CN113627066B - Displacement prediction method for reservoir bank landslide - Google Patents
Displacement prediction method for reservoir bank landslide Download PDFInfo
- Publication number
- CN113627066B CN113627066B CN202110888118.2A CN202110888118A CN113627066B CN 113627066 B CN113627066 B CN 113627066B CN 202110888118 A CN202110888118 A CN 202110888118A CN 113627066 B CN113627066 B CN 113627066B
- Authority
- CN
- China
- Prior art keywords
- landslide
- displacement
- landslide displacement
- prediction
- item
- 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.)
- Active
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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A50/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE in human health protection, e.g. against extreme weather
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Pit Excavations, Shoring, Fill Or Stabilisation Of Slopes (AREA)
Abstract
The invention discloses a displacement prediction method of reservoir bank landslide, which comprises the steps of obtaining historical data of landslide, using the historical data as a training set, wherein the historical data is concrete water level of a reservoir, rainfall data of the location of the reservoir and displacement observation values of landslide, then establishing a landslide displacement prediction model, training the landslide displacement prediction model through the training set, then performing bank landslide displacement prediction based on the trained landslide displacement prediction model, decomposing an accumulated displacement value of landslide at the previous moment into a landslide displacement period item, a landslide displacement trend item and a landslide displacement random item when the landslide displacement prediction model performs prediction, decomposing an influence factor, inputting the landslide displacement period item, the landslide displacement trend item and the landslide displacement random item into the landslide displacement prediction model to obtain a landslide displacement trend item, a landslide displacement period item and a landslide displacement random item at the prediction moment, and the accumulated displacement of the landslide at the prediction moment is obtained by adding the displacement of the landslide and the displacement of the landslide is predicted.
Description
Technical Field
The invention belongs to the technical field of displacement prediction, and particularly relates to a displacement prediction method for a bank landslide of a reservoir.
Background
After the reservoir is built, the stability of the bank landslide is changed violently in a reservoir area due to long-term rainfall and reservoir water storage period scheduling, the deformation evolution of the bank landslide is a complex nonlinear process, and the landslide is easy to cause great harm to personal safety. Therefore, how to accurately predict the landslide displacement of the bank landslide of the reservoir is a technical problem to be solved by the technical personnel in the field.
Disclosure of Invention
The invention aims to accurately predict the landslide displacement of the bank landslide of a reservoir, and provides a displacement prediction method of the bank landslide of the reservoir.
The technical scheme of the invention is a displacement prediction method for bank landslide of a reservoir, which comprises the following steps:
s1, acquiring historical data of the landslide, and taking the historical data as a training set, wherein the historical data is specifically the water level of the reservoir, rainfall data of the reservoir and landslide displacement values of the landslide;
s2, establishing a landslide displacement prediction model, and training the landslide displacement prediction model through the training set;
and S3, predicting the displacement of the landslide based on the trained landslide displacement prediction model.
Further, the training of the landslide displacement prediction model by the training set in step S2 specifically includes the following sub-steps:
s21, grouping the training sets by taking a year as a period to obtain a plurality of groups of annual training data;
s22, determining the number of the preliminary landslide displacement prediction models according to the group number of the annual training data;
s23, numbering the preliminary landslide displacement prediction model from 1, wherein the numbering is a positive integer;
s24, determining the number of groups of the annual training data for training each preliminary landslide displacement prediction model according to the serial numbers, wherein the groups of the annual training data corresponding to each preliminary landslide displacement prediction model are not repeated;
s25, training each preliminary landslide displacement prediction model;
s26, establishing an optimal combination model based on the plurality of trained preliminary landslide displacement prediction models, and taking the optimal combination model as the landslide displacement prediction model.
Further, the step S25 specifically includes the following sub-steps:
s251, acquiring a landslide displacement value and an influence factor corresponding to the last moment of the prediction moment in the training set, wherein the influence factor is specifically a reservoir water level monthly average elevation, a monthly rainfall, an accumulated rainfall two months before the prediction moment, a single-month change amplitude, a single-month change rate and a double-month change amplitude;
s252, decomposing the accumulated displacement value of the landslide corresponding to the previous moment into a landslide displacement period item, a landslide displacement trend item and a landslide displacement random item through a variational modal decomposition unit in the preliminary landslide displacement prediction model;
s253, decomposing the influence factors into periodic item influence factors and random item influence factors through the variation modal decomposition unit;
s254, inputting the landslide displacement period item, the landslide displacement random item, the period item influence factor and the random item influence factor into a gating circulation unit in the preliminary landslide displacement prediction model, and obtaining a predicted landslide displacement period item and a predicted landslide displacement random item at the prediction moment through the gating circulation unit;
s255, inputting the landslide displacement trend item into a quadratic exponential smoothing method model in the preliminary landslide displacement prediction model and obtaining a predicted landslide displacement trend item at the prediction moment;
and S256, adding the landslide displacement period predicting item, the landslide displacement random item predicting item and the landslide displacement trend item predicting to obtain a landslide displacement value at the predicting moment.
Further, the step S256 specifically includes the following sub-steps:
s2561, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement period term prediction result through a judgment coefficient and a root mean square error, taking the preliminary landslide displacement prediction model as a first preliminary landslide displacement prediction model, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement random term prediction result through the judgment coefficient and the root mean square error, and taking the preliminary landslide displacement prediction model as a second preliminary landslide displacement prediction model;
s2562, combining the first preliminary landslide displacement prediction model and the second preliminary landslide displacement prediction model together to serve as the optimal combination model, and using the optimal combination model as the landslide displacement prediction model.
Further, the prediction training of the landslide displacement prediction model is sequential prediction, the sequential prediction is that only the current prediction time is predicted during each prediction, and when an actual monitoring displacement value exists at the current prediction time, the actual monitoring displacement value is incorporated into the training set to predict the next prediction time.
Further, the step S3 specifically includes the following sub-steps:
s31, obtaining a landslide accumulated displacement value and an influence factor corresponding to the previous time of the actual prediction time;
s32, decomposing the accumulated displacement value of the landslide corresponding to the last moment of the actual prediction moment into a landslide displacement period term, a landslide displacement trend term and a landslide displacement random term through a variation modal decomposition unit in the landslide displacement prediction model;
s33, predicting the landslide displacement period item at the actual prediction time through a first preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement period item corresponding to the previous time of the actual prediction time;
s34, predicting the landslide displacement random item at the actual prediction time through a second preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement random item corresponding to the previous time of the actual prediction time;
s35, predicting the landslide displacement trend item at the actual prediction time through a quadratic exponential smoothing method model in the landslide displacement prediction model and the landslide displacement trend item corresponding to the previous time of the actual prediction time;
and S36, adding the landslide displacement period term of the actual prediction time, the landslide displacement random term of the actual time and the landslide displacement trend term of the actual time, which are obtained through prediction, to obtain the displacement prediction value of the actual prediction time.
Compared with the prior art, the invention has the following beneficial effects:
(1) the method comprises the steps of obtaining historical data of the landslide, using the historical data as a training set, wherein the historical data is specifically water level of a reservoir, rainfall data of the location of the reservoir and landslide displacement value of the landslide, then establishing a landslide displacement prediction model, training the landslide displacement prediction model through the training set, then predicting bank-shore landslide displacement based on the trained landslide displacement prediction model, decomposing a landslide displacement value at the previous moment into a landslide displacement period item, a landslide displacement trend item and a landslide displacement random item when the landslide displacement prediction model predicts, decomposing influence factors, inputting the influence factors into the landslide displacement period item, the landslide displacement trend item and the landslide displacement random item to obtain a landslide displacement trend item, a landslide displacement period item and a landslide displacement random item at the predicted moment, and adding the landslide displacement trend item, the landslide displacement period item and the landslide displacement random item to obtain the landslide displacement at the predicted moment, the prediction of the landslide displacement is realized.
(2) The method comprises the steps of grouping a training set by taking years as a period to obtain a plurality of groups of annual training data, determining the number of preliminary landslide displacement prediction models according to the group number of the annual training data, numbering the preliminary landslide displacement prediction models, determining the group number of the annual training data of each preliminary landslide displacement prediction model according to the numbering, determining the preliminary landslide displacement prediction models with optimal prediction effects on landslide displacement period terms and landslide displacement random terms, and realizing the displacement prediction accuracy of the bank landslide.
Drawings
Fig. 1 is a schematic flow chart of a displacement prediction method for bank landslides of a reservoir provided by an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
As described in the background art, when a bank of a reservoir bank is landslide, great harm is easily caused to personal safety, and therefore, the displacement prediction method for the bank landslide of the reservoir bank is provided for predicting the landslide displacement of the bank landslide of the reservoir bank.
Fig. 1 is a schematic flow chart of a displacement prediction method for bank landslide of a reservoir bank according to an embodiment of the present application, where the method includes the following steps:
and S1, acquiring historical data of the landslide, and taking the historical data as a training set, wherein the historical data is specifically the water level of the reservoir, rainfall data of the reservoir and landslide displacement values of the landslide.
Specifically, the historical data can be acquired through a network database, and the historical data of the reservoir can also be acquired through an observation research station.
And S2, establishing a landslide displacement prediction model, and training the landslide displacement prediction model through the training set.
In this embodiment of the present application, the training of the landslide displacement prediction model through the training set in step S2 specifically includes the following sub-steps:
s21, grouping the training sets by taking a year as a period to obtain a plurality of groups of annual training data;
s22, determining the number of the preliminary landslide displacement prediction models according to the group number of the annual training data;
s23, numbering the preliminary landslide displacement prediction models from 1, wherein the numbering is a positive integer;
s24, determining the number of groups of the annual training data for training each preliminary landslide displacement prediction model according to the serial numbers, wherein the groups of the annual training data corresponding to each preliminary landslide displacement prediction model are not repeated;
s25, training each preliminary landslide displacement prediction model;
s26, establishing an optimal combination model based on the plurality of trained preliminary landslide displacement prediction models, and taking the optimal combination model as the landslide displacement prediction model.
Specifically, the landslide displacement of the reservoir bank is controlled by the geological conditions of the reservoir bank, such as landform, rock-soil body structure and the like, the landslide displacement is represented as an approximate single increasing function on a time line, the long-term deformation trend of the landslide is reflected, the landslide displacement is represented as periodic fluctuation under the action of periodic influence factors (rainfall, reservoir water level fluctuation and the like), in addition, the landslide is represented as a near-sound sequence under the action of random factors (seismic force, wind load and the like), and therefore, the time sequence addition model of the landslide displacement can be obtained as follows:
X t ＝T t +P t +R t
in the formula: x t A landslide displacement monitoring value at time T, T t Is a trend term of the landslide displacement at the time t, P t Periodic term of landslide displacement at time t, R t Is a random term of the landslide displacement at time t.
Due to the limitation of the current monitoring technology, data of random factors such as seismic force, wind load and the like cannot be acquired, so that when a landslide displacement random item is researched, only influences of rainfall and random components in reservoir water level on the landslide displacement random item are considered, and it is also noted that any time in the application refers to one month.
Regarding the training of the landslide displacement period item and the landslide displacement random item, the accuracy rate changes differently according to the size of the training set, the accuracy rate of the landslide displacement period item is gradually increased along with the increase of the training set, and the prediction result of the landslide displacement random item is increased first and then decreased along with the increase of the training set, so that different numbers of training sets are required to be adopted for training the landslide displacement period item and the landslide displacement random item.
In the embodiment of the present application, the step S25 specifically includes the following sub steps:
s251, acquiring a landslide displacement value and an influence factor corresponding to the last moment of the prediction moment in the training set, wherein the influence factor is specifically a reservoir water level month average elevation, a month rainfall, a two-month accumulated rainfall before the prediction moment, a single-month change amplitude, a single-month change rate and a two-month change amplitude;
s252, decomposing the accumulated displacement value of the landslide corresponding to the previous moment into a landslide displacement period item, a landslide displacement trend item and a landslide displacement random item through a variational modal decomposition unit in the preliminary landslide displacement prediction model;
s253, decomposing the influence factors into periodic item influence factors and random item influence factors through the variation modal decomposition unit;
s254, inputting the landslide displacement period item, the landslide displacement random item, the period item influence factor and the random item influence factor into a gating circulation unit in the preliminary landslide displacement prediction model, and obtaining a predicted landslide displacement period item and a predicted landslide displacement random item at the prediction moment through the gating circulation unit;
s255, inputting the landslide displacement trend item into a quadratic exponential smoothing method model in the preliminary landslide displacement prediction model and obtaining a predicted landslide displacement trend item at the prediction moment;
and S256, adding the landslide displacement period predicting item, the landslide displacement random item predicting item and the landslide displacement trend item predicting to obtain a landslide displacement value at the predicting moment.
Variational Modal Decomposition (VMD) is an adaptive, fully non-recursive method of modal variational and signal processing. The method decomposes a real-valued input signal into a plurality of Intrinsic Mode Function (IMF) components with specific sparsity, and has the advantage of determining the number of modal components in advance. The method solves the problems of end effect and modal component aliasing existing in an Empirical Mode Decomposition (EMD) method, can reduce the non-stationarity of high-complexity and strong nonlinear time sequence data, obtains a subsequence with specific sparse characteristics, and is particularly suitable for decomposition of a landslide displacement time sequence.
The Intrinsic Mode Function (IMF) is an AM-FM signal u k (t) is as follows:
u k (t)＝A k (t)cos(φ k (t))
in the formula: phase phi k (t) is notA decreasing function of the instantaneous amplitude A k (t) are all non-negative.
Assuming that the landslide displacement time sequence or the reference influence factor is decomposed into K modal components, and ensuring that the modal components have the center frequency with limited bandwidth and the sum of the estimated bandwidths of all the modes is minimum, and the sum of all the modes is equal to the constraint of the original signal, then the constraint variation is carried out
The expression can be written as:
In the formula: k is the number of the required modal components, and K is an integer between 1 and K; { u k }＝{u 1 ,…,u K The modal component obtained by final decomposition is obtained; { omega [ [ omega ] ] k }＝{ω 1 ,…,ω K The actual center frequency of each modal component;
is a partial derivative symbol; δ (t) is a dirac function; is the convolution operator.
In order to solve the above formula, Lagrange multiplier lambda is introduced, the constraint variation problem is converted into the non-constraint variation problem, and the augmented Lagrange expression is obtained as follows:
in the formula: alpha is a secondary penalty factor, and the purpose is to reduce the interference of Gaussian noise.
At the moment, each modal component and the center frequency can be obtained by utilizing an alternating direction multiplier iterative algorithm for optimization, and saddle points of an unconstrained model are searched for and solved to obtain the optimal solution of a constrained model, so that K modal components are obtained. The decomposition of the landslide displacement and the influence factor is realized through the VMD, and the modal component K of the landslide displacement time sequence is 3 according to the time sequence addition model of the landslide displacement; the modal component K of the influence factor timing is 2 (the influence factor mainly affects the periodic term displacement of the landslide, and the random information thereof contributes to the random displacement). The value of the secondary penalty α needs to be discussed further to obtain a modal component of a suitable sparse feature.
The size of the Shannon entropy value may reflect the uniformity of the probability distribution, which is defined as shown below. For the periodic term component of the landslide displacement, the probability distribution is relatively uniform, and the sparsity is strong. Therefore, Shannon entropy is used as a fitness function, and the quadratic penalty alpha of the VMD decomposition parameter is optimized through Grey Wolf Optimization (GWO) to obtain landslide period item displacement with uniform probability distribution and strong sparsity, so that a trend item and a random item of landslide displacement can be obtained.
In the formula: h (p) is Shannon entropy value; p is a radical of i Is an uncertain probability distribution.
The goal of a Gated Recurrent Unit (GRU) recurrent neural network is to have each recurrent unit adaptively capture the dependencies of different time scales. The GRU regulates information flow inside the unit through the gate control unit, a storage unit is not arranged independently, the structure is simpler, and the training time and the updating optimization are more efficient than other recurrent neural networks. Therefore, the GRU model is adopted herein to carry out the sequential prediction on the landslide position, and is used for processing the periodic term displacement and the random term displacement of the non-linear problem.
In the application of displacement prediction, a landslide displacement time sequence and an influence factor time sequence are used as input, combined with historical information, the landslide displacement time sequence and the influence factor time sequence jointly enter an updating gate and a resetting gate of a GRU unit network, the resetting gate obtains a candidate state through tanh function calculation, and combined with the updating gate, an output state is reserved. Wherein the update gate is used to control the degree of ignoring of retained history information and the reset gate is used to decide the degree of ignoring of history information by the candidate state. If it is at the present timeThe time sequence input of the landslide displacement of the moment t is x t Then, the one-time forward calculation of the GRU unit network is:
z t ＝σ(W z ·[h t-1 ,x t ]+b z )
r t ＝σ(W r ·[h t-1 ,x t ]+b r )
in the formula: σ is a summation symbol; w z 、W r 、
Respectively an update gate z t Reset gate r t And candidate states
A weight matrix of (a); and b z 、b r 、
A bias matrix of three, respectively, in which the hidden vector h t-1 Represents the number of hidden neurons input, while gate z is updated t And candidate states
Output h for jointly determining current time of network t And obtaining the displacement value at the next moment of the landslide:
therefore, in a specific application scenario, the VMD modal component of the landslide displacement is known by taking the landslide accumulated displacement value at the above time as an input; the landslide displacement period term at the previous moment is extracted by optimizing the Shannon entropy value of the second modal component through an GWO model, namely a variation modal decomposition unit, and meanwhile, a landslide displacement trend term (first modal component) at the previous moment and a landslide displacement random term (third modal component) at the previous moment are obtained through decomposition.
The influence factor at the previous moment is decomposed by VMD to obtain the periodic term and the random term component of the influence factor, and the periodic term and the random term component are respectively used as the influence factors of the periodic term and the random term of the landslide displacement prediction model to be input. Optimization of Shannon entropy value of low frequency component (period term) by GWO to obtain optimum punishment parameter α And the fidelity of the modal component is ensured. And setting the modal number K of the VMD to be 2, wherein 2 components sequentially correspond to the time sequence of a periodic item and a random item of the influence factor, and the rest of the components are decomposed with the landslide displacement.
In this embodiment, the step S256 specifically includes the following sub-steps:
s2561, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement period term prediction result through a judgment coefficient and a root mean square error, taking the preliminary landslide displacement prediction model as a first preliminary landslide displacement prediction model, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement random term prediction result through the judgment coefficient and the root mean square error, and taking the preliminary landslide displacement prediction model as a second preliminary landslide displacement prediction model;
s2562, combining the first preliminary landslide displacement prediction model and the second preliminary landslide displacement prediction model together to serve as the optimal combination model, and taking the optimal combination model as the landslide displacement prediction model.
Specifically, a determination coefficient (R) is used 2 ) And Root Mean Square Error (RMSE) as an indicator for evaluating the performance of the predictive model. Therefore, the performance of the prediction model is analyzed based on the indexes, which are specifically defined as follows:
in the formula: x is the number of t As an actual observed value of the displacement of the landslide,
in order to predict the value of the target,
is the average of all observations, t represents the time, and N is the number of samples.
In addition, in the embodiment of the present application, the prediction training performed by the landslide displacement prediction model is sequential prediction, where the sequential prediction is performed only on the current prediction time at each prediction, and when an actual monitored displacement value exists at the current prediction time, the actual monitored displacement value is incorporated into the training set to perform prediction on the next prediction time.
And step S3, predicting the displacement of the landslide based on the trained landslide displacement prediction model.
In this embodiment, the step S3 specifically includes the following sub-steps:
s31, obtaining a landslide accumulated displacement value and an influence factor corresponding to the last moment of the actual prediction moment;
s32, decomposing the accumulated displacement value of the landslide corresponding to the last moment of the actual prediction moment into a landslide displacement period term, a landslide displacement trend term and a landslide displacement random term through a variation modal decomposition unit in the landslide displacement prediction model;
s33, predicting the landslide displacement period item at the actual prediction time through a first preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement period item corresponding to the previous time of the actual prediction time;
s34, predicting the landslide displacement random item at the actual prediction time through a second preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement random item corresponding to the previous time of the actual prediction time;
s35, predicting the landslide displacement trend item at the actual prediction time through a quadratic exponential smoothing method model in the landslide displacement prediction model and the landslide displacement trend item corresponding to the previous time of the actual prediction time;
and S36, adding the landslide displacement period term of the actual prediction time, the landslide displacement random term of the actual time and the landslide displacement trend term of the actual time, which are obtained through prediction, to obtain the displacement prediction value of the actual prediction time.
Specifically, the landslide displacement prediction model after training is a combined model substantially and comprises three parts, namely a first preliminary landslide displacement prediction model, a second preliminary landslide displacement prediction model and a quadratic index smoothing method model, the displacement prediction of landslide is more accurate through the technical scheme, the displacement prediction effects of the first preliminary landslide displacement prediction model and the second preliminary landslide displacement prediction model are in direct proportion to the number of neurons, the number of the neurons is 100, the best displacement prediction effect is achieved, and if the displacement prediction effects are continuously increased, the precision is reduced, the consumed time is increased, and the displacement prediction can be obtained through experiments.
The quadratic exponential smoothing method in the quadratic exponential smoothing method model is actually a special weighted moving average method, and is characterized in that the weight of the latest data is higher than that of the early data, and the factor of the weight decreases exponentially along with the aging of the data, so that the quadratic exponential smoothing method is more suitable for the prediction of time series with certain trends. The parameter set data smoothing factor was 0.99 and the trend smoothing factor was 0.98.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (4)
1. A displacement prediction method for reservoir bank landslide is characterized by comprising the following steps:
s1, acquiring historical data of the landslide, and taking the historical data as a training set, wherein the historical data is specifically the water level of the reservoir, rainfall data of the reservoir and landslide displacement values of the landslide;
s2, establishing a landslide displacement prediction model, and training the landslide displacement prediction model through the training set;
s3, predicting the displacement of the landslide based on the trained landslide displacement prediction model;
in step S2, the training of the landslide displacement prediction model by the training set specifically includes the following sub-steps:
s21, grouping the training sets by taking a year as a period to obtain a plurality of groups of annual training data;
s22, determining the number of the preliminary landslide displacement prediction models according to the group number of the annual training data;
s23, numbering the preliminary landslide displacement prediction model from 1, wherein the numbering is a positive integer;
s24, determining the number of groups of the annual training data for training each preliminary landslide displacement prediction model according to the serial numbers, wherein the groups of the annual training data corresponding to each preliminary landslide displacement prediction model are not repeated;
s25, training each preliminary landslide displacement prediction model;
establishing an optimal combination model based on the plurality of trained preliminary landslide displacement prediction models, and taking the optimal combination model as the landslide displacement prediction model;
the step S25 specifically includes the following sub-steps:
s251, acquiring a landslide displacement value and an influence factor corresponding to the last moment of the prediction moment in the training set, wherein the influence factor is specifically a reservoir water level month average elevation, a month rainfall, a two-month accumulated rainfall before the prediction moment, a single-month change amplitude, a single-month change rate and a two-month change amplitude;
s252, decomposing the accumulated displacement value of the landslide corresponding to the previous moment into a landslide displacement period item, a landslide displacement trend item and a landslide displacement random item through a variational modal decomposition unit in the preliminary landslide displacement prediction model;
s253, decomposing the influence factors into periodic item influence factors and random item influence factors through the variation modal decomposition unit;
s254, inputting the landslide displacement period item, the landslide displacement random item, the period item influence factor and the random item influence factor into a gating circulation unit in the preliminary landslide displacement prediction model, and obtaining a predicted landslide displacement period item and a predicted landslide displacement random item at the prediction moment through the gating circulation unit;
s255, inputting the landslide displacement trend item into a quadratic exponential smoothing method model in the preliminary landslide displacement prediction model and obtaining a predicted landslide displacement trend item at the prediction moment;
and S256, adding the landslide displacement period predicting item, the landslide displacement random item predicting item and the landslide displacement trend item predicting to obtain a displacement value predicting at the predicting moment.
2. The method for predicting the displacement of the bank landslide of the reservoir of claim 1, wherein the step S256 further comprises the following sub-steps:
s2561, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement period term prediction result through a judgment coefficient and a root mean square error, taking the preliminary landslide displacement prediction model as a first preliminary landslide displacement prediction model, determining a preliminary landslide displacement prediction model which is optimal to the landslide displacement random term prediction result through the judgment coefficient and the root mean square error, and taking the preliminary landslide displacement prediction model as a second preliminary landslide displacement prediction model;
s2562, combining the first preliminary landslide displacement prediction model and the second preliminary landslide displacement prediction model together to serve as the optimal combination model, and using the optimal combination model as the landslide displacement prediction model.
3. The method for predicting the displacement of the landslide of the reservoir bank according to claim 1, wherein the prediction training of the landslide displacement prediction model is sequential prediction, the sequential prediction is that only the current prediction time is predicted at each prediction, and when an actual monitoring displacement value exists at the current prediction time, the actual monitoring displacement value is incorporated into the training set to predict the next prediction time.
4. The method for predicting the displacement of the bank landslide of the reservoir of claim 2, wherein the step S3 specifically comprises the following substeps:
s31, obtaining a landslide accumulated displacement value and an influence factor corresponding to the last moment of the actual prediction moment;
s32, decomposing the accumulated displacement value of the landslide corresponding to the last moment of the actual prediction moment into a landslide displacement period term, a landslide displacement trend term and a landslide displacement random term through a variation modal decomposition unit in the landslide displacement prediction model;
s33, predicting the landslide displacement period item at the actual prediction time through a first preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement period item corresponding to the previous time of the actual prediction time;
s34, predicting the landslide displacement random item at the actual prediction time through a second preliminary landslide displacement prediction model in the landslide displacement prediction model, an influence factor and the landslide displacement random item corresponding to the previous time of the actual prediction time;
s35, predicting the landslide displacement trend item at the actual prediction time through a quadratic exponential smoothing method model in the landslide displacement prediction model and the landslide displacement trend item corresponding to the previous time of the actual prediction time;
and S36, adding the landslide displacement period term of the actual prediction time, the landslide displacement random term of the actual time and the landslide displacement trend term of the actual time, which are obtained through prediction, to obtain the displacement prediction value of the actual prediction time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110888118.2A CN113627066B (en) | 2021-08-03 | 2021-08-03 | Displacement prediction method for reservoir bank landslide |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110888118.2A CN113627066B (en) | 2021-08-03 | 2021-08-03 | Displacement prediction method for reservoir bank landslide |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113627066A CN113627066A (en) | 2021-11-09 |
| CN113627066B true CN113627066B (en) | 2022-08-16 |
Family
ID=78382480
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110888118.2A Active CN113627066B (en) | 2021-08-03 | 2021-08-03 | Displacement prediction method for reservoir bank landslide |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113627066B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114201923A (en) * | 2022-02-16 | 2022-03-18 | 中国地质调查局水文地质环境地质调查中心 | Processing method and device of landslide displacement prediction model |
| CN114897257B (en) * | 2022-05-31 | 2024-06-14 | 三峡大学 | Displacement step type reservoir landslide accumulated displacement prediction method based on EEMD-GRU model optimal time scale |
| CN114741976B (en) * | 2022-06-13 | 2022-09-23 | 西南交通大学 | Displacement prediction method, device and equipment and readable storage medium |
| CN115130658B (en) * | 2022-06-27 | 2023-09-22 | 武汉大学 | Reservoir landslide displacement prediction method and device based on nonlinear combination model |
| CN115512531B (en) * | 2022-09-28 | 2023-06-23 | 重庆地质矿产研究院 | Landslide hazard multi-monitoring-point fusion early warning method based on deformation order |
| CN116227365B (en) * | 2023-05-06 | 2023-07-07 | 成都理工大学 | Landslide displacement prediction method based on improved VMD-TCN |
| CN116611323B (en) * | 2023-05-10 | 2024-04-16 | 中国地质大学（武汉） | Landslide evolution modal change intelligent judgment method and system |
| CN118070049A (en) * | 2024-04-22 | 2024-05-24 | 成都理工大学 | Step type landslide displacement prediction method and system under mechanism guidance |
| CN118761331A (en) * | 2024-09-06 | 2024-10-11 | 青岛理工大学 | Landslide displacement prediction method and system based on displacement vector angle and LSTM |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112270400A (en) * | 2020-10-16 | 2021-01-26 | 西安工程大学 | Landslide displacement dynamic prediction method based on multiple influence factors |
| US20210049515A1 (en) * | 2019-08-16 | 2021-02-18 | China Institute Of Water Resources And Hydropower Research | Prediction method and system of high slope deformation |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4219100B2 (en) * | 2001-05-18 | 2009-02-04 | 財団法人鉄道総合技術研究所 | Prediction device for ground displacement due to slope failure such as landslide and surface layer failure |
| CN109885977A (en) * | 2019-03-25 | 2019-06-14 | 成都理工大学 | A kind of bank slope Deformation Prediction method |
| CN110378070A (en) * | 2019-08-05 | 2019-10-25 | 成都理工大学 | Based on PSO-SVR and the united Prediction of Displacement in Landslide method of DES |
-
2021
- 2021-08-03 CN CN202110888118.2A patent/CN113627066B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210049515A1 (en) * | 2019-08-16 | 2021-02-18 | China Institute Of Water Resources And Hydropower Research | Prediction method and system of high slope deformation |
| CN112270400A (en) * | 2020-10-16 | 2021-01-26 | 西安工程大学 | Landslide displacement dynamic prediction method based on multiple influence factors |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113627066A (en) | 2021-11-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113627066B (en) | Displacement prediction method for reservoir bank landslide | |
| Liang et al. | A novel wind speed prediction strategy based on Bi-LSTM, MOOFADA and transfer learning for centralized control centers | |
| CN109886498B (en) | EMD-GRU short-term power load prediction method based on feature selection | |
| CN110730046B (en) | Cross-frequency-band spectrum prediction method based on deep migration learning | |
| CN112733997B (en) | Hydrological time series prediction optimization method based on WOA-LSTM-MC | |
| CN110969290A (en) | Runoff probability prediction method and system based on deep learning | |
| US20220114455A1 (en) | Pruning and/or quantizing machine learning predictors | |
| CN110059867B (en) | Wind speed prediction method combining SWLSTM and GPR | |
| CN116316591A (en) | Short-term photovoltaic power prediction method and system based on hybrid bidirectional gating cycle | |
| CN113836823A (en) | Load combination prediction method based on load decomposition and optimized bidirectional long-short term memory network | |
| CN106022517A (en) | Risk prediction method and device based on nucleus limit learning machine | |
| CN104732091A (en) | Cellular automaton river bed evolution prediction method based on natural selection ant colony algorithm | |
| CN111598329A (en) | Time sequence data prediction method based on automatic parameter adjustment recurrent neural network | |
| CN115587666A (en) | Load prediction method and system based on seasonal trend decomposition and hybrid neural network | |
| Yang et al. | An ensemble prediction system based on artificial neural networks and deep learning methods for deterministic and probabilistic carbon price forecasting | |
| CN115392441A (en) | Method, apparatus, device and medium for on-chip adaptation of quantized neural network model | |
| CN115310355A (en) | Multi-energy coupling-considered multi-load prediction method and system for comprehensive energy system | |
| CN107704944B (en) | Construction method of stock market fluctuation interval prediction model based on information theory learning | |
| CN116701875A (en) | Audible noise probability prediction method and system for extra-high voltage alternating current transmission line | |
| CN114239418B (en) | Landslide displacement prediction method based on combination of multiple algorithms | |
| CN116660756A (en) | Battery capacity attenuation curve generation method based on condition generation countermeasure network | |
| CN115829123A (en) | Natural gas demand prediction method and device based on grey model and neural network | |
| CN115509752A (en) | Edge prediction method based on deep autoregressive recurrent neural network | |
| CN113095215A (en) | Solar radio filtering method and system based on improved LSTM network | |
| CN117172085B (en) | PCCP broken wire prediction method, device, computer equipment and medium |
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 |