CN114818518A - Method for analyzing monitoring information of landslide prevention danger of abrupt slope - Google Patents
Method for analyzing monitoring information of landslide prevention danger of abrupt slope Download PDFInfo
- Publication number
- CN114818518A CN114818518A CN202210754227.XA CN202210754227A CN114818518A CN 114818518 A CN114818518 A CN 114818518A CN 202210754227 A CN202210754227 A CN 202210754227A CN 114818518 A CN114818518 A CN 114818518A
- Authority
- CN
- China
- Prior art keywords
- landslide
- information
- accumulated
- steep slope
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/10—Alarms for ensuring the safety of persons responsive to calamitous events, e.g. tornados or earthquakes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Abstract
The invention discloses a method for analyzing danger monitoring information of a steep slope landslide prevention, which relates to the technical field of surveying, and realizes acquisition of data information of the steep slope landslide prevention through terminal equipment, wherein the terminal equipment at least comprises a monitoring terminal, a rainfall monitoring terminal, a ground surface crack monitoring terminal, a deep displacement monitoring terminal or a soil moisture meter; processing multi-source heterogeneous data information is achieved through a K-means clustering algorithm; receiving the information characteristics of the steep slope landslide prevention data in a wireless communication mode or a serial communication mode, extracting the information characteristics of the steep slope landslide prevention data through a convolutional neural network and a long-short term memory neural network, and analyzing the extracted information characteristics of the steep slope landslide prevention data; constructing an early warning model to realize analysis of data information of the steep slope landslide prevention, wherein the early warning model is a KM-SVM landslide real-time early warning model; the method and the device can realize analysis and mining of more specific information of the steep slope landslide prevention danger information and improve the analysis capability of data information.
Description
Technical Field
The invention relates to the field of surveying, in particular to a method for analyzing monitoring information of landslide prevention risks of a steep slope.
Background
The steep slope refers to a channel bottom slope with a gradient greater than a critical bottom slope, and also refers to a steeply rising slope. The technical scheme of the side slope landslide prevention safety measure plays an extremely important role in various occasions. Landslide (landslides) refers to the action and phenomenon that a part of rock soil on a mountain slope generates shear displacement along a certain weak structural plane (zone) under the action of gravity (including the gravity of the rock soil and the dynamic and static pressure of underground water) to integrally move to the lower part of the slope. Commonly known as 'mountain walking', 'mountain collapse', 'ground slip', 'earth slip', etc. Is one of common geological disasters. The main inducing factors include earthquake, rainfall and snow melting, scouring and soaking of surface water, continuous scouring of slope toe by surface water such as rivers and the like, unreasonable human engineering activities such as digging slope toe, loading and blasting of upper part of slope body, water storage (discharge) of reservoir, mining and the like can induce landslide, and tsunami, storm tide, freeze thawing and the like can also induce landslide. The most common natural inducing factors are heavy rainfall or heavy rainstorm, and particularly, landslide is more easily induced by heavy rainfall after long rain falls suddenly.
In real life, a plurality of data information such as landslide crack deformation rate, horizontal accumulated deformation quantity, vertical average deformation rate, accumulated acceleration, accumulated jerk, cracks, accumulated deformation quantity, daily rainfall, soil water content and the like have important influence factors on landslide. How to achieve the evaluation of these parameters and components is crucial for landslide. In the prior art, an upper body surface phenomenon is usually detected by adopting a manual empirical method and detection equipment, and although the technology has a certain reference value for safety protection, the analysis of data information cannot be realized, and the analysis and the mining of more specific information of the landslide prevention danger information of the steep slope cannot be obtained.
Disclosure of Invention
Aiming at the defects of the technology, the invention discloses a method for analyzing the monitoring information of the landslide prevention danger of the steep slope, which can realize the analysis and the mining of more specific information of the landslide prevention danger information of the steep slope and improve the analysis capability of data information.
In order to achieve the technical effects, the invention adopts the following technical scheme:
a method for analyzing monitoring information of landslide prevention risks of a steep slope comprises the following steps:
step one, acquiring landslide prevention data information of a steep slope;
in the step, the acquisition of the landslide prevention data information of the steep slope is realized through terminal equipment, wherein the terminal equipment at least comprises a monitoring terminal, a rainfall monitoring terminal, a ground surface crack monitoring terminal, a deep displacement monitoring terminal or a soil moisture meter; processing multi-source heterogeneous data information through a K-means clustering algorithm;
extracting the information characteristics of the landslide prevention data of the steep slope;
in the step, the steep slope landslide prevention data information characteristic receiving is realized in a wireless communication mode or a serial communication mode, the steep slope landslide prevention data information characteristic is extracted through a convolutional neural network and a long-short term memory neural network, and the extracted steep slope landslide prevention data information characteristic is analyzed;
constructing an early warning model to realize analysis of data information of the steep slope landslide prevention, wherein the early warning model is a KM-SVM landslide real-time early warning model;
in the step, the slope deformation rate, the acceleration, the accumulated deformation quantity, the cracks, the rainfall and the soil water content are used as comprehensive early warning indexes, and a high-precision multi-source information fusion early warning system is formed according to respective early warning thresholds of different levels, wherein when the slope crack deformation rate is greater than 25mm/d, the accumulated deformation quantity in the horizontal direction is greater than 125mm, the average deformation rate in the vertical direction is greater than 30mm/d, the accumulated acceleration and the accumulated acceleration are greater than 0, the crack meter 5-day accumulated deformation quantity is greater than 125mm, the daily rainfall is greater than 200mm or the soil water content is greater than 75%, the steep slope antiskid is in a high-risk state; when the deformation rate of the landslide crack is less than 6mm/d, the accumulated deformation amount in the horizontal direction is less than 30mm, the average deformation rate in the vertical direction is less than 10mm/d, the accumulated acceleration and the accumulated jerk are less than 0, the 5-day accumulated deformation amount of the crack meter is less than 30mm, the daily rainfall is less than 50mm or the water content of soil is within the range of 30% -45%, the abrupt slope landslide prevention has no landslide risk.
As a further technical scheme of the invention, the terminal equipment is a displacement sensor, wherein the displacement sensor adopts an STM32F105RBT 632-bit singlechip, the main frequency can reach 72MHZ, and an 8MHz active crystal oscillator is adopted as a clock module of the singlechip.
As a further technical scheme of the invention, the time interval for acquiring the steep slope landslide prevention data information is 30 minutes.
As a further technical scheme, in the step one, an encoder and an Adam optimizer are arranged in a K-means clustering algorithm, the encoder is used for realizing encoding of different data information based on a PLC control module, and the Adam optimizer realizes updating of input data information after gradient oscillation and filtering oscillation of historical steep slope antiskid dangerous data information.
As a further technical scheme of the present invention, the convolutional neural network comprises an input matrix, a convolutional kernel, a feature extraction module, a data normalization module and an output module, wherein an output end of the input matrix is connected with an input end of the convolutional kernel, an output end of the convolutional kernel is connected with an input end of the feature extraction module, an output end of the feature extraction module is connected with an input end of the data normalization module, and an output end of the data normalization module is connected with an input end of the output module;
the long-short term memory neural network model comprises a forgetting gate, an input gate and an output gate which are provided with an accelerator, wherein the accelerator is an acceleration module based on PFGA control.
As a further technical scheme, the KM-SVM landslide real-time early warning model comprises a Kalman filtering module and a support vector machine module, wherein the Kalman filtering module is used for predicting dangerous data information, and the support vector machine module is used for diagnosing the dangerous data information.
As a further technical scheme of the invention, the Kalman filtering module realizes data prediction through a dynamic data processing recursion algorithm, and the data types comprise landslide crack deformation rate, horizontal direction accumulated deformation quantity, vertical direction average deformation rate, accumulated acceleration, accumulated jerk, cracks, accumulated deformation quantity, daily rainfall and soil water content; observing the data information through the latest moment to observe the random difference data information of the landslide information state, wherein a prediction model of a Kalman filtering module is as follows:
in the formula (1), the first and second groups,
a state vector representing the landslide information at the current time,
、
the state vector is transferred to the matrix during the update process for landslide information,
is the control input when the Kalman filtering module processes the data information,
representing the excitation noise of the kalman filtering module,
the vector sequence number of the landslide information is represented, and the landslide information state updating process can be represented as follows:
in the formula (2), the first and second groups,
representing a landslide information vector data transfer matrix,
a state update function representing the landslide information,
a sequence of monitored information indicative of a landslide,
representing the best estimation value of the landslide information state; linear recursion filtering is carried out through Kalman filtering, the optimal estimator of the current data is calculated, vectors of all landslide information are represented,
representing a landslide information state vector at the previous moment;
and representing the influence quantity parameter of the optimal estimation value of the landslide information state.
As a further technical scheme of the invention, the method comprises the following steps
When the deformation rate of the landslide crack is between 0 and 4, the deformation rate of the landslide crack is more than 25mm/d, the accumulated deformation amount in the horizontal direction is more than 125mm, the average deformation rate in the vertical direction is more than 30mm/d, the accumulated acceleration and the accumulated jerk are more than 0, the 5-day accumulated deformation amount of a crack meter is more than 125mm, the daily rainfall is more than 200mm or the water content of soil is more than 75 percent;
As a further technical solution of the present invention, the support vector machine maps the input data information of the landslide crack deformation rate, the horizontal direction accumulated deformation amount, the vertical direction average deformation rate, the accumulated acceleration, the accumulated jerk, the crack, the accumulated deformation amount, the daily rainfall and the soil water content vector to the high latitude space through the nonlinear transformation, and the data information is expressed as:
in the formula (3), the first and second groups,
、
representing parameters determined by classifying the deformation rate of the landslide cracks, the accumulated deformation quantity in the horizontal direction, the average deformation rate in the vertical direction, the accumulated acceleration, the accumulated jerk, the cracks, the accumulated deformation quantity, the daily rainfall and the soil water content support vector,
a penalty factor for the SVM is represented,
which represents the relaxation factor of the SVM,
、
a training sequence representing the input is shown,
representing the number of samples; the classification equation for an SVM can be expressed as:
in the formula (4), the first and second groups,
representing the kernel function of an SVM for computing the vector inner product computationThe early warning model searches for the optimal characteristic step length by constructing a sequence, and the calculation formula is expressed as follows:
in the formula (5), the first and second groups,
characteristic information representing a landslide, N representing a serial number of a data sample, p representing an arbitrary integer value, q representing a characteristic step size,
a formula representing the optimal characteristic step output is shown,
the data information sequence of the optimal characteristic step size.
As a further technical scheme of the invention, the data information sequence of the optimal characteristic step length is in direct proportion to the dangerous case degree.
The invention has the beneficial and positive effects that:
different from the conventional technology, the method and the device realize the acquisition of the landslide prevention data information of the steep slope through terminal equipment, wherein the terminal equipment at least comprises a monitoring terminal, a rainfall monitoring terminal, a ground surface crack monitoring terminal, a deep displacement monitoring terminal or a soil moisture meter; processing multi-source heterogeneous data information is achieved through a K-means clustering algorithm; receiving the information characteristics of the steep slope landslide prevention data in a wireless communication mode or a serial communication mode, extracting the information characteristics of the steep slope landslide prevention data through a convolutional neural network and a long-short term memory neural network, and analyzing the extracted information characteristics of the steep slope landslide prevention data; constructing an early warning model to realize analysis of data information of the steep slope landslide prevention, wherein the early warning model is a KM-SVM landslide real-time early warning model;
the method and the device can realize analysis and mining of more specific information of the steep slope landslide prevention danger information and improve the analysis capability of data information. The slope deformation rate, the acceleration, the accumulated deformation quantity, the cracks, the rainfall and the soil water content can be used as comprehensive early warning indexes, a high-precision multi-source information fusion early warning system is formed according to early warning threshold values of different levels, and when the slope crack deformation rate is larger than 25mm/d, the accumulated deformation quantity in the horizontal direction is larger than 125mm, the average deformation rate in the vertical direction is larger than 30mm/d, the accumulated acceleration and the accumulated acceleration are larger than 0, the 5-day accumulated deformation quantity of a crack meter is larger than 125mm, the daily rainfall is larger than 200mm or the soil water content is larger than 75%, the steep slope landslide is in a high-risk state; when the deformation rate of the landslide crack is less than 6mm/d, the accumulated deformation amount in the horizontal direction is less than 30mm, the average deformation rate in the vertical direction is less than 10mm/d, the accumulated acceleration and the accumulated jerk are less than 0, the 5-day accumulated deformation amount of the crack meter is less than 30mm, the daily rainfall is less than 50mm or the water content of soil is within the range of 30% -45%, the abrupt slope landslide prevention has no landslide risk.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive exercise, wherein:
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a system diagram of an embodiment of the present invention;
FIG. 3 is a schematic diagram of a device terminal according to an embodiment of the present invention;
fig. 4 is a schematic diagram of the KM-SVM landslide real-time early warning model of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, and it should be understood that the embodiments described herein are merely for the purpose of illustrating and explaining the present invention and are not intended to limit the present invention.
As shown in fig. 1, a method for analyzing risk monitoring information of a steep slope antiskid includes:
step one, acquiring landslide prevention data information of a steep slope;
in the step, the acquisition of the landslide prevention data information of the steep slope is realized through terminal equipment, wherein the terminal equipment at least comprises a monitoring terminal, a rainfall monitoring terminal, a ground surface crack monitoring terminal, a deep displacement monitoring terminal or a soil moisture meter; processing multi-source heterogeneous data information is achieved through a K-means clustering algorithm;
extracting the information characteristics of the landslide prevention data of the steep slope;
in the step, the steep slope landslide prevention data information characteristic receiving is realized in a wireless communication mode or a serial communication mode, the steep slope landslide prevention data information characteristic is extracted through a convolutional neural network and a long-short term memory neural network, and the extracted steep slope landslide prevention data information characteristic is analyzed;
constructing an early warning model to realize analysis of data information of the steep slope landslide prevention, wherein the early warning model is a KM-SVM landslide real-time early warning model;
in the step, the slope deformation rate, the acceleration, the accumulated deformation quantity, the cracks, the rainfall and the soil water content are used as comprehensive early warning indexes, and a high-precision multi-source information fusion early warning system is formed according to respective early warning thresholds of different levels, wherein when the slope crack deformation rate is greater than 25mm/d, the accumulated deformation quantity in the horizontal direction is greater than 125mm, the average deformation rate in the vertical direction is greater than 30mm/d, the accumulated acceleration and the accumulated acceleration are greater than 0, the crack meter 5-day accumulated deformation quantity is greater than 125mm, the daily rainfall is greater than 200mm or the soil water content is greater than 75%, the steep slope antiskid is in a high-risk state; when the deformation rate of the landslide crack is less than 6mm/d, the accumulated deformation amount in the horizontal direction is less than 30mm, the average deformation rate in the vertical direction is less than 10mm/d, the accumulated acceleration and the accumulated jerk are less than 0, the 5-day accumulated deformation amount of the crack meter is less than 30mm, the daily rainfall is less than 50mm or the water content of soil is within the range of 30% -45%, the abrupt slope landslide prevention has no landslide risk.
In one embodiment, in the step one, the terminal device is a displacement sensor, wherein the displacement sensor adopts an STM32F105RBT 632 single chip microcomputer, the main frequency can reach 72MHZ, and an 8MHZ active crystal oscillator is adopted as a clock module of the single chip microcomputer.
In a specific embodiment, as shown in fig. 2, the stability and the anti-interference capability of the system can be effectively improved. The data transmission mode of the displacement sensor comprises a CAN bus, an Ethernet, optical fiber communication, wireless network communication and the like, the supported transmission distance is long, the real-time performance of data acquisition and action is high, the electromagnetic interference resistance is high, and the transmission rate CAN reach 1Mbps/40m at most. The CAN bus adopts a serial communication mode, has data processing priority and judgment functions, improves data communication efficiency, and CAN automatically identify nodes with serious errors in communication so as not to influence the data communication of other nodes on the bus.
In order to adapt to complex working environment, EMC protection and switch conversion are added into the power module, and fuses, piezoresistors, safety capacitors and anti-reverse diodes are used in an EMC protection circuit. A self-recovery fuse is added at the front end of the power input to perform overcurrent protection, the load can recover to work after being disconnected and electrified again, and the voltage dependent resistor plays a role in inputting an overvoltage protection rear-stage circuit. LM2596-ADJ is used as a switching power supply conversion chip in the switch conversion module, the input voltage is 40V at most, the highest 3A current can be output, the load capacity is high, and the voltage regulation output can be realized by sending data through a communication line. The MB85RS25 memory is used in the displacement sensor to realize the storage and reading of each setting parameter, and the memory is controlled by the SPI interface of the main control module, so that the set parameters can be stored in a power-down mode.
In the first step, the time interval for acquiring the landslide prevention data information of the steep slope is 30 minutes.
In the first step, an encoder and an Adam optimizer are arranged in the K-means clustering algorithm, the encoder is used for realizing encoding of different data information based on a PLC control module, and the Adam optimizer realizes updating of input data information after gradient oscillation and filtering oscillation of historical steep slope antiskid dangerous data information.
The method can acquire multisource heterogeneous data of the landslide prevention of the steep slope, wherein the rainfall monitoring terminal automatically acquires rainfall data of a monitoring area and records the rainfall value and the measurement time. The earth surface crack monitoring terminal automatically collects earth surface crack variable quantity, the deep displacement monitoring terminal collects inclination measuring sensor signals at regular time according to a set period, and the signals are converted into displacement values. The main monitoring information of the steep slope and the landslide prevention obtained by the monitoring analysis method comprises information such as ground surface cracks, ground surface displacement, a main sliding zone, underground water level, internal seepage, internal stress, precipitation, ground surface deformation and the like. In a specific embodiment, different interfaces are adopted for multi-source heterogeneous steep slope antiskid monitoring data, real-time data, text data and relational data in original data are automatically classified and collected, and collection driving of multiple protocols such as MODBUS, OPC, DAT, CSV, ADO and ODBC is achieved through data exchange.
In the second step, the convolutional neural network comprises an input matrix, a convolutional kernel, a feature extraction module, a data normalization module and an output module, wherein the output end of the input matrix is connected with the input end of the convolutional kernel, the output end of the convolutional kernel is connected with the input end of the feature extraction module, the output end of the feature extraction module is connected with the input end of the data normalization module, and the output end of the data normalization module is connected with the input end of the output module.
In a specific embodiment, in the convolutional neural network model, the calculation speed of the convolutional neural network model can be improved by fusing an input matrix, a convolutional kernel, a feature extraction module, a data normalization module, an output module and the like,
in the second step, the long-short term memory neural network model comprises a forgetting gate, an input gate and an output gate which are provided with an accelerator, wherein the accelerator is an acceleration module based on PFGA control. By the methods, the working efficiency of the long-term and short-term memory neural network model can be accelerated. And the data computing capacity is improved.
In the above steps, the convolution kernel and the input matrix are convolved to extract the characteristics of the input matrix, the characteristic diagram obtained after the convolution operation is used as the input of the next layer of convolution layer, the next layer of convolution performs convolution operation on the characteristic diagram to obtain a new characteristic diagram, and then the data distribution of the output characteristic diagram is adjusted through the normalization layer of the model, so that the model can learn the rules in the monitoring data more easily.
In the above steps, the LSTM (long short term memory, LSTM) long and short term memory neural network model belongs to one kind of recurrent neural network, and can be used to process data information with time sequence attributes, and the wind power data has time sequence characteristics, so that the prediction of the wind power can be realized based on the LSTM. The accumulated speed of the information is controlled by introducing a threshold mechanism, and the accumulated information before forgetting can be selected, so that the problem of gradient explosion or gradient dispersion is effectively solved.
In one embodiment, the forgetting gate is used to determine which long-term memories need to be erased in order to reduce the effect of the previous cell state on the next cell state; wherein the input gate is used to determine which part of the information in the input portion needs to be retained and to add new information to the cell state; the output gate needs to output the new cell state to the next step and calculate the new hidden state. By the structure, the LSTM unit can learn and recognize important input (input gate function), store the state of long-term evolution, save necessary time (forget gate function), further extract the memory required by current output, and realize strong adaptability to long-term sequences.
In the third step, the KM-SVM landslide real-time early warning model comprises a Kalman filtering module and a support vector machine module, wherein the Kalman filtering module is used for predicting dangerous data information, and the support vector machine module is used for diagnosing the dangerous data information.
In the above embodiment, the kalman filtering module performs data prediction by using a dynamic data processing recursion algorithm, where the data types include a landslide crack deformation rate, a horizontal accumulated deformation amount, a vertical average deformation rate, an accumulated acceleration, an accumulated jerk, a crack, an accumulated deformation amount, a daily rainfall, and a soil water content;
observing the data information through the latest moment to observe the random difference data information of the landslide information state, wherein a prediction model of a Kalman filtering module is as follows:
in the formula (1), the first and second groups,
a state vector representing the landslide information at the current time,
、
the state vector is transferred to the matrix during the update process for landslide information,
is the control input when the Kalman filtering module processes the data information,
representing the excitation noise of the kalman filtering module,
the vector sequence number of the landslide information is represented, and the landslide information state updating process can be represented as follows:
in the formula (2), the first and second groups,
representing a landslide information vector data transfer matrix,
a state update function representing the landslide information,
a sequence of monitored information indicative of a landslide,
representing the best estimation value of the landslide information state; linear recursive filtering is carried out through Kalman filtering, and work is calculatedThe best estimator of the previous data, the vector representing all landslide information,
representing a landslide information state vector at the previous moment;
and representing the influence quantity parameter of the optimal estimation value of the landslide information state.
In the above embodiment, when
When the deformation rate of the landslide crack is between 0 and 4, the deformation rate of the landslide crack is larger than 25mm/d, the accumulated deformation amount in the horizontal direction is larger than 125mm, the average deformation rate in the vertical direction is larger than 30mm/d, the accumulated acceleration and the accumulated jerk are larger than 0, the 5-day accumulated deformation amount of a crack meter is larger than 125mm, the daily rainfall is larger than 200mm, or the water content of soil is larger than 75%.
In a specific embodiment, when the deformation rate of the landslide crack is more than 25mm/d, the accumulated deformation amount in the horizontal direction is more than 125mm, the average deformation rate in the vertical direction is more than 30mm/d, the accumulated acceleration and the accumulated jerk are more than 0, the 5-day accumulated deformation amount of the crack meter is more than 125mm, the daily rainfall is more than 200mm or the soil water content is more than 75 percent, the method can be converted into the method
The data conversion of the data information is realized by changing the value, and the influence quantity of the external data information is converted into function representation, thereby realizing the expression of the data information.
In the above embodiment, when
When the deformation rate of the landslide crack is larger than 4, the deformation rate of the landslide crack is smaller than 6mm/d, the accumulated deformation amount in the horizontal direction is smaller than 30mm, the average deformation rate in the vertical direction is smaller than 10mm/d, the accumulated acceleration and the accumulated jerk are smaller than 0, the 5-day accumulated deformation amount of the crack meter is smaller than 30mm, the daily rainfall is smaller than 50mm, or the water content of the soil is within the range of 30% -45%.
In a specific embodiment, Kalman Filtering (KF) is combined with a Support Vector Machine (SVM) to realize prediction and evaluation of various data information of the steep slope landslide prevention. Kalman filtering (Kalman filtering) is an algorithm that uses a linear system state equation to optimally estimate the state of a system by inputting and outputting observation data through the system. The optimal estimation can also be seen as a filtering process, since the observed data includes the effects of noise and interference in the system. Data filtering is a data processing technique for removing noise and restoring true data, and Kalman filtering can estimate the state of a dynamic system from a series of data with measurement noise under the condition that measurement variance is known. And then, obtaining the data of whether the deformation rate of the landslide crack is greater than 25mm/d, whether the accumulated deformation quantity in the horizontal direction is greater than 125mm, whether the average deformation rate in the vertical direction is greater than 30mm/d, whether the accumulated acceleration and the accumulated jerk are greater than 0, whether the 5-day accumulated deformation quantity of the crack meter is greater than 125mm, whether the daily rainfall is greater than 200mm or whether the soil water content is greater than 75%, wherein the data is data information monitored and analyzed according to the long-term steep slope antiskid risk data information. When in these ranges, a certain risk will occur.
In a specific embodiment, the support vector machine maps the input landslide crack deformation rate, horizontal accumulated deformation quantity, vertical average deformation rate, accumulated acceleration, accumulated jerk, crack, accumulated deformation quantity, daily rainfall and soil water content vector data information to a high-latitude space through nonlinear transformation, and the data information is expressed as:
in the formula (3), the first and second groups,
、
presentation pairThe support vector classification determining parameters of the deformation rate of the landslide cracks, the accumulated deformation quantity in the horizontal direction, the average deformation rate in the vertical direction, the accumulated acceleration, the accumulated jerk, the cracks, the accumulated deformation quantity, the daily rainfall and the soil water content,
a penalty factor for the SVM is represented,
which represents the relaxation factor of the SVM,
、
a training sequence representing the input is shown,
representing the number of samples; the classification equation for SVMs may be expressed as:
in the formula (4), the first and second groups,
the kernel function of the SVM is represented and used for calculating vector inner product calculation, the early warning model searches the optimal characteristic step length by constructing a sequence, and the calculation formula is represented as follows:
in the formula (5), the first and second groups,
characteristic information representing a landslide, N representing a serial number of a data sample, p representing an arbitrary integer value, q representing a characteristic step size,
a formula representing the optimal characteristic step output is shown,
the data information sequence of the optimal characteristic step size.
In the above embodiment, the data information sequence of the optimal feature step size is proportional to the risk level. The larger the data information sequence output of the optimal characteristic step length is, the larger the details are, and the smaller the data information sequence of the optimal characteristic step length is, the smaller the dangerous case is.
In one embodiment, the data information analysis is implemented by a loss function (loss function). When a classification problem does not have linear separability, the use of hyperplanes as decision boundaries results in a classification penalty, i.e., a portion of the support vectors are no longer located on the interval boundaries, but instead enter the interval boundaries or fall on the wrong side of the decision boundary. The relevance research of classification loss and substitution loss through the loss function can be quantified, and the result obtained by solving the proxy loss minimization is also the solution of 0-1 loss minimization when the proxy loss is a continuous convex function and is the upper bound of the 0-1 loss function under any value. A Support Vector Machine (SVM) is a generalized linear classifier (generalized linear classifier) that binary classifies data according to a supervised learning (supervised learning) mode, and a decision boundary of the SVM is a maximum-margin hyperplane (maximum-margin hyperplane) that solves learning samples. The SVM calculates an empirical risk (empirical risk) using a hinge loss function (change loss) and adds a regularization term to a solution system to optimize a structural risk (structural risk), which is a classifier with sparsity and robustness. SVMs can be classified non-linearly by a kernel method, which is one of the common kernel learning (kernel learning) methods.
The early warning model sends out real-time early warning by analyzing and extracting precursor features in landslide monitoring data in a certain period, information such as displacement, speed and acceleration of landslide is used as an input vector and sent to a filter to obtain a predicted value, and the early warning is carried out by using the trained early warning model.
Although specific embodiments of the present invention have been described above, it will be understood by those skilled in the art that these specific embodiments are merely illustrative and that various omissions, substitutions and changes in the form of the detail of the methods and systems described above may be made by those skilled in the art without departing from the spirit and scope of the invention. For example, it is within the scope of the present invention to combine the steps of the above-described methods to perform substantially the same function in substantially the same way to achieve substantially the same result. Accordingly, the scope of the invention is to be limited only by the following claims.
Claims (10)
1. A method for analyzing monitoring information of landslide prevention danger of a steep slope is characterized by comprising the following steps: the method comprises the following steps:
step one, acquiring landslide prevention data information of a steep slope;
in the step, the acquisition of the landslide prevention data information of the steep slope is realized through terminal equipment, wherein the terminal equipment at least comprises a monitoring terminal, a rainfall monitoring terminal, a ground surface crack monitoring terminal, a deep displacement monitoring terminal or a soil moisture meter; processing multi-source heterogeneous data information is achieved through a K-means clustering algorithm;
extracting the information characteristics of the landslide prevention data of the steep slope;
in the step, the steep slope landslide prevention data information characteristic receiving is realized in a wireless communication mode or a serial communication mode, the steep slope landslide prevention data information characteristic is extracted through a convolutional neural network and a long-short term memory neural network, and the extracted steep slope landslide prevention data information characteristic is analyzed;
constructing an early warning model to realize analysis of data information of the steep slope landslide prevention, wherein the early warning model is a KM-SVM landslide real-time early warning model;
in the step, the slope deformation rate, the acceleration, the accumulated deformation quantity, the cracks, the rainfall and the soil water content are used as comprehensive early warning indexes, and a high-precision multi-source information fusion early warning system is formed according to respective early warning thresholds of different levels, wherein when the slope crack deformation rate is greater than 25mm/d, the accumulated deformation quantity in the horizontal direction is greater than 125mm, the average deformation rate in the vertical direction is greater than 30mm/d, the accumulated acceleration and the accumulated acceleration are greater than 0, the crack meter 5-day accumulated deformation quantity is greater than 125mm, the daily rainfall is greater than 200mm or the soil water content is greater than 75%, the steep slope antiskid is in a high-risk state; when the deformation rate of the landslide crack is less than 6mm/d, the accumulated deformation amount in the horizontal direction is less than 30mm, the average deformation rate in the vertical direction is less than 10mm/d, the accumulated acceleration and the accumulated jerk are less than 0, the 5-day accumulated deformation amount of the crack meter is less than 30mm, the daily rainfall is less than 50mm or the water content of soil is within the range of 30% -45%, the abrupt slope landslide prevention has no landslide risk.
2. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 1, wherein: the terminal equipment is a displacement sensor, wherein the displacement sensor adopts an STM32F105RBT 632-bit singlechip, the main frequency can reach 72MHZ, and an 8MHz active crystal oscillator is adopted as a clock module of the singlechip.
3. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 1, wherein: the time interval for acquiring the landslide prevention data information of the steep slope is 30 minutes.
4. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 1, wherein: in the first step, an encoder and an Adam optimizer are arranged in the K-means clustering algorithm, the encoder is used for realizing encoding of different data information based on a PLC control module, and the Adam optimizer realizes updating of input data information after gradient oscillation and filtering oscillation of historical steep slope antiskid dangerous data information.
5. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 1, wherein: the convolutional neural network comprises an input matrix, a convolutional kernel, a feature extraction module, a data normalization module and an output module, wherein the output end of the input matrix is connected with the input end of the convolutional kernel, the output end of the convolutional kernel is connected with the input end of the feature extraction module, the output end of the feature extraction module is connected with the input end of the data normalization module, and the output end of the data normalization module is connected with the input end of the output module;
the long-short term memory neural network model comprises a forgetting gate, an input gate and an output gate which are provided with an accelerator, wherein the accelerator is an acceleration module based on PFGA control.
6. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 1, wherein: the KM-SVM landslide real-time early warning model comprises a Kalman filtering module and a support vector machine module, wherein the Kalman filtering module is used for predicting dangerous data information, and the support vector machine module is used for diagnosing the dangerous data information.
7. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 6, wherein: the Kalman filtering module realizes data prediction through a dynamic data processing recursion algorithm, and the data types comprise landslide crack deformation rate, horizontal accumulated deformation quantity, vertical average deformation rate, accumulated acceleration, cracks, accumulated deformation quantity, daily rainfall and soil water content; observing the data information through the latest moment to observe the random difference data information of the landslide information state, wherein a prediction model of a Kalman filtering module is as follows:
in the formula (1), the first and second groups,
a state vector representing the landslide information at the current time,
、
the state vector is transferred to the matrix during the update process for landslide information,
is the control input when the Kalman filtering module processes the data information,
representing the excitation noise of the kalman filtering module,
the vector sequence number of the landslide information is represented, and the landslide information state updating process can be represented as follows:
in the formula (2), the first and second groups,
representing a landslide information vector data transfer matrix,
a state update function representing the landslide information,
a sequence of monitored information indicative of a landslide,
representing the best estimation value of the landslide information state; linear recursion filtering is carried out through Kalman filtering, the optimal estimator of the current data is calculated, vectors of all landslide information are represented,
indicating the state direction of the landslide information at the previous momentAn amount;
and representing the influence quantity parameter of the optimal estimation value of the landslide information state.
8. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 7, wherein: when in use
When the deformation rate of the landslide crack is between 0 and 4, the deformation rate of the landslide crack is more than 25mm/d, the accumulated deformation amount in the horizontal direction is more than 125mm, the average deformation rate in the vertical direction is more than 30mm/d, the accumulated acceleration and the accumulated jerk are more than 0, the 5-day accumulated deformation amount of a crack meter is more than 125mm, the daily rainfall is more than 200mm or the water content of soil is more than 75 percent;
when the deformation rate of the landslide crack is larger than 4, the deformation rate of the landslide crack is smaller than 6mm/d, the accumulated deformation amount in the horizontal direction is smaller than 30mm, the average deformation rate in the vertical direction is smaller than 10mm/d, the accumulated acceleration and the accumulated jerk are smaller than 0, the 5-day accumulated deformation amount of the crack meter is smaller than 30mm, the daily rainfall is smaller than 50mm, or the water content of the soil is within the range of 30% -45%.
9. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 6, wherein: the support vector machine maps input landslide crack deformation rate, horizontal direction accumulated deformation quantity, vertical direction average deformation rate, accumulated acceleration, cracks, accumulated deformation quantity, daily rainfall and soil water content vector data information to a high latitude space through nonlinear conversion, and the data is expressed as follows through a hyperplane:
in the formula (3), the first and second groups,
、
representing parameters determined by classifying the deformation rate of the landslide cracks, the accumulated deformation quantity in the horizontal direction, the average deformation rate in the vertical direction, the accumulated acceleration, the accumulated jerk, the cracks, the accumulated deformation quantity, the daily rainfall and the soil water content support vector,
a penalty factor for the SVM is represented,
which represents the relaxation factor of the SVM,
、
a training sequence representing the input is shown,
representing the number of samples; the classification equation for SVMs may be expressed as:
in the formula (4), the first and second groups,
the kernel function of the SVM is represented and used for calculating vector inner product calculation, the early warning model searches the optimal characteristic step length by constructing a sequence, and the calculation formula is represented as follows:
in the formula (5), the first and second groups,
characteristic information representing a landslide, N representing a serial number of a data sample, p representing an arbitrary integer value, q representing a characteristic step size,
a formula representing the optimal characteristic step output is shown,
the data information sequence of the optimal characteristic step size.
10. The method for analyzing the monitoring information of the danger of the steep slope landslide, according to claim 9, wherein: the data information sequence of the optimal characteristic step size is proportional to the danger degree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210754227.XA CN114818518B (en) | 2022-06-30 | 2022-06-30 | Method for analyzing monitoring information of landslide prevention danger of abrupt slope |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210754227.XA CN114818518B (en) | 2022-06-30 | 2022-06-30 | Method for analyzing monitoring information of landslide prevention danger of abrupt slope |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114818518A true CN114818518A (en) | 2022-07-29 |
| CN114818518B CN114818518B (en) | 2022-09-13 |
Family
ID=82523407
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210754227.XA Active CN114818518B (en) | 2022-06-30 | 2022-06-30 | Method for analyzing monitoring information of landslide prevention danger of abrupt slope |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114818518B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115273410A (en) * | 2022-09-09 | 2022-11-01 | 西北大学 | Sudden landslide monitoring and early warning system based on big data |
| CN115410342A (en) * | 2022-08-26 | 2022-11-29 | 安徽省地质矿产勘查局332地质队 | Landslide disaster intelligent early warning method based on crack meter real-time monitoring |
| CN115638767A (en) * | 2022-11-07 | 2023-01-24 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Ground settlement monitoring method |
| CN116956046A (en) * | 2023-09-07 | 2023-10-27 | 西南交通大学 | Earthquake landslide hazard analysis method and device based on cyclic neural network |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070152804A1 (en) * | 1997-10-22 | 2007-07-05 | Intelligent Technologies International, Inc. | Accident Avoidance Systems and Methods |
| CN105627978A (en) * | 2016-01-27 | 2016-06-01 | 中国铁道科学研究院电子计算技术研究所 | Rainy season land slide deformation monitoring method and rainy season land slide deformation monitoring system |
| US20170091637A1 (en) * | 2015-09-30 | 2017-03-30 | Hampton Creek, Inc. | Discovery systems for identifying entities that have a target property |
| CN108010280A (en) * | 2017-12-26 | 2018-05-08 | 成都理工大学 | A kind of sudden Loess Landslide method for early warning and its application |
| CN108317949A (en) * | 2018-02-07 | 2018-07-24 | 桂林电子科技大学 | A kind of RTK high-precision differences positioning deformation monitoring system and method |
| CN111896953A (en) * | 2020-07-30 | 2020-11-06 | 华能澜沧江水电股份有限公司 | Time sequence InSAR basin bank deformation prediction method based on self-adaptive KF |
| CN114119485A (en) * | 2021-10-28 | 2022-03-01 | 深圳特科动力技术有限公司 | Mixed vision slope detection method |
-
2022
- 2022-06-30 CN CN202210754227.XA patent/CN114818518B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070152804A1 (en) * | 1997-10-22 | 2007-07-05 | Intelligent Technologies International, Inc. | Accident Avoidance Systems and Methods |
| US20170091637A1 (en) * | 2015-09-30 | 2017-03-30 | Hampton Creek, Inc. | Discovery systems for identifying entities that have a target property |
| CN105627978A (en) * | 2016-01-27 | 2016-06-01 | 中国铁道科学研究院电子计算技术研究所 | Rainy season land slide deformation monitoring method and rainy season land slide deformation monitoring system |
| CN108010280A (en) * | 2017-12-26 | 2018-05-08 | 成都理工大学 | A kind of sudden Loess Landslide method for early warning and its application |
| CN108317949A (en) * | 2018-02-07 | 2018-07-24 | 桂林电子科技大学 | A kind of RTK high-precision differences positioning deformation monitoring system and method |
| CN111896953A (en) * | 2020-07-30 | 2020-11-06 | 华能澜沧江水电股份有限公司 | Time sequence InSAR basin bank deformation prediction method based on self-adaptive KF |
| CN114119485A (en) * | 2021-10-28 | 2022-03-01 | 深圳特科动力技术有限公司 | Mixed vision slope detection method |
Non-Patent Citations (1)
| Title |
|---|
| 武永波: "基于物联网的滑坡灾害监测预警技术研究", 《中国优秀博硕士学位论文全文数据库(博士)基础科学辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115410342A (en) * | 2022-08-26 | 2022-11-29 | 安徽省地质矿产勘查局332地质队 | Landslide disaster intelligent early warning method based on crack meter real-time monitoring |
| CN115410342B (en) * | 2022-08-26 | 2023-08-11 | 安徽省地质矿产勘查局332地质队 | Landslide hazard intelligent early warning method based on real-time monitoring of crack meter |
| CN115273410A (en) * | 2022-09-09 | 2022-11-01 | 西北大学 | Sudden landslide monitoring and early warning system based on big data |
| CN115273410B (en) * | 2022-09-09 | 2023-08-25 | 西北大学 | Sudden landslide monitoring and early warning system based on big data |
| CN115638767A (en) * | 2022-11-07 | 2023-01-24 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Ground settlement monitoring method |
| CN115638767B (en) * | 2022-11-07 | 2023-10-03 | 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) | Ground subsidence monitoring method |
| CN116956046A (en) * | 2023-09-07 | 2023-10-27 | 西南交通大学 | Earthquake landslide hazard analysis method and device based on cyclic neural network |
| CN116956046B (en) * | 2023-09-07 | 2023-12-29 | 西南交通大学 | Earthquake landslide hazard analysis method and device based on cyclic neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114818518B (en) | 2022-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114818518B (en) | Method for analyzing monitoring information of landslide prevention danger of abrupt slope | |
| CN111882869B (en) | Deep learning traffic flow prediction method considering adverse weather | |
| CN112800634B (en) | Rainfall estimation method and system coupling dry-wet state identification and multi-source information fusion | |
| Xing et al. | A hybrid prediction model of landslide displacement with risk-averse adaptation | |
| CN110929946A (en) | Agricultural flood disaster monitoring and forecasting system and method based on surface hydrologic process | |
| CN111866452B (en) | Video real-time monitoring method for emergency disposal of landslide and damming dam | |
| CN113886917B (en) | CNN-LSTM model-based railway line region ground settlement prediction early warning method | |
| CN112287528B (en) | Flood control high water level determination method for sandy river reservoir based on high beach trough | |
| US11174624B2 (en) | Method for identifying number of shoveling-and-loading processes of loader and device for identifying the number of the shoveling-and-loading processes of the loader | |
| Sahagun et al. | Wireless sensor nodes for flood forecasting using artificial neural network | |
| Liu et al. | Slope displacement prediction using sequential intelligent computing algorithms | |
| CN112508060B (en) | Landslide body state judging method and system based on graph convolution neural network | |
| CN106772354B (en) | Method for tracking target and device based on parallel fuzzy gaussian sum particle filter | |
| Ghaderi et al. | Shoreline change analysis along the coast of Bandar Abbas city, Iran using remote sensing images | |
| CN117113830B (en) | Mountain restoration evaluation prediction system based on data analysis | |
| CN115019476B (en) | Landslide space-time information monitoring and early warning method based on multi-source information fusion | |
| CN112946240B (en) | Landslide geological disaster gene identification and prediction system | |
| CN110610187A (en) | Soil slope collapse accident diagnosis method based on Bayesian network | |
| Das et al. | Real-Time Threshold-based Landslide Prediction System for Hilly Region using Wireless Sensor Networks | |
| CN116990847B (en) | Beidou GNSS receiver resolving method and system based on edge calculation | |
| CN111860973B (en) | Debris flow intelligent early warning method based on multi-objective optimization | |
| CN111275298B (en) | Geological disaster risk early warning method based on power grid | |
| EVELPIDOU et al. | Fuzzy modelling of slope erosion by runoff. Case study in Corinth basin, Greece | |
| CN117952278A (en) | Data processing system and method for data collection management | |
| Li et al. | Study on geological meaning and application of mine water inflow series after phase space reconstruction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |