needs to be calculated: the complex absolute value of each frequency component is taken and normalized by dividing by the number of sampling points and multiplying by a coefficient of 2 to make it reflect the actual physical magnitude (only the first half of the effective frequency points are retained). To improve detail recognition, the amplitude can be logarithmically processed to expand the dynamic range. Next, a frequency axis is generated based on the sampling frequency and signal length. The frequency interval is determined by dividing the sampling frequency by the total number of sampling points, and the upper limit of the effective frequency range is half of the sampling frequency. Finally, a spectrum is plotted with frequency as the horizontal axis and normalized amplitude as the vertical axis. The peak position corresponds to the main frequency component where the energy in the signal is concentrated (such as the natural frequency of the dangerous rock mass), which intuitively reveals the frequency characteristics of the vibration signal and provides a key basis for subsequent risk identification.

[0061] It's important to note that the Fast Fourier Transform (FFT) is a highly efficient algorithm that rapidly converts time-domain signals (such as time-varying physical quantities like sound and vibration) into frequency-domain analysis results, accurately identifying hidden periodic features or frequency components within the signal. By cleverly breaking down complex computational problems into multiple smaller subproblems, it reduces the computational time of traditional methods from growing with the square of the data volume to a near-linear growth, significantly improving efficiency.

[0062] S207B: In the spectrum diagram, calculate the natural frequency of the dangerous rock mass according to the dynamic model of the dangerous rock mass.

[0063] Alternatively, the natural frequency of the dangerous rock mass can be calculated according to the following formula:

[0064] in, f Represents the natural frequency of the dangerous rock mass.

[0065] S208B: Calculate the vibration amplitude change rate of the dangerous rock mass based on the vibration amplitude.

[0066] Optionally, the vibration amplitude change rate of the dangerous rock mass is calculated according to the following formula:

[0067] in, AVR Indicates the rate of change of vibration amplitude of dangerous rock mass, A t Indicates t The vibration amplitude of the dangerous rock mass at the time.

[0068] S209B: Calculate the rate of change of the natural frequency of the dangerous rock mass based on the natural frequency.

[0069] Optionally, the natural frequency change rate of the dangerous rock mass is calculated according to the following formula:

[0070] in, FVR It represents the natural frequency change rate of the dangerous rock mass, f t-1 Indicates t The natural frequency of the dangerous rock mass at the moment -1, f t Indicates t The natural frequency of the dangerous rock mass at the moment.

[0071] S210B: When the vibration amplitude change rate is less than the vibration amplitude change rate threshold and the natural frequency change rate is less than the natural frequency change rate threshold, it is preliminarily identified that the dangerous rock mass does not have the risk of collapse and landslide.

[0072] S211B: When the vibration amplitude change rate is greater than or equal to the vibration amplitude change rate threshold or the natural frequency change rate is greater than or equal to the natural frequency change rate threshold, it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide.

[0073] It should be noted that those skilled in the art can set the vibration amplitude change rate threshold and the natural frequency change rate threshold according to actual needs, and the present invention does not limit this.

[0074] This invention, through spectrum analysis and dynamic monitoring based on vibration velocity signals, enables flexible and accurate preliminary identification of collapse and landslide risks in dangerous rock masses. By dynamically adjusting the vibration amplitude threshold and combining frequency domain analysis of the signal with Fast Fourier Transform (FFT), the frequency characteristics and natural frequency variations of the rock mass can be revealed, thereby helping to identify potential risks. Furthermore, by constructing a dynamic model of the dangerous rock mass and combining it with monitoring of the vibration amplitude and natural frequency change rate, the precise assessment of rock mass stability is enhanced.

[0075] S3: When it is preliminarily determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1.

[0076] S4: When it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide, proceed to the next step.

[0077] S5: Obtain environmental monitoring data.

[0078] In a possible implementation, the environmental monitoring data includes rainfall and groundwater level.

[0079] Optionally, rainfall is obtained through a rain gauge.

[0080] Optionally, the groundwater level is obtained through a groundwater level monitoring well.

[0081] In this invention, environmental monitoring data on rainfall and groundwater levels can provide key external factors related to the risk of collapse and landslides in dangerous rock masses. Rainfall directly affects the water infiltration and stability of the rock mass, while changes in groundwater levels can reflect the water pressure within the rock mass and the potential landslide risk. By monitoring these environmental factors in real time, a more comprehensive assessment of rock mass stability can be achieved, improving the accuracy of predictions of collapse and landslide risks.

[0082] S6: Based on rock mass monitoring data and environmental monitoring data, a deep learning algorithm is used to precisely identify the risk of dangerous rock mass collapse and landslides.

[0083] In this paper, a deep learning algorithm, combining rock mass monitoring data with environmental monitoring data, can precisely identify the risk of dangerous rock mass collapse and landslides. This algorithm can fully utilize multi-source information to improve prediction accuracy. Deep learning algorithms can automatically extract features from complex, multi-dimensional data and identify potential risk patterns, thereby achieving more accurate risk assessment.

[0084] In one possible implementation, the deep learning algorithm is specifically a long short-term memory neural network, and S6 specifically includes: S601: Eliminate outliers in the displacement data based on the Pauta criterion.

[0085] It should be noted that the Pauta criterion (also known as the Laida criterion or the 3σ criterion) is a statistical outlier identification method based on the normal distribution assumption. Its core idea is to calculate the mean and standard deviation of the data, define a reasonable range, and identify extreme values ​​outside this range as outliers and eliminate them. Specifically, for a data set that follows a normal or nearly normal distribution, approximately 99.7% of the values ​​will be distributed within three times the standard deviation (3σ) of the mean (μ), that is, (μ-3σ,μ+3σ). Data outside this range are considered gross errors caused by extremely low probability events and need to be excluded.

[0086] In this paper, we eliminate outliers in displacement data based on the Pauta criterion, effectively improving data quality and accuracy. By eliminating extreme values ​​outside a reasonable range, we can prevent abnormal data from interfering with the training process of deep learning algorithms, thereby ensuring that the model can more accurately identify potential risk patterns.

[0087] In a possible implementation, S601 specifically includes: S6011: Calculate the mean of the displacement data.

[0088] Optionally, calculate the mean of the displacement data according to the following formula:

[0089] in, m represents the mean of the displacement data, N Indicates the total number of displacement data, x i Indicates the i displacement data.

[0090] S6012: Calculate the standard deviation of the displacement data.

[0091] Optionally, calculate the standard deviation of the displacement data according to the following formula:

[0092] in, Indicates the standard deviation of the displacement data.

[0093] S6013: Calculate the residual absolute value of the displacement data according to the mean value of the displacement data.

[0094] Optionally, the absolute value of the residual error of the displacement data is calculated according to the following formula:

[0095] Among them, Δ x Represents the absolute value of the residual of the displacement data.

[0096] S6014: Dynamically determine the effective value range based on the mean and standard deviation of the displacement data and the Pauta criterion.

[0097] Optionally, the effective value range is dynamically determined according to the following formula:

[0098] in, I Indicates the valid value range.

[0099] S6015: The displacement data corresponding to the residual absolute value exceeding the valid value interval is treated as an outlier and eliminated.

[0100] Specifically, when the absolute value of the residual is not within the valid value interval, the displacement data corresponding to the absolute value of the residual is eliminated as an abnormal value.

[0101] In this paper, outlier removal from displacement data based on the Pauta criterion significantly improves data quality and accuracy. Calculating the mean and standard deviation of the displacement data and dynamically determining the effective value range helps identify and eliminate outliers caused by measurement errors or environmental interference, thereby preventing them from interfering with subsequent deep learning model training. By removing outliers, the model's accuracy and stability are improved, reducing the risk of misjudgment and ensuring a more precise prediction of the risk of dangerous rock mass collapse and landslides.

[0102] S602: Performing discrete wavelet transform on the displacement data after eliminating outliers to obtain target displacement data.

[0103] It should be noted that Discrete Wavelet Transform (DWT) is a multi-scale analysis method that decomposes signals into different frequency components. Its core idea is to gradually downsample the signal through low-pass filters and high-pass filters to separate low-frequency approximate components (reflecting the overall trend of the signal) and high-frequency detail components (capturing local mutations or noise in the signal).

[0104] In the present invention, by performing discrete wavelet transform (DWT) on the displacement data after eliminating outliers, the signal can be decomposed into low-frequency and high-frequency components, thereby capturing the overall trend and local details of the signal respectively.

[0105] In a possible implementation, S602 specifically includes: S6021: Extract the low-frequency components of the displacement data after eliminating outliers through a low-pass filter.

[0106] Optionally, the low-frequency component of the displacement data after eliminating outliers is extracted according to the following formula:

[0107] in, S low [ k ] represents the extracted k low-frequency components, Indicates the n The displacement data of the sampling points after eliminating outliers, h 1 represents the convolution kernel of the low-pass filter.

[0108] S6022: Using a high-pass filter, extract the high-frequency components of the displacement data after eliminating outliers.

[0109] Optionally, the high-frequency component of the displacement data after eliminating outliers is extracted according to the following formula:

[0110] in, S high [ k ] represents the extracted k high-frequency components, h 2 represents the convolution kernel of the high-pass filter.

[0111] S6023: Reconstruct the displacement data after eliminating outliers based on the low-frequency components and the high-frequency components to obtain target displacement data.

[0112] Optionally, the displacement data after eliminating outliers is reconstructed according to the following formula to obtain target displacement data:

[0113] in, Indicates the n The target displacement data of each sampling point.

[0114] In this method, by using low-pass and high-pass filters to extract the low- and high-frequency components of displacement data and reconstructing the data after eliminating outliers, the stability and accuracy of the data can be effectively improved. The low-frequency component helps capture overall trends, reveal long-term changes in the rock mass, and identify changes in stability. The high-frequency component, on the other hand, can identify local mutations and noise, promptly identifying potential risks. By removing noise interference and accurately reconstructing target displacement data, this method provides more reliable input for subsequent deep learning algorithms, thereby enhancing the accuracy and reliability of risk identification.

[0115] S603: Perform weighted summation on the target displacement data to calculate cumulative displacement data.

[0116] Optionally, the accumulated displacement data is calculated according to the following formula.

[0117]

[0118] in, oh n Indicates the n The weight coefficient of the target displacement data of each sampling point.

[0119] In this method, by weighting and summing the target displacement data and calculating the cumulative displacement data, the overall displacement trend of the dangerous rock mass can be more accurately reflected based on the weight coefficients of different sampling points. The weighted summation can adjust the weights based on the importance of different time points or the reliability of the data, thereby highlighting the data that is more critical to risk identification.

[0120] S604: Decompose the accumulated displacement data using the Prophet algorithm to obtain trend terms and periodic terms:

[0121] in, y ( t ) means in t The cumulative displacement data at the moment, g ( t ) means in t Trend items at the moment, s ( t ) means in t The periodic term at time, e ( t ) means in t The error term at time, C ( t ) means in t The trend saturation value at the moment, exp( ) represents the exponential function, r represents the initial growth rate, β ( t ) means int The mutation point indicator vector at the moment, T represents the transpose operation, d represents the change in growth rate, m Indicates the initial offset, c j Indicates the j The offset corresponding to the mutation point, M represents the order of the Fourier series, cos() represents the cosine function, P represents the period length, sin() represents the sine function, Indicates the The first Fourier coefficient of order , Indicates the The second Fourier coefficient of order .

[0122] It's important to note that Prophet is an algorithm used for time series forecasting. It uses additive modeling to decompose time series data into long-term trends (such as linear or nonlinear growth trends), cyclical fluctuations (such as seasonality), and possible holiday effects. Through this decomposition, the Prophet algorithm can capture both long-term trends and cyclical fluctuations in the data and leverage this information for future forecasts. In the S604 application, the Prophet algorithm helps identify long-term trends and cyclical fluctuations in displacement data, providing data support for subsequent risk identification.

[0123] In this paper, by using the Prophet algorithm to decompose accumulated displacement data, we can accurately identify long-term trends and cyclical fluctuations in the data. This decomposition method helps break down complex displacement data into trend and cyclical terms, thereby more clearly revealing the stability changes and cyclical fluctuations of dangerous rock masses. The trend term reflects the long-term trend of the rock mass, while the cyclical term reveals the impact of cyclical environmental factors (such as seasonal changes or climate factors) on the rock mass. This method improves the accuracy of risk prediction and provides more reliable basic data for subsequent risk identification and early warning.

[0124] S605: Based on environmental monitoring data and periodic items, a long-short-term memory neural network is used to perform detailed identification of the risk of dangerous rock mass collapse and landslide.

[0125] It's important to note that the Long Short-Term Memory (LSTM) neural network is a special type of recurrent neural network (RNN) designed to process and predict time series data. Unlike traditional RNNs, LSTMs effectively address long-term dependencies, meaning they can remember critical information over long time intervals. They control the flow of information by introducing three primary "gate" mechanisms: a forget gate, an input gate, and an output gate. The forget gate determines which information to discard, the input gate determines which information to store, and the output gate determines the next output. LSTMs are particularly well-suited for processing long-term sequence data, such as meteorological and stock market data. In the prediction of dangerous rock mass collapse and landslide risks, LSTMs can accurately identify changing risk trends based on historical data and environmental monitoring data.

[0126] In this paper, LSTM can effectively process time series data with long time spans, accurately capturing key patterns of change in historical data. The LSTM's "gate" mechanism automatically and selectively stores and forgets information, allowing the model to remember long-term dependencies related to collapse and landslide risk. In this way, LSTM can accurately predict rock mass risk trends based on historical data and real-time environmental changes, providing more accurate risk assessment and early warning capabilities, thereby enhancing the timeliness and accuracy of disaster prevention.

[0127] In a possible implementation, S605 specifically includes: S6051: Use environmental monitoring data and periodic terms as input to the long short-term memory neural network and output the predicted value of the periodic term:

[0128] in, f t Indicates t The output of the forget gate at the moment, s represents the Sigmoid activation function, W f represents the weight matrix of the forget gate, h t-1 Indicates t -1 moment of hidden state, X t Indicates t Input data at time, b f represents the bias term of the forget gate, i t Indicates t The output of the input gate at time t, W i represents the weight matrix of the input gate, bi represents the bias term of the input gate, Indicates t The candidate cell state at the moment, tanh () represents the tanh activation function, W c The weight matrix representing the candidate cell state, b c represents the bias term of the candidate cell state, c t Indicates t The state of cells at any moment, represents element-wise multiplication, o t Indicates t The output of the output gate at time t, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate, h t Indicates t The hidden state at the moment.

[0129] S6052: Determine the final displacement prediction value by performing an addition operation based on the periodic item prediction value and the trend item:

[0130] in, D represents the final displacement prediction value, R Indicates rainfall, W represents the groundwater level, LSTM( ) represents the output of the long short-term memory neural network, i.e., the predicted value of the periodic term.

[0131] S6053: Determine the displacement rate and displacement acceleration based on the final displacement prediction value.

[0132]

[0133] in, v t Indicates t The displacement rate at time Δ D Represents the displacement change, Δ t Indicates the time change, that is, the difference between the current moment and the previous moment. D t Indicates t The final displacement prediction value at time , D t-1 Indicates t The final displacement prediction value at time -1, a t Indicatest The displacement acceleration at the time, Δ v Indicates the speed change, v t-1 Indicates t Displacement rate at time -1.

[0134] S6054: Detailed identification of the risk of dangerous rock mass collapse and landslide based on displacement acceleration and displacement rate.

[0135] In a possible implementation, S6054 specifically includes: When the displacement acceleration is less than the displacement acceleration threshold and the displacement rate is less than the displacement rate threshold, it is precisely identified that the dangerous rock mass does not have the risk of collapse and landslide.

[0136] When the displacement acceleration is greater than or equal to the displacement acceleration threshold or the displacement rate is greater than or equal to the displacement rate threshold, the dangerous rock mass is accurately identified as having the risk of collapse and landslide.

[0137] It should be noted that those skilled in the art can set the displacement acceleration threshold and the displacement rate threshold according to actual needs, and the present invention does not limit this.

[0138] This method uses a long short-term memory (LSTM) neural network, combined with environmental monitoring data and periodic terms, to accurately identify changing trends in the risk of collapse and landslides in dangerous rock masses. LSTM leverages its advantage in processing time series data to capture long-term dependencies in the data, thereby accurately predicting cyclical changes. By combining the periodic term with the trend term, the final displacement prediction takes into account the influence of environmental factors such as rainfall and groundwater levels, improving prediction accuracy. Accurately calculating displacement rate and acceleration facilitates dynamic assessment of changes in rock mass stability and provides a more precise basis for risk warning.

[0139] S7: When it is determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1.

[0140] S8: When the risk of collapse and landslide of dangerous rock mass is carefully identified, an early warning signal is issued.

[0141] Reference Manual Figure 2 , which shows a structural schematic diagram of a dangerous rock collapse and landslide monitoring and early warning system based on deep learning provided by the present invention.

[0142] The present invention further provides a deep learning-based dangerous rock mass collapse and landslide monitoring and early warning system 20, which is applied to the above-mentioned deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method, including: Processor 201; The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, the deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method as described in the method embodiment is implemented.

[0143] The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning system 20 provided by the present invention can execute the above-mentioned deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method, and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0144] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, by obtaining rock mass monitoring data and performing preliminary identification of the risk of collapse and landslide of dangerous rock masses based on the rock mass monitoring data, it is no longer dependent on manual detection, reduces manpower input, and can achieve real-time monitoring and accurate early warning. Through environmental monitoring data and based on the rock mass monitoring data and environmental monitoring data, a deep learning algorithm is used to perform fine identification of the risk of collapse and landslide of dangerous rock masses. It no longer focuses solely on the physical characteristics of the dangerous rock mass, but takes environmental factors into consideration. When faced with a complex and changeable natural environment, it can provide timely and accurate early warning information.

[0145] It should be understood that the processor in the embodiments of the present invention may be a central processing unit (CPU), but may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0146] It should also be understood that the memory in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0147] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0148] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0149] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0150] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0151] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0152] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0153] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0154] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0155] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0156] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0157] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the method for monitoring and early warning of dangerous rock mass collapse and landslides based on deep learning as described in the method embodiment is implemented.

[0158] The computer-readable storage medium provided by the present invention can implement the steps and effects of the dangerous rock collapse and landslide monitoring and early warning method based on deep learning in the above-mentioned method embodiment. To avoid repetition, the present invention will not go into details.

[0159] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least: In an embodiment of the present invention, by obtaining rock mass monitoring data and performing preliminary identification of the risk of collapse and landslide of dangerous rock masses based on the rock mass monitoring data, it is no longer dependent on manual detection, reduces manpower input, and can achieve real-time monitoring and accurate early warning. Through environmental monitoring data and based on the rock mass monitoring data and environmental monitoring data, a deep learning algorithm is used to perform fine identification of the risk of collapse and landslide of dangerous rock masses. It no longer focuses solely on the physical characteristics of the dangerous rock mass, but takes environmental factors into consideration. When faced with a complex and changeable natural environment, it can provide timely and accurate early warning information.

[0160] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

[0161] There are a few points to note: (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention. Other structures may refer to conventional designs.

[0162] (2) For the sake of clarity, the thickness of layers or regions in the drawings used to describe the embodiments of the present invention are exaggerated or reduced, that is, these drawings are not drawn to scale. It is understood that when an element such as a layer, film, region, or substrate is referred to as being "on" or "under" another element, the element may be "directly on" or "under" the other element or intervening elements may be present.

Claims

1. A deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method, characterized by: include: S1: Obtain rock mass monitoring data; S2: Preliminary identification of the risk of dangerous rock mass collapse and landslide based on the rock mass monitoring data; S3: When it is preliminarily determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1; S4: When it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide, proceed to the next step; S5: Obtain environmental monitoring data; S6: Based on the rock mass monitoring data and the environmental monitoring data, a deep learning algorithm is used to perform a detailed identification of the risk of dangerous rock mass collapse and landslide; S7: When it is determined that the dangerous rock mass does not have the risk of collapse or landslide, return to S1; S8: When the dangerous rock mass is precisely identified as posing a risk of collapse and landslide, an early warning signal is issued.

2. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 1 is characterized in that: The rock mass monitoring data includes: a triaxial acceleration signal, a vibration velocity signal and a plurality of displacement data of the dangerous rock mass; The environmental monitoring data include: rainfall and groundwater level.

3. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 2 is characterized in that: The S2 specifically includes: S201A: Determine a triaxial vibration acceleration component and a triaxial gravity acceleration component of the dangerous rock mass according to the triaxial acceleration signal; S202A: Calculating the strong vibration acceleration of the dangerous rock mass according to the triaxial vibration acceleration components; S203A: Calculating the inclination angle of the dangerous rock mass according to the three-axis gravity acceleration components; S204A: Calculating the inclination rate of the dangerous rock mass according to the inclination angle; S205A: When the strong vibration acceleration is less than the strong vibration acceleration threshold and the inclination rate is less than the inclination rate threshold, it is preliminarily determined that the dangerous rock mass does not have the risk of collapse and landslide; S206A: When the strong vibration acceleration is greater than or equal to the strong vibration acceleration threshold or the inclination rate is greater than or equal to the inclination rate threshold, it is preliminarily identified that the dangerous rock mass has a risk of collapse and landslide.

4. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 2 is characterized in that: The S2 specifically includes: S201B: Calculating the vibration amplitude of the dangerous rock mass according to the vibration velocity signal; S202B: Dynamically determining a vibration amplitude threshold according to the vibration amplitude; S203B: When the vibration amplitude is less than or equal to the vibration amplitude threshold, it is preliminarily determined that the dangerous rock mass does not have the risk of collapse or landslide; S204B: When the vibration amplitude is greater than the vibration amplitude threshold, proceed to the next step; S205B: Constructing a dynamic model of dangerous rock masses; S206B: Performing fast Fourier transform on the preprocessed vibration velocity signal to generate a spectrum diagram; S207B: Calculating the natural frequency of the dangerous rock mass in the frequency spectrum according to the dangerous rock mass dynamic model; S208B: Calculating a vibration amplitude change rate of the dangerous rock mass according to the vibration amplitude; S209B: Calculating a natural frequency change rate of the dangerous rock mass according to the natural frequency; S210B: When the vibration amplitude change rate is less than the vibration amplitude change rate threshold and the natural frequency change rate is less than the natural frequency change rate threshold, it is preliminarily determined that the dangerous rock mass does not have a risk of collapse and landslide; S211B: When the vibration amplitude change rate is greater than or equal to the vibration amplitude change rate threshold or the natural frequency change rate is greater than or equal to the natural frequency change rate threshold, it is preliminarily identified that the dangerous rock mass has the risk of collapse and landslide.

5. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 2 is characterized in that: The deep learning algorithm is specifically a long short-term memory neural network, and S6 specifically includes: S601: Eliminate outliers in the displacement data based on the Pauta criterion; S602: Performing discrete wavelet transform on the displacement data after eliminating outliers to obtain target displacement data; S603: Perform weighted summation on the target displacement data to calculate cumulative displacement data; S604: Decompose the accumulated displacement data using the Prophet algorithm to obtain trend terms and period terms: in, y ( t ) means in t The cumulative displacement data at the moment, g ( t ) means in t Trend items at the moment, s ( t ) means in t The periodic term at time, ε ( t ) means in t The error term at time, C ( t ) means in t The trend saturation value at the moment, exp( ) represents the exponential function, r represents the initial growth rate, β ( t ) means in t The mutation point indicator vector at the moment, T represents the transpose operation, δ represents the change in growth rate, m Indicates the initial offset, γ j Indicates the j The offset corresponding to the mutation point, M represents the order of the Fourier series, cos() represents the cosine function, P represents the period length, sin() represents the sine function, Indicates the The first Fourier coefficient of order , Indicates the The second Fourier coefficient of the order; S605: Based on the environmental monitoring data and the periodic item, the long short-term memory neural network is used to perform detailed identification of the risk of dangerous rock mass collapse and landslide.

6. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 5 is characterized in that: The S601 specifically includes: S6011: Calculate the mean of the displacement data; S6012: Calculate the standard deviation of the displacement data; S6013: Calculate the residual absolute value of the displacement data according to the mean value of the displacement data; S6014: Dynamically determine a valid value interval based on the mean and standard deviation of the displacement data and the Pauta criterion; S6015: Eliminate the displacement data corresponding to the residual absolute value exceeding the valid value interval as the abnormal value.

7. The method for monitoring and early warning of dangerous rock mass collapse and landslide based on deep learning according to claim 5, characterized in that: The S602 specifically includes: S6021: extracting the low-frequency component of the displacement data after eliminating outliers through a low-pass filter; S6022: extracting high-frequency components of the displacement data after eliminating outliers through a high-pass filter; S6023: Reconstruct the displacement data after eliminating outliers based on the low-frequency component and the high-frequency component to obtain the target displacement data.

8. The method for monitoring and early warning of dangerous rock mass collapse and landslide based on deep learning according to claim 5, characterized in that: The S605 specifically includes: S6051: Using the environmental monitoring data and the periodic term as inputs to the long short-term memory neural network, outputting a predicted value of the periodic term: in, f t Indicates t The output of the forget gate at the moment, σ represents the Sigmoid activation function, W f represents the weight matrix of the forget gate, h t-1 Indicates t -1 moment of hidden state, X t Indicates t Input data at time, b f represents the bias term of the forget gate, i t Indicates t The output of the input gate at time t, W i represents the weight matrix of the input gate, b i represents the bias term of the input gate, Indicates t The candidate cell state at the moment, tanh () represents the tanh activation function, W c The weight matrix representing the candidate cell state, b c represents the bias term of the candidate cell state, c t Indicates t The state of cells at any moment, represents element-wise multiplication, o t Indicates t The output of the output gate at time t, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate, h t Indicates t The hidden state at the moment; S6052: Determine a final displacement prediction value by performing an addition operation based on the periodic item prediction value and the trend item: in, D represents the final displacement prediction value, R Indicates rainfall, W represents the groundwater level, and LSTM() represents the output of the long short-term memory neural network, i.e., the predicted value of the periodic term; S6053: Determine a displacement rate and a displacement acceleration according to the final displacement prediction value; in, v t Indicates t The displacement rate at time Δ D Represents the displacement change, Δ t Indicates the time change, that is, the difference between the current moment and the previous moment. D t Indicates t The final displacement prediction value at time , D t-1 Indicates t The final displacement prediction value at time -1, a t Indicates t The displacement acceleration at the time, Δ v Indicates the speed change, v t-1 Indicates t Displacement rate at time -1; S6054: Perform detailed identification of the risk of dangerous rock mass collapse and landslide based on the displacement acceleration and the displacement rate.

9. The deep learning-based dangerous rock mass collapse and landslide monitoring and early warning method according to claim 8 is characterized in that: The S6054 specifically includes: When the displacement acceleration is less than the displacement acceleration threshold and the displacement rate is less than the displacement rate threshold, it is precisely identified that the dangerous rock mass does not have the risk of collapse and landslide; When the displacement acceleration is greater than or equal to the displacement acceleration threshold or the displacement rate is greater than or equal to the displacement rate threshold, it is precisely identified that the dangerous rock mass has the risk of collapse and landslide.

10. A deep learning-based dangerous rock mass collapse and landslide monitoring and early warning system, characterized by: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method for monitoring and early warning of dangerous rock mass collapse and landslide based on deep learning as described in any one of claims 1 to 9 is implemented.

Images