Method and system for predicting deformation of excavation and anchoring of high rock slopes based on machine learning

Abstract

The present invention discloses a method and system for predicting deformation of rock high slope excavation and anchoring based on machine learning, including: obtaining geological survey reports of high arch dam area, slope excavation progress and monitoring information, and constructing a multi-stage excavation and anchoring model for rock high slope; ranking the importance of rock mass physical and mechanical parameters of the constructed model based on a machine learning algorithm, and identifying key rock mass physical and mechanical parameters as random variables to be updated; constructing a proxy model between random variables and slope responses in each excavation and anchoring stage based on a machine learning algorithm; constructing a Bayesian framework, incorporating monitoring information into the posterior distribution of rock mass physical and mechanical parameters in the inversion; inputting the inverted posterior distribution of rock mass physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict slope deformation in subsequent excavation and anchoring stages. The present invention realizes the prediction of excavation and anchoring deformation of rock high slopes by using monitoring information to make up for the lack of rock mass physical and mechanical parameter test data.

Description

Technical Field

[0001] The present invention belongs to the technical field of slope deformation prediction, and in particular relates to a method and system for predicting the excavation and anchoring deformation of a high rock slope based on machine learning. Background Art

[0002] In recent years, high arch dams with good economic performance and safety have been widely favored by engineers, and a large number of high arch dams have been built and put into operation in the southwest region, such as Jinping I Hydropower Station, Xiluodu Hydropower Station, Baihetan Hydropower Station and Wudongde Hydropower Station. Due to the huge scale of excavation during the construction of high arch dams, the excavation deformation and stability problems of high rock slopes in the dam area are particularly prominent, which is one of the main engineering and technical problems in hydropower construction.

[0003] Generally speaking, engineering usually predicts slope deformation based on finite element models. However, since slope rock is a natural material with the characteristics of concealment and spatiotemporal variability, and is affected by limited test data, the rock mass physical and mechanical parameters input into the finite element model to calculate slope deformation cannot represent the actual rock mass physical and mechanical properties, resulting in significant errors in predicting slope deformation based only on the finite element model. In recent years, with the development of monitoring technology, a large number of monitoring instruments are usually installed during the construction of high arch dams to obtain real-time monitoring data such as slope displacement, stress, and seepage pressure. These data reflect the real state of the slope in real time and are more reliable than the rock mass physical and mechanical parameters obtained based on survey data and experiments. How to use monitoring information to invert the rock mass physical and mechanical parameters and combine finite element models with machine learning algorithms to predict slope deformation in real time needs further development. Summary of the invention

[0004] The purpose of the present invention is to address the deficiencies in the prior art and provide a method for predicting the deformation of high rock slope excavation anchoring based on machine learning. The method can use monitoring information to make up for the lack of test data on the physical and mechanical parameters of the rock mass to predict the deformation of high rock slope excavation anchoring.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A method for predicting deformation of rock high slope excavation anchoring based on machine learning, characterized in that it comprises the following steps:

[0007] Step 1: Obtain the geological survey report of the high arch dam area, the slope excavation progress and monitoring information, and use them to build a multi-stage excavation and anchoring model for rock high slopes;

[0008] Step 2: sort the importance of the rock mass physical and mechanical parameters used to construct the model based on the machine learning algorithm, and identify the key rock mass physical and mechanical parameters as random variables to be updated;

[0009] Step 3, constructing a proxy model between random variables and slope response in each excavation and anchoring stage based on a machine learning algorithm;

[0010] Step 4: Establish a functional relationship between the monitoring information and the proxy model, construct a Bayesian framework, and incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass;

[0011] Step 5: Input the inverted posterior distribution of the rock mass physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage.

[0012] Furthermore, the method of step 1 for constructing a multi-stage excavation and anchoring model for a high rock slope is as follows: based on the geological survey report of the dam area, the progress of slope excavation and monitoring information, a mesh of a finite element model is generated in ANSYS, and the mesh is imported into FLAC3D through a plug-in to construct a multi-stage excavation and anchoring model for a high rock slope.

[0013] Furthermore, for the multi-stage excavation and anchoring model of high rock slope, the Mohr-Coulomb constitutive model is adopted, normal constraints are imposed on the four sides of the model, and the bottom side is fixed; different groups are used to distinguish the various types of rock mass and structural surface materials in the model, and the null model is used to simulate the excavation material and the cable model is used to simulate the prestressed anchor cable.

[0014] Furthermore, the implementation method of step 2 is:

[0015] Generate a rock mass physical and mechanical parameter data set based on Latin hypercube sampling, generate a corresponding command stream based on the rock mass physical and mechanical parameter data set, and input the command stream into the FLAC3D model to calculate the corresponding slope response data set;

[0016] The random forest algorithm is used for analysis. The rock mass physical and mechanical parameter dataset is defined as input, and the slope response dataset is defined as output. The rock mass physical and mechanical parameters are ranked according to their importance based on the importance measurement method in the random forest algorithm, and the rock mass physical and mechanical parameters that have a key impact on the slope response are identified as random variables to be updated.

[0017] Furthermore, the physical and mechanical parameters of the rock mass include elastic modulus E, Poisson's ratio μ, gravity γ, cohesion c and internal friction angle .

[0018] Further, step 3 includes:

[0019] The key physical and mechanical parameters of the rock mass identified in step 2 are defined as random variables x = [x 1 ,x 2 ,…,x d]; the support vector machine algorithm is used to train the proxy model SVM(x) between the random variables and the slope response in each excavation and anchoring stage, and the proxy model SVM(x) is used to replace the FLAC3D model.

[0020] Furthermore, in step 4, the monitoring information Y established is 1 ,y 2 ,…,y n ] and the surrogate model SVM(x) is as follows:

[0021] Y=SVM(x)+ε

[0022] Where ε represents the measurement error.

[0023] Furthermore, based on Bayesian theory, the functional relationship of monitoring information is incorporated into the following formula, and the adaptive differential evolution Metropolis algorithm DREAM with multi-chain parallel computing is used. (ZS) To invert the posterior distribution of the physical and mechanical parameters of the rock mass:

[0024] p(x|Y)=cL(x|Y)p(x)

[0025]

[0026]

[0027] Where p(x|Y) represents the posterior probability density function of the parameter; p(x) represents the prior probability density function of the parameter, and the parameters obey the log-normal distribution and are independent of each other; c is the normalization constant; L(x|Y) represents the likelihood function; subscript j represents the jth rock mass physical and mechanical parameter; μ N and σ N represents the logarithmic mean and standard deviation of the parameter; the subscript n represents the nth excavation and anchoring stage; μ ε and σ ε represents the mean and standard deviation of the measurement error ε.

[0028] Furthermore, step 5 specifically includes:

[0029] The inverted posterior distribution of rock mass physical and mechanical parameters is input into the proxy model SVM(x) of each excavation stage to predict the slope deformation in the subsequent excavation and anchoring stage. According to the calculation results, the mean value is taken as the predicted value of slope deformation in each subsequent excavation and anchoring stage and compared with the actual monitoring value of the slope.

[0030] Another object of the present invention is to provide a system according to the above-mentioned method for predicting deformation of rock high slope excavation anchoring based on machine learning, comprising:

[0031] Data acquisition module, used to obtain geological survey reports of high arch dam area, slope excavation progress and monitoring information;

[0032] A model building module is used to build a multi-stage excavation and anchoring model for a high rock slope based on the data acquired by the data acquisition module;

[0033] A random variable screening module is used to sort the importance of the rock mass physical and mechanical parameters used to construct the model based on a machine learning algorithm, and identify key rock mass physical and mechanical parameters as random variables to be updated;

[0034] A proxy model training module is used to train the proxy model between random variables and slope responses in each excavation and anchoring stage based on a machine learning algorithm;

[0035] The module for obtaining the posterior distribution of physical and mechanical parameters is used to establish the functional relationship between the monitoring information and the proxy model, and to construct a Bayesian framework to incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass;

[0036] The slope deformation prediction module is used to input the posterior distribution of the rock mass physical and mechanical parameters inverted by the posterior distribution acquisition module of the physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method for predicting deformation of excavation and anchoring of rocky high slopes based on machine learning. Under the premise of considering the uncertainty of the physical and mechanical parameters of the rock mass of the rocky high slopes, the Bayesian method is combined with actual engineering monitoring information to gradually invert the posterior distribution of the physical and mechanical parameters of the rock mass and predict the slope deformation in the subsequent excavation and anchoring stage, thereby solving the problem that it is difficult to reliably invert the physical and mechanical parameters of the rock mass based on limited survey data and test data. The method of the present invention ranks the importance of the physical and mechanical parameters of the rock mass input into the FLAC3D model based on a machine learning algorithm, identifies the key physical and mechanical parameters of the rock mass as random variables to be updated to reduce the number of inversion parameters, and at the same time, the method trains a proxy model to reduce the number of calls to the complex model, thereby greatly improving the computational efficiency of the Bayesian update. The present invention provides a basis for realizing a method for predicting deformation of excavation and anchoring of rocky high slopes based on machine learning, and has good engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of a method for predicting deformation of rock high slope excavation anchoring based on machine learning according to an embodiment of the present invention;

[0039] Figure 2 This is a diagram of a multi-stage excavation and anchoring model for a high rock slope according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention will be described clearly and completely below in combination with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0042] The present invention will be further described below in conjunction with specific embodiments, but the present invention is not limited thereto.

[0043] like Figure 1 As shown, an embodiment of the present invention provides a method for predicting deformation of rock high slope excavation anchoring based on machine learning, comprising the following steps:

[0044] Step 1: Obtain the geological survey report of the high arch dam area, the slope excavation progress and monitoring information, and use them to build a multi-stage excavation and anchoring model for rock high slopes;

[0045] According to the actual engineering situation, the geological survey report of the high arch dam area, the slope excavation progress and monitoring information are obtained, and the mesh of the finite element model is generated in ANSYS. It is then imported into FLAC3D through the plug-in to construct a multi-stage excavation and anchoring model for rock high slopes, such as Figure 2 As shown in the figure. The multi-stage excavation and anchoring model of high rock slope adopts the Mohr-Coulomb constitutive model, with normal constraints applied around the model and the bottom side fixed. The simulated rock mass and structural surface materials in the multi-stage excavation and anchoring model of high rock slope are distinguished by different groups, and the null model is used to simulate the excavation material and the cable model is used to simulate the prestressed anchor cable.

[0046] Step 2: sort the importance of the rock mass physical and mechanical parameters used to construct the model based on the machine learning algorithm, and identify the key rock mass physical and mechanical parameters as random variables to be updated;

[0047] According to the geological survey report of the dam area, engineering experience and literature, the physical and mechanical parameters of the rock mass are determined, including elastic modulus E, Poisson's ratio μ, gravity γ, cohesion c and internal friction angle. , the corresponding command stream is generated based on the rock mass physical and mechanical parameter data set, and the command stream is input into the FLAC3D model to calculate the corresponding slope response data set; the random forest algorithm is used for analysis, and the rock mass physical and mechanical parameter data set is defined as input, and the slope response data set is defined as output; the rock mass physical and mechanical parameters are ranked according to their importance based on the importance measurement method in the random forest algorithm, and the rock mass physical and mechanical parameters that have a key impact on the slope response are identified as random variables to be updated.

[0048] In this embodiment, the physical and mechanical parameters obtained are specifically shown in Table 1, wherein the elastic modulus E and cohesion c with high variability are considered as uncertain parameters, and the input set of the random forest algorithm is constructed by Latin hypercube sampling. The importance of the rock mass physical and mechanical parameters is ranked based on the importance measurement method in the random forest algorithm. The rock mass physical and mechanical parameters that have a key impact on the displacement of the slope monitoring point are identified as random variables to be updated, and the prior distribution of the selected random variables is shown in Table 2.

[0049] Table 1 Physical and mechanical parameters of slope rock mass

[0050]

[0051]

[0052] Table 2 Prior distribution table of random variables to be updated

[0053]

[0054] Step 3, constructing a proxy model between random variables and slope response in each excavation and anchoring stage based on a machine learning algorithm;

[0055] According to the order of parameter importance, the five identified random variables to be updated x = [x 1 ,x 2 ,…,x d ] is constructed as the input set, and the corresponding FLAC3D model calculation results are used as the output set. The support vector machine algorithm is used to construct the proxy model between the random variables and the displacement of the slope monitoring points in each excavation and anchoring stage. The input set and the output set are constructed as a data set, and the data set is divided into a training set and a prediction set. The training set is used to train the proxy model and the prediction set is used to verify each proxy model. The Bayesian update method based on the response of geotechnical structures usually includes two steps, namely, obtaining the posterior distribution of uncertain model parameters and updating the structural response based on the posterior distribution. In this process, a large number of numerical model simulations need to be performed. Therefore, the proxy model is used instead of the FLAC3D model in the subsequent Bayesian update to achieve high computational efficiency while ensuring the calculation accuracy.

[0056] Step 4: Establish a functional relationship between the monitoring information and the proxy model, construct a Bayesian framework, and incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass;

[0057] The monitoring information Y established is 1 ,y 2 ,…,y n ] and the surrogate model SVM(x) is as follows:

[0058] Y=SVM(x)+ε

[0059] Where ε represents the measurement error;

[0060] Among them, the measurement error ε obeys a standard normal distribution with a mean of 1 mm and a variance of 0. Based on the Bayesian theory p(x|Y)=cL(x|Y)p(x), combined with prior knowledge and actual engineering monitoring information, the parameter prior probability density function p(x) and the likelihood function L(x|Y) are constructed to calculate the parameter posterior probability density function p(x|Y). Since the random variable x is high-dimensional and nonlinear, the posterior distribution p(x|Y) usually does not have an analytical solution. This embodiment uses the adaptive differential evolution Metropolis algorithm DREAM with multi-chain parallel computing. (ZS) Invert the posterior distribution of the random variable x. Specifically, the adaptive differential evolution Metropolis algorithm DREAM with multi-chain parallel computing is used. (ZS) The posterior distribution of the inverted rock mass physical and mechanical parameters is:

[0061] p(x|Y)=cL(x|Y)p(x)

[0062]

[0063]

[0064] Where p(x|Y) represents the posterior probability density function of the parameter; p(x) represents the prior probability density function of the parameter, and the parameters obey the log-normal distribution and are independent of each other; c is the normalization constant; L(x|Y) represents the likelihood function; subscript j represents the jth rock mass physical and mechanical parameter; μ N and σ N represents the logarithmic mean and standard deviation of the parameter; the subscript n represents the nth excavation and anchoring stage; μ ε and σ ε represents the mean and standard deviation of the measurement error ε.

[0065] DREAM (ZS) The algorithm calculates multiple different Markov chains in parallel. If the parameter state of the i-th chain is x i , new suggestion point x i,prop =xi +Δ i , where the increment Δ i For the incremental part to be updated, the acceptance rate of the recommended samples is determined by the Metropolis ratio α. In this embodiment, after the Markov chain converges, the next 25% of the samples are intercepted as the posterior distribution samples of the random variable x, so as to realize the parameter posterior distribution inversion that uses the monitoring information to make up for the insufficient test data of the physical and mechanical parameters of the rock mass.

[0066] Step 5, input the inverted posterior distribution of the rock mass physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage;

[0067] The inverted posterior distribution of the rock mass physical and mechanical parameters is input into the proxy model SVM(x) of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage. According to the calculation results, the mean value is taken as the predicted value of the slope deformation in each subsequent excavation and anchoring stage and compared with the actual monitoring value of the slope, as shown in Table 3.

[0068] Table 3 Comparison of slope deformation prediction and monitoring values

[0069]

[0070] The above embodiments are merely examples of the technical solutions of the present invention. The method for predicting deformation of high rock slope excavation anchoring based on machine learning involved in this embodiment is not limited to the contents described in the above embodiments, but is subject to the scope defined in the claims. This embodiment uses a machine learning algorithm to identify key rock mass physical and mechanical parameters and construct an agent model. By incorporating field monitoring information into a Bayesian framework to predict deformation of high rock slope excavation anchoring, it provides a basis for safety supervision of dam area engineering.

[0071] The present invention also provides a system according to the above-mentioned method for predicting deformation of rock high slope excavation anchoring based on machine learning, comprising:

[0072] Data acquisition module, used to obtain geological survey reports of high arch dam area, slope excavation progress and monitoring information;

[0073] A model building module is used to build a multi-stage excavation and anchoring model for a high rock slope based on the data acquired by the data acquisition module;

[0074] A random variable screening module is used to sort the importance of the rock mass physical and mechanical parameters used to construct the model based on a machine learning algorithm, and identify key rock mass physical and mechanical parameters as random variables to be updated;

[0075] A proxy model training module is used to train the proxy model between random variables and slope responses in each excavation and anchoring stage based on a machine learning algorithm;

[0076] The module for obtaining the posterior distribution of physical and mechanical parameters is used to establish the functional relationship between the monitoring information and the proxy model, and to construct a Bayesian framework to incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass;

[0077] The slope deformation prediction module is used to input the posterior distribution of the rock mass physical and mechanical parameters inverted by the posterior distribution acquisition module of the physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage.

[0078] The above are only preferred embodiments of the present invention, and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of the specification of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for predicting deformation of rock high slope excavation anchoring based on machine learning, characterized in that: The steps include: Step 1: Obtain the geological survey report of the high arch dam area, the slope excavation progress and monitoring information, and use them to build a multi-stage excavation and anchoring model for rock high slopes; Step 2: sort the importance of the rock mass physical and mechanical parameters used to construct the model based on the machine learning algorithm, and identify the key rock mass physical and mechanical parameters as random variables to be updated; Step 3, constructing a proxy model between random variables and slope response in each excavation and anchoring stage based on a machine learning algorithm; Step 4: Establish a functional relationship between the monitoring information and the proxy model, construct a Bayesian framework, and incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass; Step 5, input the inverted posterior distribution of the rock mass physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage; Wherein, step 3 comprises: The key physical and mechanical parameters of the rock mass identified in step 2 are defined as random variables x = [x1, x2, …, x d ]; The support vector machine algorithm is used to train the proxy model SVM(x) between the random variables and the slope response in each excavation and anchoring stage, and the proxy model SVM(x) is used to replace the FLAC3D model of the multi-stage excavation and anchoring model of the rock high slope; In step 4, the monitoring information Y = [y1, y2, ..., y n ] and the surrogate model SVM(x) is as follows: Y=SVM(x)+ε Where ε represents the measurement error.

2. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 1 is characterized in that: Step 1 The method for constructing a multi-stage excavation and anchoring model for a high rock slope is as follows: based on the geological survey report of the dam area, the slope excavation progress and monitoring information, a mesh of a finite element model is generated in ANSYS, and then imported into FLAC3D through a plug-in to construct a multi-stage excavation and anchoring model for a high rock slope.

3. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 2 is characterized in that: For the multi-stage excavation and anchoring model of high rock slope, the Mohr-Coulomb constitutive model is adopted. Normal constraints are applied around the model and the bottom side is fixed. Different groups are used to distinguish the various types of rock mass and structural surface materials in the model, and the null model is used to simulate the excavation material and the cable model is used to simulate the prestressed anchor cable.

4. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 1 is characterized in that: Step 2 is implemented as follows: Generate a rock mass physical and mechanical parameter data set based on Latin hypercube sampling, generate a corresponding command stream based on the rock mass physical and mechanical parameter data set, and input the command stream into the FLAC3D model to calculate the corresponding slope response data set; The random forest algorithm is used for analysis. The rock mass physical and mechanical parameter dataset is defined as input, and the slope response dataset is defined as output. The rock mass physical and mechanical parameters are ranked according to their importance based on the importance measurement method in the random forest algorithm, and the rock mass physical and mechanical parameters that have a key impact on the slope response are identified as random variables to be updated.

5. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 4 is characterized in that: The physical and mechanical parameters of rock mass include elastic modulus E, Poisson's ratio μ, gravity γ, cohesion c and internal friction angle 6. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 1 is characterized in that: Based on Bayesian theory, the functional relationship of monitoring information is incorporated into the following formula, and the adaptive differential evolution Metropolis algorithm DREAM with multi-chain parallel computing is used. (ZS) To invert the posterior distribution of the physical and mechanical parameters of the rock mass: p(x|Y)=cL(x|Y)p(x) Where p(x|Y) represents the posterior probability density function of the parameter; p(x) represents the prior probability density function of the parameter, and the parameters obey the log-normal distribution and are independent of each other; c is the normalization constant; L(x|Y) represents the likelihood function; subscript j represents the jth rock mass physical and mechanical parameter; μ N and σ N represents the logarithmic mean and standard deviation of the parameter; the subscript n represents the nth excavation and anchoring stage; μ ε and σ ε represents the mean and standard deviation of the measurement error ε.

7. The method for predicting deformation of rock high slope excavation anchoring based on machine learning according to claim 1 is characterized in that: Step 5 specifically includes: The inverted posterior distribution of rock mass physical and mechanical parameters is input into the proxy model SVM(x) of each excavation stage to predict the slope deformation in the subsequent excavation and anchoring stage. According to the calculation results, the mean value is taken as the predicted value of slope deformation in the subsequent excavation and anchoring stages and compared with the actual monitoring value of the slope.

8. A system for implementing the method for predicting deformation of rock high slope excavation anchoring based on machine learning as described in any one of claims 1 to 7, characterized in that: include: Data acquisition module, used to obtain geological survey reports of high arch dam area, slope excavation progress and monitoring information; A model building module is used to build a multi-stage excavation and anchoring model for a high rock slope based on the data acquired by the data acquisition module; A random variable screening module is used to sort the importance of the rock mass physical and mechanical parameters used to construct the model based on a machine learning algorithm, and identify key rock mass physical and mechanical parameters as random variables to be updated; A proxy model training module is used to train the proxy model between random variables and slope responses in each excavation and anchoring stage based on a machine learning algorithm; The module for obtaining the posterior distribution of physical and mechanical parameters is used to establish the functional relationship between the monitoring information and the proxy model, and to construct a Bayesian framework to incorporate the functional relationship into the inversion of the posterior distribution of the physical and mechanical parameters of the rock mass; The slope deformation prediction module is used to input the posterior distribution of the rock mass physical and mechanical parameters inverted by the posterior distribution acquisition module of the physical and mechanical parameters into the proxy model of each excavation and anchoring stage to predict the slope deformation in the subsequent excavation and anchoring stage.

Images