A method for analyzing earth-rock dam failure under earthquake and flood

Abstract

The invention relates to the technical field of earth-rock dam failure analysis, and discloses an earth-rock dam failure analysis method under the action of earthquake and flood, comprising the following steps: step S101, dividing the earth-rock dam to be monitored into M areas to be monitored; step S102, collecting vibration signal data, hydrological data and dam body data of each area to be monitored, and generating a comprehensive feature matrix; step S103, inputting the comprehensive feature matrix into an earth-rock dam risk prediction model, and the output value represents the dam failure risk score of the earth-rock dam to be monitored; step S104, taking corresponding early warning measures according to the dam failure risk score; the invention comprehensively considers the influence of the vibration characteristics of seismic waves, the hydrological characteristics of the earth-rock dam and the dam body structure characteristics on the earth-rock dam in time and space through the earth-rock dam risk prediction model, and obtains the stress condition of the earth-rock dam failure in the whole life cycle with the help of finite element analysis software, so as to improve the accuracy of the earth-rock dam failure risk prediction.

Description

Technical Field

[0001] The invention relates to the technical field of earth-rock dam breach analysis, and more specifically, to an earth-rock dam breach analysis method under the action of earthquake and flood. Background Art

[0002] There are three main research methods for deformation and damage of earth-rock dams caused by floods and earthquakes. The first is an intuitive analysis study through on-site investigation. Researchers conduct on-site surveys on earth-rock dams to obtain the structure, materials, stress conditions, etc. of the earth-rock dams to assist researchers in analyzing the stability of the earth-rock dams. The second is a prototype reproduction analysis study through appropriate scale and similar material model tests. With the help of model tests, researchers can observe the destruction process, damage mechanism and the influence of different parameters on the stability of earth-rock dams. The third is a mechanism study through numerical theoretical methods. Researchers consider factors such as the constitutive relationship of the dam body and boundary conditions by establishing mathematical models in order to conduct mechanical analysis and simulation of the earth-rock dam.

[0003] However, due to the limitations of on-site conditions, the field investigation and analysis mainly focused on the degree of damage to the dam body and the downstream disaster losses, making it difficult to observe the entire process of earth-rock dam failure. Model tests and numerical theoretical methods are limited by factors such as model scale and similar materials, and lack a systematic study on the catastrophic laws of the entire process of earth-rock dam destruction after flood and earthquake loads.

[0004] Therefore, there is an urgent need for an analysis method for earth-rock dam failure under earthquakes and floods to cover the entire life cycle of earth-rock dam failure. Summary of the invention

[0005] The present invention provides a method for analyzing earth-rock dam failure under earthquake and flood effects, which solves the technical problems in the above-mentioned background technology.

[0006] The present invention provides a method for analyzing earth-rock dam failure under earthquake and flood effects, comprising the following steps:

[0007] Step S101, dividing the edge of the earth-rock dam to be monitored into M areas to be monitored at the same distance intervals;

[0008] Step S102, within a preset collection time period T, collect vibration signal data, hydrological data and dam body data of each area to be monitored at a preset time interval t, and generate a comprehensive feature matrix;

[0009] The vibration signal data is represented by the acceleration corresponding to N time steps, where N represents the number of time steps, N=T / t;

[0010] The comprehensive feature matrix is ​​expressed as: in They represent the comprehensive feature vectors from the first monitored area to the Mth monitored area at the first time step, They represent the comprehensive feature vectors from the 1st time step to the Nth time step of the 1st monitored area, Represents the comprehensive feature vector of the Nth time step of the Mth area to be monitored;

[0011] in and They represent the vibration vector, hydrological vector and dam body vector of the Mth monitored area at the Nth time step respectively;

[0012] Step S103, inputting the comprehensive characteristic matrix into the earth-rock dam risk prediction model, and the output value represents the dam break risk score of the earth-rock dam to be monitored; the larger the dam break risk score, the higher the dam break risk;

[0013] The earth-rock dam risk prediction model is a nonlinear function model of vibration signal data, hydrological data, dam body data and dam break risk score;

[0014] The dam break risk score is a discrete value between 0 and 100;

[0015] Step S104, dividing the dam break risk score into three risk levels, and taking corresponding early warning measures respectively.

[0016] Furthermore, M, T, and t are all custom parameters, the default value of M is set to 8, the default value of T is set to 60 seconds, and the default value of t is set to 2 seconds.

[0017] Furthermore, the vibration vector of the Mth area to be monitored at the Nth time step is expressed as:

[0018]

[0019] in They respectively represent the energy values ​​corresponding to the frequencies of 0.01 Hz, 0.02 Hz to 19.99 Hz and 20 Hz in the Nth time step of the Mth area to be monitored.

[0020] Further, generating a vibration vector based on the vibration signal data comprises the following steps:

[0021] Step S201, performing pre-emphasis processing on the vibration signal data through a high-pass filter;

[0022] Step S202, performing noise reduction processing on the vibration signal data after pre-emphasis processing by wavelet transform;

[0023] Step S203: converting the vibration signal data after noise reduction processing into a vibration vector through fast Fourier transform.

[0024] Furthermore, the hydrological vector of the Mth monitored area at the Nth time step is expressed as:

[0025] in They respectively represent the water level, water level change rate, rainfall, rainfall change rate, seepage and seepage change rate of the Nth time step of the Mth monitored area; the water level change rate of the Nth time step is equal to the difference between the water level of the Nth time step and the water level of the N-1th time step divided by the preset time interval t; the rainfall change rate of the Nth time step is equal to the difference between the rainfall of the Nth time step and the rainfall of the N-1th time step divided by the preset time interval t; the seepage change rate of the Nth time step is equal to the difference between the infiltration rate of the Nth time step and the infiltration rate of the N-1th time step divided by the preset time interval t. The water level change rate, rainfall change rate and seepage change rate of the first time step are all set to 0.

[0026] Furthermore, the dam body vector of the Mth monitored area at the Nth time step is expressed as:

[0027] They respectively represent the pressure, displacement, number of cracks, sum of crack lengths, sum of crack widths, sum of crack depths and compressive strength of earth-rock dam materials in the Mth monitored area at the Nth time step; the compressive strength of earth-rock dam materials is a fixed value, which is obtained by researchers through experimental analysis during the design and construction of the earth-rock dam, and the unit is MPa.

[0028] Furthermore, the earth-rock dam risk prediction model includes: N hidden layers, each hidden layer includes: a first hidden unit, a second hidden unit and a third hidden unit;

[0029] The first hidden unit includes H channels, where H is a custom parameter;

[0030] Each channel of the first hidden unit of the i-th hidden layer inputs the comprehensive feature vector from the first to-be-monitored area to the M-th to-be-monitored area at the i-th time step in the comprehensive feature matrix, and each channel outputs the first updated feature;

[0031] The second hidden unit of the i-th hidden layer inputs the first updated features output by the H channels of the first hidden unit of the i-th hidden layer, and outputs the second updated features;

[0032] The third hidden unit of the i-th hidden layer inputs the second updated feature output by the second hidden unit of the i-th hidden layer, and outputs a third updated feature;

[0033] The third hidden unit is constructed based on a gated recurrent unit;

[0034] The third updated feature output by the third hidden unit of the Nth hidden layer is input into the classifier, and the classification space of the classifier represents the dam break risk score of the earth-rock dam to be monitored;

[0035] Definition: 1≤i≤N.

[0036] Furthermore, the calculation formula of the earth-rock dam risk prediction model includes:

[0037] The calculation formula for the first hidden unit of the i-th hidden layer is as follows:

[0038]

[0039]

[0040]

[0041]

[0042] Where 1≤j≤H, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, The comprehensive feature vector from the first to-be-monitored area to the Mth to-be-monitored area at the i-th time step in the comprehensive feature matrix of the j-th channel input of the first hidden unit of the i-th hidden layer, and denote the first conversion feature, the second conversion feature, and the third conversion feature of the jth channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first conversion parameter, the second conversion parameter and the third conversion parameter of the j-th channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first weight parameter and the second weight parameter of the first hidden unit of the i-th hidden layer, respectively, T denotes a transpose operation, tanh denotes a hyperbolic tangent activation function, and softmax denotes a softmax activation function;

[0043] The calculation formula for the second hidden unit of the i-th hidden layer is as follows:

[0044]

[0045] Where Z i represents the second updated feature output by the second hidden unit of the i-th hidden layer, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, W represents the concatenation of the first updated features of the H channels output by the first hidden unit of the i-th hidden layer.i Represents the weight parameter of the second hidden unit of the i-th hidden layer.

[0046] Furthermore, a training sample for training the earth-rock dam risk prediction model is obtained by finite element analysis software, wherein the training sample includes training data and training labels, and the following steps are included:

[0047] Step S301, constructing a digital model of the earth-rock dam to be monitored at a reduced scale with real geometric shape and material properties by finite element analysis software, and using triangular facets to represent each area to be monitored of the digital model of the earth-rock dam;

[0048] Step S302, adjusting the vibration signal data and hydrological data of each area to be monitored until the digital model of the earth-rock dam to be monitored breaks;

[0049] Step S303, recording the vibration signal data, hydrological data and dam body data of each monitored area after each adjustment as training data of training samples;

[0050] Step S304, obtaining the maximum principal stress and the maximum shear stress in all triangular facets corresponding to each monitored area after each adjustment through finite element analysis software, and recording the maximum resultant force value of each monitored area after each adjustment;

[0051] The calculation formula of the maximum resultant force F is as follows:

[0052]

[0053] Where stress represents the maximum principal stress and shear represents the maximum shear stress;

[0054] Step S305, repeating steps S302 to S304, taking the maximum resultant force value as a benchmark, normalizing the maximum resultant force value of each monitored area after each adjustment and mapping it to a dam break risk score, and using the dam break risk score as a training label for the training sample.

[0055] Furthermore, the risk levels include: low risk level, medium risk level and high risk level. The dam break risk score of 0 to 30 corresponds to the low risk level, and daily monitoring and inspection measures are taken for the low risk level; the dam break risk score of 31 to 75 corresponds to the medium risk level, and measures such as increasing monitoring points and improving monitoring frequency are taken for the medium risk level; the dam break risk score of 76 to 100 corresponds to the high risk level, and emergency warning and disposal measures are immediately initiated for the high risk level, including evacuating personnel, blocking traffic and notifying relevant departments.

[0056] The beneficial effects of the present invention are as follows: the present invention comprehensively considers the influence of the vibration characteristics of seismic waves, the hydrological characteristics of the earth-rock dam and the structural characteristics of the dam body on the earth-rock dam to be monitored in time and space by constructing a risk prediction model for earth-rock dams, and obtains the stress conditions of the earth-rock dam during the entire life cycle with the help of finite element analysis software, and establishes a nonlinear mapping relationship between the vibration characteristics of seismic waves, the hydrological characteristics of the earth-rock dam and the structural characteristics of the dam body and the stress conditions, thereby improving the accuracy of dam break risk prediction for the earth-rock dam to be monitored. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of a method for analyzing earth-rock dam failure under earthquake and flood effects of the present invention;

[0058] Figure 2 is a flow chart of generating a vibration vector based on vibration signal data of the present invention;

[0059] Figure 3 The present invention is a flow chart of obtaining training samples for training an earth-rock dam risk prediction model through finite element analysis software. DETAILED DESCRIPTION

[0060] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0061] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0062] like Figure 1-Figure 3As shown, a method for analyzing earth-rock dam failure under earthquake and flood action includes the following steps:

[0063] Step S101, dividing the edge of the earth-rock dam to be monitored into M areas to be monitored at the same distance intervals;

[0064] Step S102, within a preset collection time period T, collect vibration signal data, hydrological data and dam body data of each area to be monitored at a preset time interval t, and generate a comprehensive feature matrix;

[0065] The vibration signal data is represented by the acceleration corresponding to N time steps, where N represents the number of time steps, N=T / t;

[0066] The comprehensive feature matrix is ​​expressed as:

[0067] in They represent the comprehensive feature vectors from the first monitored area to the Mth monitored area at the first time step, They represent the comprehensive feature vectors from the 1st time step to the Nth time step of the 1st monitored area, Represents the comprehensive feature vector of the Nth time step of the Mth area to be monitored;

[0068] in and They represent the vibration vector, hydrological vector and dam body vector of the Mth monitored area at the Nth time step respectively;

[0069] Step S103, inputting the comprehensive characteristic matrix into the earth-rock dam risk prediction model, and the output value represents the dam break risk score of the earth-rock dam to be monitored;

[0070] A larger dam failure risk score indicates a higher dam failure risk;

[0071] The earth-rock dam risk prediction model is a nonlinear function model of vibration signal data, hydrological data, dam body data and dam break risk score;

[0072] The dam break risk score is a discrete value between 0 and 100;

[0073] Step S104, dividing the dam break risk score into three risk levels, and taking corresponding early warning measures respectively.

[0074] In one embodiment of the present invention, M, T and t are all custom parameters, the default value of M is set to 8, the default value of T is set to 60 seconds, and the default value of t is set to 2 seconds; specifically, M can be set according to the survey and evaluation of the earth-rock dam to be monitored by relevant experts.

[0075] In one embodiment of the present invention, the vibration vector of the Nth time step of the Mth area to be monitored is expressed as:

[0076] in They respectively represent the energy values ​​corresponding to the frequencies of 0.01 Hz, 0.02 Hz to 19.99 Hz and 20 Hz in the Nth time step of the Mth area to be monitored.

[0077] It should be noted that the frequency range of low-frequency seismic waves is below 0.1 Hz, the propagation speed is slow, and they can produce large ground movements, which have a greater impact on buildings and structures; the frequency range of medium-frequency seismic waves is between 0.1 Hz and 10 Hz, the propagation speed is fast, and there is a stronger sense of tremor; the frequency range of high-frequency seismic waves is above 10 Hz, the propagation speed is very fast, and there is a weaker vibration; therefore, the vibration vector is represented by the energy value corresponding to the frequency of the seismic wave of 0.01 to 20 Hz, which can more accurately characterize the degree of influence of seismic waves in different frequency ranges on earth-rock dams.

[0078] In one embodiment of the present invention, a vibration sensor is installed in each monitored area to collect vibration signal data of seismic waves in real time. The vibration sensor may be an accelerometer or a seismograph, and the vibration signal data is uploaded to a cloud server via mobile communication or satellite communication to generate a vibration vector.

[0079] In one embodiment of the present invention, Figure 2 As shown, generating a vibration vector based on vibration signal data includes the following steps:

[0080] Step S201, pre-emphasize the vibration signal data through a high-pass filter, that is, increase the high-frequency part of the vibration signal data, which helps to improve the signal-to-noise ratio;

[0081] Step S202, performing noise reduction processing on the vibration signal data after the pre-emphasis processing by wavelet transform, that is, removing the noise part in the vibration signal data;

[0082] Step S203, converting the vibration signal data after the noise reduction process into a vibration vector through fast Fourier transform, that is, realizing the conversion from the time domain to the frequency domain.

[0083] In one embodiment of the present invention, the hydrological vector of the Mth monitored area at the Nth time step is expressed as:

[0084] in They respectively represent the water level, water level change rate, rainfall, rainfall change rate, seepage and seepage change rate of the Nth time step of the Mth monitored area; the water level change rate of the Nth time step is equal to the difference between the water level of the Nth time step and the water level of the N-1th time step divided by the preset time interval t; the rainfall change rate of the Nth time step is equal to the difference between the rainfall of the Nth time step and the rainfall of the N-1th time step divided by the preset time interval t; the seepage change rate of the Nth time step is equal to the difference between the infiltration rate of the Nth time step and the infiltration rate of the N-1th time step divided by the preset time interval t. The water level change rate, rainfall change rate and seepage change rate of the first time step are all set to 0.

[0085] In one embodiment of the present invention, the water level of each monitored area is collected in real time by a water level sensor; the rainfall of each monitored area is collected in real time by a rain gauge; and the seepage volume of each monitored area is collected in real time by a seepage flow meter.

[0086] In one embodiment of the present invention, the dam body vector of the Mth area to be monitored at the Nth time step is expressed as: They respectively represent the pressure, displacement, number of cracks, sum of crack lengths, sum of crack widths, sum of crack depths and compressive strength of earth-rock dam materials in the Mth monitored area at the Nth time step; the compressive strength of earth-rock dam materials is a fixed value, which is obtained by researchers through experimental analysis during the design and construction of the earth-rock dam, and the unit is MPa.

[0087] In one embodiment of the present invention, pressure is collected in real time by installing a pressure sensor inside the dam body; displacement is collected in real time by installing a displacement sensor on the surface of the dam body, and the unit is micrometer; the number of cracks, the sum of crack lengths, the sum of crack widths and the sum of crack depths are obtained by collecting images of the dam body surface and inputting them into an existing target detection model, and the existing target detection model may be a YOLO series model or Faster R-CNN, etc.; a training data set for training the target detection model may be obtained by relevant experts using existing annotation tools to annotate images of the dam body surface, and the existing annotation tools may be LabelImg or CVAT, etc. The training of the target detection model is a conventional technical means and will not be elaborated here.

[0088] In one embodiment of the present invention, the earth-rock dam risk prediction model includes: N hidden layers, each hidden layer includes: a first hidden unit, a second hidden unit and a third hidden unit;

[0089] The first hidden unit includes H channels, where H is a custom parameter, and preferably, H is set to 8;

[0090] Each channel of the first hidden unit of the i-th hidden layer inputs the comprehensive feature vector from the first to-be-monitored area to the M-th to-be-monitored area at the i-th time step in the comprehensive feature matrix, and each channel outputs the first updated feature;

[0091] The second hidden unit of the i-th hidden layer inputs the first updated features output by the H channels of the first hidden unit of the i-th hidden layer, and outputs the second updated features;

[0092] The third hidden unit of the i-th hidden layer inputs the second updated feature output by the second hidden unit of the i-th hidden layer, and outputs a third updated feature;

[0093] The third hidden unit is constructed based on a gated recurrent unit;

[0094] The third updated feature output by the third hidden unit of the Nth hidden layer is input into the classifier, and the classification space of the classifier represents the dam break risk score of the earth-rock dam to be monitored;

[0095] Definition: 1≤i≤N.

[0096] It should be noted that the classifier is a Softmax classifier, and the output is the probability distribution of different dam break risk scores, that is, the probability value corresponding to the discrete value between 0 and 100, and the discrete value with the largest corresponding probability value is taken as the dam break risk score of the earth-rock dam to be monitored. For example; the probability value corresponding to the value 75 is 98%, which exceeds 50%. Then the dam break risk score of the earth-rock dam to be monitored is 75, indicating that the dam break risk of the earth-rock dam to be monitored is relatively high.

[0097] In one embodiment of the present invention, the calculation formula of the earth-rock dam risk prediction model includes:

[0098] The calculation formula for the first hidden unit of the i-th hidden layer is as follows:

[0099]

[0100]

[0101]

[0102]

[0103] Where 1≤j≤H, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, The comprehensive feature vector from the first to-be-monitored area to the Mth to-be-monitored area at the i-th time step in the comprehensive feature matrix of the j-th channel input of the first hidden unit of the i-th hidden layer, and denote the first conversion feature, the second conversion feature, and the third conversion feature of the jth channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first conversion parameter, the second conversion parameter and the third conversion parameter of the j-th channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first weight parameter and the second weight parameter of the first hidden unit of the i-th hidden layer, respectively, T denotes a transpose operation, tanh denotes a hyperbolic tangent activation function, and softmax denotes a softmax activation function;

[0104] The calculation formula for the second hidden unit of the i-th hidden layer is as follows:

[0105]

[0106] Where Z i represents the second updated feature output by the second hidden unit of the i-th hidden layer, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, W represents the concatenation of the first updated features of the H channels output by the first hidden unit of the i-th hidden layer. i Represents the weight parameter of the second hidden unit of the i-th hidden layer.

[0107] In one embodiment of the present invention, the calculation formula of the third hidden unit of the i-th hidden layer includes:

[0108] The reset gate r of the i-th hidden layer i The calculation formula is as follows:

[0109]

[0110] in and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the reset gate of the i-th hidden layer;

[0111] The update gate g of the i-th hidden layer i The calculation formula is as follows:

[0112]

[0113] in and They respectively represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the update gate of the i-th hidden layer;

[0114] Candidate hidden states of the i-th hidden layer The calculation formula is as follows:

[0115]

[0116] in and They represent the first weight parameter, the second weight parameter and the bias parameter corresponding to the candidate hidden state of the i-th hidden layer, respectively, and r i represents the reset gate of the i-th hidden layer, ⊙ represents point-by-point multiplication, and tanh represents the hyperbolic tangent function;

[0117] The third updated feature h output by the i-th hidden layer i The calculation formula is as follows:

[0118]

[0119] where g i and Represent the update gate and candidate hidden state of the i-th hidden layer respectively, and ⊙ represents point-by-point multiplication;

[0120] Definition: x i represents the second updated feature of the third hidden unit input of the i-th hidden layer, h i-1 represents the third updated feature output by the third hidden unit of the i-1th hidden layer, σ represents the sigmoid activation function, x0=0, h0=0.

[0121] In one embodiment of the present invention, the third hidden unit may also be constructed based on an RNN (recurrent neural network) unit or an LSTM (long short-term memory network) unit.

[0122] In one embodiment of the present invention, Figure 3 As shown, a training sample for training an earth-rock dam risk prediction model is obtained through finite element analysis software. The training sample includes training data and training labels, and includes the following steps:

[0123] Step S301, constructing a digital model of the earth-rock dam to be monitored at a reduced scale with real geometric shape and material properties by finite element analysis software, and using triangular facets to represent each area to be monitored of the digital model of the earth-rock dam;

[0124] Step S302, adjusting the vibration signal data and hydrological data of each area to be monitored until the digital model of the earth-rock dam to be monitored breaks;

[0125] Step S303, recording the vibration signal data, hydrological data and dam body data of each monitored area after each adjustment as training data of training samples;

[0126] Step S304, obtaining the maximum principal stress and the maximum shear stress in all triangular facets corresponding to each monitored area after each adjustment through finite element analysis software, and recording the maximum resultant force value of each monitored area after each adjustment;

[0127] The calculation formula of the maximum resultant force F is as follows:

[0128]

[0129] Where stress represents the maximum principal stress and shear represents the maximum shear stress;

[0130] Step S305, repeating steps S302 to S304, taking the maximum resultant force value as a benchmark, normalizing the maximum resultant force value of each monitored area after each adjustment and mapping it to a dam break risk score, and using the dam break risk score as a training label for the training sample.

[0131] It should be noted that the present invention represents the stress condition of the area to be monitored by comprehensively considering the maximum principal stress and the maximum shear stress, and normalizes the maximum resultant force value to a dam break risk score to more accurately represent the dam break risk of the area to be monitored.

[0132] In one embodiment of the present invention, the risk levels include: low risk level, medium risk level and high risk level. The dam break risk score of 0 to 30 corresponds to the low risk level, and daily monitoring and patrol measures are taken for the low risk level; the dam break risk score of 31 to 75 corresponds to the medium risk level, and measures such as increasing monitoring points and improving monitoring frequency are taken for the medium risk level; the dam break risk score of 76 to 100 corresponds to the high risk level, and emergency warning and disposal measures are immediately initiated for the high risk level, including evacuating personnel, blocking traffic, notifying relevant departments and the public, etc.

[0133] It should be noted that the present invention comprehensively considers the influence of the vibration characteristics of seismic waves, the hydrological characteristics of the earth-rock dam and the structural characteristics of the dam body on the earth-rock dam to be monitored in time and space by constructing a risk prediction model for earth-rock dams, and obtains the stress conditions of the earth-rock dam during the entire life cycle of dam failure with the help of finite element analysis software, and establishes a nonlinear mapping relationship between the vibration characteristics of seismic waves, the hydrological characteristics of the earth-rock dam and the structural characteristics of the dam body and the stress conditions, thereby improving the accuracy of dam failure risk prediction for the earth-rock dam to be monitored.

[0134] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A method for analyzing earth-rock dam failure under earthquake and flood, characterized in that: The following steps are involved: Step S101, dividing the edge of the earth-rock dam to be monitored into M areas to be monitored at the same distance intervals; Step S102, within a preset collection time period T, collect vibration signal data, hydrological data and dam body data of each area to be monitored at a preset time interval t, and generate a comprehensive feature matrix; The vibration signal data is represented by the acceleration corresponding to N time steps, where N represents the number of time steps, N=T / t; The comprehensive feature matrix is ​​expressed as: in They represent the comprehensive feature vectors of the first monitored area to the Mth monitored area at the first time step, They represent the comprehensive feature vectors from the 1st time step to the Nth time step of the 1st monitored area, Represents the comprehensive feature vector of the Nth time step of the Mth area to be monitored; in and They represent the vibration vector, hydrological vector and dam body vector of the Mth monitored area at the Nth time step respectively; Step S103, inputting the comprehensive characteristic matrix into the earth-rock dam risk prediction model, and the output value represents the dam break risk score of the earth-rock dam to be monitored; A larger dam failure risk score indicates a higher dam failure risk; The earth-rock dam risk prediction model is a nonlinear function model of vibration signal data, hydrological data, dam body data and dam break risk score; The dam break risk score is a discrete value between 0 and 100; Step S104, dividing the dam break risk score into three risk levels, and taking corresponding early warning measures respectively; The earth-rock dam risk prediction model includes: N hidden layers, each hidden layer includes: a first hidden unit, a second hidden unit and a third hidden unit; The first hidden unit includes H channels, where H is a custom parameter; Each channel of the first hidden unit of the i-th hidden layer inputs the comprehensive feature vector from the first to-be-monitored area to the M-th to-be-monitored area at the i-th time step in the comprehensive feature matrix, and each channel outputs the first updated feature; The second hidden unit of the i-th hidden layer inputs the first updated features output by the H channels of the first hidden unit of the i-th hidden layer, and outputs the second updated features; The third hidden unit of the i-th hidden layer inputs the second updated feature output by the second hidden unit of the i-th hidden layer, and outputs a third updated feature; The third hidden unit is constructed based on a gated recurrent unit; The third updated feature output by the third hidden unit of the Nth hidden layer is input into the classifier, and the classification space of the classifier represents the dam break risk score of the earth-rock dam to be monitored; Definition: 1≤i≤N; The calculation formula of the earth-rock dam risk prediction model includes: The calculation formula for the first hidden unit of the i-th hidden layer is as follows: Where 1≤j≤H, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, represents the comprehensive feature vector of the first monitored area to the Mth monitored area in the i-th time step in the comprehensive feature matrix of the j-th channel input of the first hidden unit of the i-th hidden layer, V i j , Q i j and K i j denote the first conversion feature, the second conversion feature, and the third conversion feature of the jth channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first conversion parameter, the second conversion parameter and the third conversion parameter of the j-th channel of the first hidden unit of the i-th hidden layer, respectively. and denote the first weight parameter and the second weight parameter of the first hidden unit of the i-th hidden layer, respectively, T denotes a transpose operation, tanh denotes a hyperbolic tangent activation function, and softmax denotes a softmax activation function; The calculation formula for the second hidden unit of the i-th hidden layer is as follows: Where Z i represents the second updated feature output by the second hidden unit of the i-th hidden layer, represents the first updated feature of the jth channel output of the first hidden unit of the i-th hidden layer, W represents the concatenation of the first updated features of the H channels output by the first hidden unit of the i-th hidden layer. i Represents the weight parameter of the second hidden unit of the i-th hidden layer.

2. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1 is characterized in that: M, T, and t are all custom parameters. The default value of M is set to 8, the default value of T is set to 60 seconds, and the default value of t is set to 2 seconds.

3. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1 is characterized in that: The vibration vector of the Nth time step of the Mth area to be monitored is expressed as: in They respectively represent the energy values ​​corresponding to the frequencies of 0.01 Hz, 0.02 Hz to 19.99 Hz and 20 Hz in the Nth time step of the Mth area to be monitored.

4. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 3 is characterized in that: Generating a vibration vector based on the vibration signal data includes the following steps: Step S201, performing pre-emphasis processing on the vibration signal data through a high-pass filter; Step S202, performing noise reduction processing on the vibration signal data after pre-emphasis processing by wavelet transform; Step S203: converting the vibration signal data after noise reduction processing into a vibration vector through fast Fourier transform.

5. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1, characterized in that: The hydrological vector of the Nth time step of the Mth monitored area is expressed as: in They respectively represent the water level, water level change rate, rainfall, rainfall change rate, seepage and seepage change rate of the Nth time step of the Mth monitored area; the water level change rate of the Nth time step is equal to the difference between the water level of the Nth time step and the water level of the N-1th time step divided by the preset time interval t; the rainfall change rate of the Nth time step is equal to the difference between the rainfall of the Nth time step and the rainfall of the N-1th time step divided by the preset time interval t; the seepage change rate of the Nth time step is equal to the difference between the infiltration rate of the Nth time step and the infiltration rate of the N-1th time step divided by the preset time interval t. The water level change rate, rainfall change rate and seepage change rate of the first time step are all set to 0.

6. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1, characterized in that: The dam body vector of the Nth time step in the Mth monitored area is expressed as: They respectively represent the pressure, displacement, number of cracks, sum of crack lengths, sum of crack widths, sum of crack depths and compressive strength of earth-rock dam materials in the Mth monitored area at the Nth time step; the compressive strength of earth-rock dam materials is a fixed value, which is obtained by researchers through experimental analysis during the design and construction of the earth-rock dam, and the unit is MPa.

7. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1, characterized in that: The training samples for training the earth-rock dam risk prediction model are obtained through finite element analysis software. The training samples include training data and training labels, including the following steps: Step S301, constructing a digital model of the earth-rock dam to be monitored at a reduced scale with real geometric shape and material properties by finite element analysis software, and using triangular facets to represent each area to be monitored of the digital model of the earth-rock dam; Step S302, adjusting the vibration signal data and hydrological data of each area to be monitored until the digital model of the earth-rock dam to be monitored breaks; Step S303, recording the vibration signal data, hydrological data and dam body data of each monitored area after each adjustment as training data of training samples; Step S304, obtaining the maximum principal stress and the maximum shear stress in all triangular facets corresponding to each monitored area after each adjustment through finite element analysis software, and recording the maximum resultant force value of each monitored area after each adjustment; The calculation formula of the maximum resultant force F is as follows: Where stress represents the maximum principal stress and shear represents the maximum shear stress; Step S305, repeating steps S302 to S304, taking the maximum resultant force value as a benchmark, normalizing the maximum resultant force value of each monitored area after each adjustment and mapping it to a dam break risk score, and using the dam break risk score as a training label for the training sample.

8. The method for analyzing earth-rock dam failure under earthquake and flood according to claim 1, characterized in that: The risk levels include: low risk level, medium risk level and high risk level. The dam break risk score of 0 to 30 corresponds to the low risk level, and daily monitoring and inspection measures are taken for the low risk level; the dam break risk score of 31 to 75 corresponds to the medium risk level, and measures such as increasing monitoring points and increasing monitoring frequency are taken for the medium risk level; the dam break risk score of 76 to 100 corresponds to the high risk level, and emergency warning and disposal measures are immediately initiated for the high risk level, including evacuating personnel, blocking traffic and notifying relevant departments.

Images