CN120316887A - Tunnel deformation regression prediction model training method, result generation method and equipment - Google Patents
Tunnel deformation regression prediction model training method, result generation method and equipmentInfo
- Publication number
- CN120316887A CN120316887A CN202510796888.2A CN202510796888A CN120316887A CN 120316887 A CN120316887 A CN 120316887A CN 202510796888 A CN202510796888 A CN 202510796888A CN 120316887 A CN120316887 A CN 120316887A
- Authority
- CN
- China
- Prior art keywords
- tunnel
- model
- random field
- data
- tunnel deformation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Civil Engineering (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Architecture (AREA)
- Computing Systems (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明提供了一种隧道变形回归预测模型训练方法、结果生成方法及设备，涉及岩土工程技术领域，包括：通过初始三维隧道数值模型和参数统计分布特征得到随机场样本数据；基于中心点法，将随机场样本数据对应赋值给初始三维隧道数值模型得到目标三维隧道数值模型；通过目标三维隧道数值模型得到隧道变形响应数据，返回通过初始三维隧道数值模型和所述参数统计分布特征得到随机场样本数据的步骤，重复上述步骤直至达到预设样本获取次数得到隧道变形数据集；基于Bagging算法，对隧道变形数据集进行重复抽样，训练机器学习模型得到隧道变形回归预测模型。本发明实现了提高隧道变形回归预测模型的精度。
The present invention provides a tunnel deformation regression prediction model training method, result generation method and equipment, which relate to the field of geotechnical engineering technology, including: obtaining random field sample data through an initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics; assigning the random field sample data to the initial three-dimensional tunnel numerical model based on the center point method to obtain a target three-dimensional tunnel numerical model; obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and the parameter statistical distribution characteristics, repeating the above steps until a preset number of sample acquisition times is reached to obtain a tunnel deformation data set; based on the Bagging algorithm, repeatedly sampling the tunnel deformation data set, training a machine learning model to obtain a tunnel deformation regression prediction model. The present invention achieves the improvement of the accuracy of the tunnel deformation regression prediction model.
Description
Technical Field
The invention relates to the technical field of geotechnical engineering, in particular to a tunnel deformation regression prediction model training method, a result generation method and equipment.
Background
In the tunnel construction design process, tunnel deformation conditions at different stages need to be acquired, so that constructors are helped to know potential risks, and corresponding measures are taken to reduce adverse effects. Different from artificial synthetic materials, tunnel deformation data has natural uncertainty, and even two similar points can have difference, so that a large amount of data is required to meet the prediction of tunnel deformation conditions, and the prediction of tunnel deformation conditions is usually carried out by using a machine learning model in the related technology. However, because tunnel engineering has the characteristics of long construction period, complex construction environment, variable geological conditions and the like, effective monitoring data which can be accumulated in actual engineering is very limited. Such small sample data often has difficulty meeting the requirements of the training data volume of the model, resulting in lower accuracy of the tunnel deformation regression prediction model.
Disclosure of Invention
The invention solves the problem of how to improve the precision of the tunnel deformation regression prediction model.
In order to solve the problems, the invention provides a tunnel deformation regression prediction model training method, a result generation method and equipment.
In a first aspect, the present invention provides a method for training a regression prediction model of tunnel deformation, including:
Obtaining random field sample data through an initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics;
based on a central point method, correspondingly assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model;
Obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set, wherein the tunnel deformation data set comprises at least one group of random field sample data and corresponding tunnel deformation response data;
And repeatedly sampling the tunnel deformation data set based on a Bagging algorithm, and training a machine learning model to obtain a tunnel deformation regression prediction model.
Optionally, before the random field sample data is obtained through the initial three-dimensional tunnel numerical model and the parameter statistical distribution feature, the method further comprises:
Acquiring three-dimensional tunnel modeling data, wherein the three-dimensional tunnel modeling data comprises the geometric dimension, grid data, load conditions, boundary conditions, material properties and Shi Gongbu of a tunnel;
Inputting the three-dimensional tunnel modeling data into a numerical simulation model to obtain the initial three-dimensional tunnel numerical model;
and obtaining the parameter statistical distribution characteristics through the initial three-dimensional tunnel numerical model based on a space statistical analysis algorithm.
Optionally, the obtaining random field sample data through the initial three-dimensional tunnel numerical model and the parameter statistical distribution feature includes:
obtaining the size of a random field mathematical model according to the initial three-dimensional tunnel numerical model;
based on a Cholesky matrix decomposition method, obtaining a lower triangular matrix through the size of the random field mathematical model and the parameter statistical distribution characteristics;
obtaining a random number sequence vector by using a pseudo-random number generator, wherein the random number sequence vector obeys standard normal distribution;
and obtaining the random field sample data through the lower triangular matrix and the random number sequence vector.
Optionally, the obtaining a lower triangular matrix through the size of the random field mathematical model and the parameter statistical distribution feature includes:
Obtaining correlation coefficients of all space points in the random field mathematical model size according to the parameter statistical distribution characteristics, wherein the parameter statistical distribution characteristics comprise autocorrelation distances of the space points;
Wherein, the correlation coefficient is:
,
Wherein, the The relative distances of the two space points in the directions of the x axis, the y axis and the z axis are respectively、、The correlation coefficient at the time of the time,、、The autocorrelation distances of the spatial points in the directions of the x axis, the y axis and the z axis are respectively;
obtaining a correlation matrix through all the correlation coefficients;
Wherein, the correlation matrix is:
,
Wherein, the For the purpose of the correlation matrix,For the correlation coefficients of the i-th and j-th said spatial points, i=1, 2, & gt, n, j=1, 2, & gt, n, n being the number of spatial points;
Decomposing the correlation matrix to obtain the lower triangular matrix;
Wherein, the lower triangular matrix is:
,
Wherein, the For the lower triangular matrix to be described,Is the transpose of the lower triangular matrix.
Optionally, the assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model includes:
obtaining random field center coordinates of each grid cell according to the random field sample data;
obtaining the central coordinates of the numerical model of each grid unit according to the initial three-dimensional tunnel numerical model;
And judging the position relation between the central coordinates of each random field and the central coordinates of the numerical model, and assigning the data of the grid cells in the random field sample data to the corresponding grid cells in the initial three-dimensional tunnel numerical model according to the judging result to obtain the target three-dimensional tunnel numerical model.
Optionally, the tunnel deformation regression prediction model includes at least one sub-model, the resampling the tunnel deformation dataset, training a machine learning model to obtain the tunnel deformation regression prediction model, includes:
Carrying out data preprocessing on the tunnel deformation data set to obtain a processed tunnel deformation data set;
Randomly extracting samples from the processed tunnel deformation data set to obtain a target data set;
Training the machine learning model according to the target data set to obtain the sub-model;
And returning to the step of obtaining a target data set by randomly extracting samples from the tunnel deformation data set, repeating the steps until the preset training times are reached, and taking all the submodels as the tunnel deformation regression prediction model.
Optionally, the training the machine learning model according to the target data set to obtain the sub-model includes:
Inputting the target data set into the machine learning model for training until the model converges to obtain an initial training model;
and testing the model accuracy of the initial training model, and obtaining the sub-model when the model accuracy reaches an accuracy threshold.
In a second aspect, the present invention further provides a method for generating a tunnel deformation regression prediction result, including:
Acquiring target random field sample data of a tunnel to be predicted;
Inputting the target random field sample data into the tunnel deformation regression prediction model obtained by the tunnel deformation regression prediction model training method to obtain tunnel deformation response prediction data;
And carrying out confidence interval estimation on the tunnel deformation response prediction data to obtain a tunnel deformation regression prediction result of the tunnel to be predicted.
In a third aspect, the present invention provides an electronic device comprising a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to implement the tunnel deformation regression prediction model training method according to the first aspect or the tunnel deformation regression prediction result generation method according to the second aspect when executing the computer program.
In a fourth aspect, the present invention provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the tunnel deformation regression prediction model training method as described in the first aspect or the tunnel deformation regression prediction result generation method as described in the second aspect.
The tunnel deformation regression prediction model training method, the result generation method and the device have the beneficial effects that the random field sample data is obtained by introducing the parameter statistical distribution characteristics, so that the actual engineering sample characteristics can be reflected more truly. And assigning the random field sample data to the initial three-dimensional tunnel numerical model based on a central point method to obtain a target three-dimensional tunnel numerical model, wherein the central point method improves modeling efficiency on the premise of guaranteeing the physical meaning of the model. And obtaining tunnel deformation response data according to the target three-dimensional tunnel numerical model, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set. By continuously generating new random field sample data and running numerical simulation, a large number of data pairs are accumulated gradually to obtain a tunnel deformation data set, and a high-quality training sample set is constructed under the condition of data scarcity. And obtaining a tunnel deformation regression prediction model by carrying out put-back sampling on the tunnel deformation data set and training a plurality of base models, thereby improving the precision of the tunnel deformation regression prediction model.
Drawings
FIG. 1 is a schematic flow chart of a training method of a tunnel deformation regression prediction model according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a three-dimensional tunnel numerical model according to an embodiment of the present invention;
FIG. 3 is a schematic flow chart of a method for generating a regression prediction result of tunnel deformation according to an embodiment of the present invention;
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. While the invention is susceptible of embodiment in the drawings, it is to be understood that the invention may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided to provide a more thorough and complete understanding of the invention. It should be understood that the drawings and embodiments of the invention are for illustration purposes only and are not intended to limit the scope of the present invention.
It should be understood that the various steps recited in the method embodiments of the present invention may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the invention is not limited in this respect.
The term "comprising" and variations thereof as used herein is meant to be open-ended, i.e., "including but not limited to," based at least in part on, "one embodiment" means "at least one embodiment," another embodiment "means" at least one additional embodiment, "some embodiments" means "at least some embodiments," and "optional" means "optional embodiment. Related definitions of other terms will be given in the description below. It should be noted that the concepts of "first", "second", etc. mentioned in this disclosure are only used to distinguish between different devices, modules or units, and are not intended to limit the order or interdependence of functions performed by these devices, modules or units.
It should be noted that references to "a" and "an" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.
As shown in fig. 1, a training method for a tunnel deformation regression prediction model provided by an embodiment of the present invention includes:
and 110, obtaining random field sample data through the initial three-dimensional tunnel numerical model and the parameter statistical distribution characteristics.
Specifically, the parameter statistical distribution characteristics refer to acquiring mechanical parameters (such as elastic modulus, poisson ratio, cohesive force, internal friction angle and density) of various rock-soil bodies in a field through acquisition or experimental analysis, and determining the statistical distribution characteristics (such as mean value, variance and space autocorrelation distance) of the parameters by combining a space statistical analysis method, so that a reliable statistical basis is provided for subsequent random field modeling.
And 120, correspondingly assigning the random field sample data to the initial three-dimensional tunnel numerical model based on a central point method to obtain a target three-dimensional tunnel numerical model.
In particular, the random field is an extension of a random process, and the random field sample data is a set of random variables defined on a multi-dimensional set of parameters, where each point corresponds to a parameter value and each cell includes a plurality of points. And assigning the parameter value of each unit in the random field sample data to the corresponding unit of the initial three-dimensional tunnel numerical model by adopting the value at the center point of the unit. And directly binding the random field parameters with the middle points of grids of the initial three-dimensional tunnel numerical model, transmitting the constructed space variability and statistical characteristics to the initial three-dimensional tunnel numerical model, and replacing a deterministic analysis model in numerical software with an uncertainty analysis model to obtain the target three-dimensional tunnel numerical model.
And 130, obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, and returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and the parameter statistical distribution characteristics, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set, wherein the tunnel deformation data set comprises at least one group of random field sample data and corresponding tunnel deformation response data.
Specifically, the tunnel deformation response data includes dome subsidence data and peripheral convergence data. Repeatedly generating the random field sample data, and then continuously importing the initial three-dimensional tunnel numerical model to generate a new target three-dimensional tunnel numerical model. Because the generation of the random field is different each time, the corresponding three-dimensional tunnel numerical model of the target is also different. And extracting a group of possible parameters each time, substituting the group of parameters into an analytical model of the tunnel to calculate, simulating the tunnel condition under the group of data, and further acquiring tunnel deformation response data to obtain the deformation of the tunnel. This is repeated hundreds and thousands of times to see how much the tunnel response will change under different geological conditions. And finally, counting the results, wherein the number of times is the number of times exceeding a safety limit value, namely the number of times of failure, accounting for the total number of times, namely the ratio of times of failure, and repeatedly carrying out sampling, calculating and counting processes to carry out randomness analysis. A large number of different parameter combinations or random field samples are randomly generated and are sequentially imported into a numerical model by using a central point method, so that each group of samples corresponds to one complete numerical simulation, and tunnel deformation response data corresponding to different random field sample data are obtained.
And 140, repeatedly sampling the tunnel deformation data set based on a Bagging algorithm, and training a machine learning model to obtain a tunnel deformation regression prediction model.
Specifically, a Bagging algorithm is introduced, a plurality of support vector regression models are selected as a base learner, an integrated model is formed by training on different subsamples, and the accuracy and the stability of a machine learning model are improved by combining predictions of the plurality of models. A plurality of different training sets are generated from the tunnel deformation dataset using a subsampled approach, each training set being used to train a separate basis model.
In the embodiment, the random field sample data is obtained by introducing the parameter statistical distribution characteristic, so that the actual engineering sample characteristics can be reflected more truly. And assigning the random field sample data to the initial three-dimensional tunnel numerical model based on a central point method to obtain a target three-dimensional tunnel numerical model, wherein the central point method improves modeling efficiency on the premise of guaranteeing the physical meaning of the model. And obtaining tunnel deformation response data according to the target three-dimensional tunnel numerical model, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set. By continuously generating new random field sample data and running numerical simulation, a large number of data pairs are accumulated gradually to obtain a tunnel deformation data set, and a high-quality training sample set is constructed under the condition of data scarcity. And obtaining a tunnel deformation regression prediction model by carrying out put-back sampling on the tunnel deformation data set and training a plurality of base models, thereby improving the precision of the tunnel deformation regression prediction model.
Optionally, before the random field sample data is obtained through the initial three-dimensional tunnel numerical model and the parameter statistical distribution feature, the method further comprises:
Acquiring three-dimensional tunnel modeling data, wherein the three-dimensional tunnel modeling data comprises the geometric dimension, grid data, load conditions, boundary conditions, material properties and Shi Gongbu of a tunnel;
Inputting the three-dimensional tunnel modeling data into a numerical simulation model to obtain the initial three-dimensional tunnel numerical model;
and obtaining the parameter statistical distribution characteristics through the initial three-dimensional tunnel numerical model based on a space statistical analysis algorithm.
Specifically, establishing an initial three-dimensional tunnel numerical model refers to establishing a three-dimensional tunnel numerical model to be simulated in numerical software, wherein the modeling process comprises geometric dimension determination, grid data, load conditions, boundary conditions, constitutive models, material properties, construction steps and the like. The constitutive model and the material property refer to description of an elastoplastic constitutive model for a stratum rock-soil body, and a Mohr-Coulomb model or Drucker-Prager model is preferably adopted to simulate the stress-strain relation of the stratum rock-soil body, wherein the material property comprises mechanical parameters such as elastic modulus, poisson's ratio, cohesive force, internal friction angle, density and the like of the rock-soil body, and a linear elastic or nonlinear material model is selected according to the stress characteristics of the tunnel lining structure to endow the tunnel lining structure with corresponding material strength and rigidity parameters. The Mohr-Coulomb model (Mohr-Coulomb model) is a classical theoretical model for describing material damage conditions, and is widely applied in the fields of geomechanics, geotechnical engineering and the like. It is based on the shear failure theory, assuming that the failure of the material is caused by shear stress, and that the failure condition can be represented by a linear relationship. Drucker-Prager model (Deruker-Prague model) is an improvement on the Mohr-Coulomb model, mainly for overcoming the latter's problem of insensitivity to intermediate principal stresses. Based on the theory of plasticity, it is suitable for a wider range of stress states, especially those cases where the influence of intermediate principal stresses needs to be considered. Unlike the Mohr-Coulomb model, its failure plane appears as a cone in the principal stress space, which allows it to better accommodate complex stress distribution conditions, providing greater accuracy and applicability.
In some more specific embodiments, first, it is necessary to determine the geometry of the tunnel from actual engineering drawings or designs, including but not limited to the cross-sectional shape (circular, rectangular, etc.), diameter or width, length, and any particular structure such as junctions, branches, etc. of the tunnel. The grid data is a process of discretizing the tunnel and surrounding rock-soil mass, and is divided into a plurality of cells so as to facilitate numerical calculation. Reasonable grid design is critical to computational accuracy. Generally, finer meshes should be employed near the tunnel walls and in areas of varying geological conditions to capture local stress and deformation characteristics, while relatively coarse meshes may be used away from these critical areas to reduce computation. The geometric dimension and the grid division are set according to the actual engineering site conditions, wherein the site model is generally 3-6 times of the tunnel hole diameter so as to fully reflect the boundary effect, in the grid division process, the tunnel structure and the adjacent areas adopt denser unit division so as to improve the calculation accuracy, and the areas far away from the tunnel adopt relatively coarse unit division so as to improve the calculation efficiency. The loading conditions mainly comprise self gravity, ground stress (original rock stress field), underground water pressure and other factors. Accurate setting of these loads is very important for simulating the true stress state of the tunnel. The ground stress is generally estimated based on field measured data or empirical formulas. Boundary conditions define behavior at the edges of the model. It is common practice to set the sides as normal fixed (i.e. not allowing displacement perpendicular to the boundary), the bottom as fully fixed (not allowing displacement in both horizontal and vertical directions), and the top as the case may be, to apply a distributed force or a free surface treatment. The load condition and the boundary condition are that dead weight load of the rock-soil body and other additional loads in actual engineering are applied to simulate an initial stress state, the boundary condition is set to be that the bottom of the model is fully fixed, the side limit is used for horizontal displacement or rolling boundary, and the top of the model is a free surface boundary so as to reasonably simulate the mechanical boundary environment of an actual stratum. Shi Gongbu refers to various stages in the simulated tunnel excavation and support process. This typically involves gradually removing units representing the excavated area and adding support structures (e.g., shotcrete, anchor rods, steel arches, etc.) at the appropriate time.
In some more specific embodiments, taking a three-dimensional tunnel engineering as an example, modeling is performed by selecting a finite element method that is currently very mainstream, as shown in fig. 2. The method mainly comprises the steps of taking the diameter of a tunnel to be 6m according to the general size of the tunnel when the ground surface is piled up, dividing grid units after a finite element model is built, setting a gravity load and a top additional load in a field, setting the gravity acceleration to be 9.8m/s 2 and the additional load to be 74kPa, setting boundary conditions, fully fixing the bottom, setting rolling boundaries on the side boundaries and setting the top to be a free surface, and selecting a Mohr-Coulomb model by the constitutive model, wherein only a single soil layer is considered for simplicity. Through reference, the elastic modulus is 20MPa, the Poisson ratio is 0.33, the soil body weight is 18.5kN/m 3, and the cohesion and the internal friction angle are 25kPa and 25 degrees respectively. The tunnel lining adopts a linear elastic model, the elastic modulus is 25GPa, the Poisson ratio is 0.2, and the weight is 24.5kN/m 3. In the setting of the construction step, firstly, the ground stress balance is carried out, then the calculation is carried out after the whole tunnel is excavated, and when the balance state is reached, the additional load of the earth surface is applied until the calculation is converged. The deformation is monitored as the maximum settlement deformation of the tunnel vault under the influence of surface loading.
In the optional embodiment, a three-dimensional tunnel model capable of reflecting real conditions is established in numerical software, so that possible problems in the tunnel construction process can be effectively predicted, the design scheme is optimized, and the engineering safety is improved.
Optionally, the obtaining random field sample data through the initial three-dimensional tunnel numerical model and the parameter statistical distribution feature includes:
obtaining the size of a random field mathematical model according to the initial three-dimensional tunnel numerical model;
based on a Cholesky matrix decomposition method, obtaining a lower triangular matrix through the size of the random field mathematical model and the parameter statistical distribution characteristics;
obtaining a random number sequence vector by using a pseudo-random number generator, wherein the random number sequence vector obeys standard normal distribution;
and obtaining the random field sample data through the lower triangular matrix and the random number sequence vector.
Specifically, the parameter statistical distribution feature comprises an average value and a standard deviation of a geotechnical parameter, the random field sample data is obtained through the lower triangular matrix and the random number sequence vector, and the method comprises the following steps:
obtaining a standard normal random field through the product of the lower triangular matrix and the random number sequence vector;
wherein the standard normal random field is:
,
wherein Z is the standard normal random field, For the lower triangular matrix to be described,Is the random column vector;
The random field sample data is obtained through the equal probability transformation of the average value, the standard deviation and the standard normal random field, and the random field sample data is used for representing the random field obeying any average value and standard deviation;
Wherein the random field sample data is:
,
Wherein, the For the random field sample data,As a result of the mean value of the values,Is the standard deviation.
In some more specific embodiments, the model mesh division information including the cell size, the node number and the cell number is read according to the basic geometric parameters of the explicit model of the initial three-dimensional tunnel numerical model, including the tunnel length, the tunnel width or diameter and the tunnel height, so as to obtain the random field mathematical model size. And generating a group of column vectors obeying the standard normal distribution by using a pseudo-random number generator, wherein the dimension is n, and multiplying the lower triangular matrix by the random column vectors to obtain the standard normal random field. Finally, by means of the equal probability transformation, random fields with other mean values, standard deviations and distribution forms (such as lognormal distribution) are generated. The repeated sampling generates random column vectors, i.e., different random fields can be generated.
In some more specific embodiments, the statistical distribution of the geotechnical parameters is used to represent data obtained by actual site survey, and the mean value of the geotechnical parameters is calculatedStandard deviation ofAnd fit its probability distribution form. The data obtained from actual site surveys refer to the mean value of mechanical parameters such as modulus of elasticity, cohesion, friction angle, and standard deviation. Fitting its probability distribution form is used to represent the collected sample data, ensuring that the data is representative. Then, visualization is performed through a histogram or Kernel Density Estimation (KDE), and preliminary judgment is made as to which distribution form (e.g., normal, log-normal, gamma, index, etc.) the data approximately takes on. Next, a statistical tool (e.g., scipy. Stats of Python) is used to parametrically fit the candidate distributions, and the fit effect of each distribution is evaluated by a goomagorov-Smirnov test (KS test) or an AIC/BIC goodness test index. Finally, the distribution form with the best fitting effect is selected as a result. scipy.stats is a module in the SciPy library of Python, dedicated to statistical calculations. The method provides a large number of functions such as probability distribution, statistical test, descriptive statistics, parameter estimation and the like, and is a very common tool in scientific research, data analysis and machine learning. The Kolmogorov-Smirnov test (K-S test for short) is a commonly used non-parametric hypothesis test method for determining whether one sample is subject to a certain theoretical distribution (e.g., normal distribution, uniform distribution, etc.), or whether two samples are from the same population. AIC (Akaike Information Criterion, red pool information amount criterion) and BIC (Bayesian Information Criterion ) are two commonly used model selection criteria. They are mainly used to evaluate the balance between complexity and goodness of fit of statistical models, helping researchers to choose the best one among multiple models.
Optionally, the obtaining a lower triangular matrix through the size of the random field mathematical model and the parameter statistical distribution feature includes:
Obtaining correlation coefficients of all space points in the random field mathematical model size according to the parameter statistical distribution characteristics, wherein the parameter statistical distribution characteristics comprise autocorrelation distances of the space points;
Wherein, the correlation coefficient is:
,
Wherein, the The relative distances of the two space points in the directions of the x axis, the y axis and the z axis are respectively、、The correlation coefficient at the time of the time,、、The autocorrelation distances of the spatial points in the directions of the x axis, the y axis and the z axis are respectively;
obtaining a correlation matrix through all the correlation coefficients;
Wherein, the correlation matrix is:
,
Wherein, the For the purpose of the correlation matrix,For the correlation coefficients of the i-th and j-th said spatial points, i=1, 2, & gt, n, j=1, 2, & gt, n, n being the number of spatial points;
Decomposing the correlation matrix to obtain the lower triangular matrix;
Wherein, the lower triangular matrix is:
,
Wherein, the For the lower triangular matrix to be described,Is the transpose of the lower triangular matrix.
Specifically, cholesky matrix decomposition is a method of decomposing a symmetric positive definite matrix into a lower triangular matrix and its transpose. Generating a random field by utilizing a matrix decomposition method according to the acquired statistical distribution characteristics of the parameters in the same size range as the initial three-dimensional tunnel numerical model, and constructing a mechanical parameter uncertainty mathematical model of each unit in the soil, wherein the random field reflects the spatial variability of the parameters and meets the statistical characteristic requirement, thereby providing quantitative uncertainty basis for subsequent numerical analysis and machine learning prediction.
Optionally, the assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model includes:
obtaining random field center coordinates of each grid cell according to the random field sample data;
obtaining the central coordinates of the numerical model of each grid unit according to the initial three-dimensional tunnel numerical model;
And judging the position relation between the central coordinates of each random field and the central coordinates of the numerical model, and assigning the data of the grid cells in the random field sample data to the corresponding grid cells in the initial three-dimensional tunnel numerical model according to the judging result to obtain the target three-dimensional tunnel numerical model.
Specifically, the method of using a center point method to import the random field into a numerical model is to extract the coordinates of the center position of each grid cell in the random field sample data, then extract the coordinates of the center position of one grid cell in an initial three-dimensional tunnel numerical model (such as a finite element model), then calculate the distance between the center point of the grid cell in the random field sample data and the center point of the grid cell in the initial three-dimensional tunnel numerical model, and judge that when the distance between the two cells is minimum, the mechanical parameter value of the random field model grid is endowed to the grid cell of the corresponding position of the numerical model. Since the distance determination is performed by using the center position coordinates of the grid cells in the model mapping process, this mapping method is called a center point method.
In this alternative embodiment, the units divided by the initial three-dimensional tunnel numerical model are corresponding by a center point method, that is, the parameter value of each unit in the random field is given to the corresponding unit of the initial three-dimensional tunnel numerical model by adopting the value at the center point of the unit. Directly binding the random field sample data with the unit midpoint of the initial three-dimensional tunnel numerical model, transmitting the constructed space variability and statistical characteristics to the tunnel numerical model, and replacing a deterministic analysis model in numerical software with an uncertainty analysis model.
Optionally, the tunnel deformation regression prediction model includes at least one sub-model, the resampling the tunnel deformation dataset, training a machine learning model to obtain the tunnel deformation regression prediction model, includes:
Carrying out data preprocessing on the tunnel deformation data set to obtain a processed tunnel deformation data set;
Randomly extracting samples from the processed tunnel deformation data set to obtain a target data set;
Training the machine learning model according to the target data set to obtain the sub-model;
And returning to the step of obtaining a target data set by randomly extracting samples from the tunnel deformation data set, repeating the steps until the preset training times are reached, and taking all the submodels as the tunnel deformation regression prediction model.
Specifically, the data preprocessing is performed on the tunnel deformation data set to obtain a processed tunnel deformation data set, namely, after the data preprocessing, the format unification and the random rearrangement are performed, the data set is divided into a training set and a testing set according to a certain proportion, wherein the training set is used for training, verifying and optimizing a machine learning model, and the testing set is used for evaluating a final prediction result and determining a confidence interval. On the basis of the processed tunnel deformation data set, a plurality of base learners (such as support vector regression, random forest, decision tree and the like) are constructed by adopting a Bagging integration idea, the training samples are randomly resampled, and each base learner is independently trained on the sub-data set obtained by sampling, so that the prediction results of all base models are integrated and fused. The random field sample data includes elastic modulus, friction angle, and cohesive force parameters.
In some more specific embodiments, the random field data is input as elastic modulus with the elastic modulus as random field sample data, and the number of input features is related to the finite element mesh. The finite element model is built to have 12000 grid cells, so that the machine learning features 12000 elastic modulus parameters, and each 12000 random field data corresponds to one output value (namely corresponding tunnel deformation response data obtained by numerical simulation calculation). These features and corresponding targets are input into an integrated regression model for training, so that the learning of the mapping relation between the elastic modulus and the tunnel deformation response data is realized.
In the optional embodiment, model training is performed according to the Bagging integration thought, a plurality of training subsets are generated through substitution sampling, a plurality of basic learners are trained respectively, and then results of the learners are integrated, so that generalization capability is improved.
Optionally, the training the machine learning model according to the target data set to obtain the sub-model includes:
Inputting the target data set into the machine learning model for training until the model converges to obtain an initial training model;
and testing the model accuracy of the initial training model, and obtaining the sub-model when the model accuracy reaches an accuracy threshold.
In some more specific embodiments, an initial tunnel three-dimensional numerical calculation model is first established, the young modulus of the soil body is selected as a random field parameter to be simulated, the mean value of the elastic modulus is the same as the value in the first step, namely 20MPa, the standard deviation is 6MPa, the distribution type is selected as lognormal distribution, the relevant distances in three directions take the same value, and the relevant distances in the three directions are as follows:
;
And establishing a soil body mechanical parameter random field model with the same size of the initial tunnel three-dimensional numerical calculation model, and obtaining corresponding random field sample data. Determining the size of a random field mathematical model according to an initial tunnel three-dimensional numerical calculation model, calculating correlation coefficients of all space points in a field, constructing an autocorrelation matrix, then decomposing the correlation matrix according to a matrix decomposition method to obtain a lower triangular matrix, sampling a group of random variables which are subjected to standard normal distribution by using pseudo-random numbers to obtain a standard normal distribution random field, and finally performing equal probability transformation to obtain a group of random field sample data with the average value of 20MPa, the standard deviation of 6MPa and the distribution type of lognormal distribution. The random field sample data comprises a plurality of random field samples, different random variables are repeatedly sampled, the implementation process is circulated, a plurality of groups of random field mathematical models can be obtained, and the selection calculation times are 300, so that 300 groups of standard normal random variables are required to be repeatedly sampled to obtain 300 different random field samples.
And importing the random field into an initial tunnel three-dimensional numerical calculation model by using a center point method, calculating relative distances according to unit center point coordinates of the random field and the initial tunnel three-dimensional numerical calculation model, judging a unit point which is most matched with the finite element numerical model, and sequentially carrying out random field parameter assignment so as to realize grid unit mapping of the random field and the initial tunnel three-dimensional numerical calculation model and obtain a plurality of target tunnel three-dimensional numerical calculation models. And respectively carrying out numerical simulation on the three-dimensional numerical calculation model of the target tunnel to obtain tunnel deformation response under the set of random field parameters. And sequentially solving the three-dimensional numerical calculation model of the target tunnel corresponding to the 300 groups of random fields to obtain tunnel deformation response corresponding to the 300 groups of random field data. The tunnel deformation data sets obtained through calculation are sorted and divided, 300 random field data and 300 corresponding deformation responses are in one-to-one correspondence and are coded, a part of the data sets are randomly extracted to be used as training sets, the rest data sets are used as test sets, the data sets are divided according to the proportion of 2:1 in the embodiment, namely, the training sets are 100, the test sets are 200, and the data sets are used as data bases for training and verification of a subsequent machine learning model. A basic learner of a support vector regression (Support Vector Regression, SVR) model is selected as a machine learning model, 100 random field samples in a training set are used as input features of the machine learning model, 100 tunnel deformation responses in the training set are used as output targets, and preliminary training is carried out to continuously adjust and optimize super parameters, so that the prediction performance of the conventional support vector regression model at the moment is optimal. Based on the optimal super parameters, a Bagging method in integrated learning is fused, a plurality of data are randomly selected as subsets in 100 training sets, 70 subsets (namely 70%) are selected, a support vector regression model is trained by utilizing the super parameter combination just obtained to obtain a base learner, then 70 subsets (which cannot be repeated before) are randomly selected from 100 training sets to train to obtain a second base learner, the above process is repeated to obtain a plurality of base learners, 50 base learners are obtained by repeating 50 times, 50 support vector regression models are used as base learners for predicting tunnel deformation, and the base learners are all trained on different subsets, so that the 50 base learners are mutually independent. Thus, 50 different prediction results can be obtained in 50 base learners every time one random field data is input.
Corresponding to the above-mentioned training method of the tunnel deformation regression prediction model, as shown in fig. 3, a further embodiment of the present invention further provides a method for generating a tunnel deformation regression prediction result, including:
step 310, obtaining target random field sample data of a tunnel to be predicted;
step 320, inputting the target random field sample data into a tunnel deformation regression prediction model obtained by the tunnel deformation regression prediction model training method as described above to obtain tunnel deformation response prediction data;
And 330, estimating a confidence interval of the tunnel deformation response prediction data to obtain a tunnel deformation regression prediction result of the tunnel to be predicted.
Specifically, the tunnel deformation response prediction data are summarized, the central trend (such as mean or median) of the prediction result and the corresponding discrete degree are calculated, and the upper confidence limit and the lower confidence limit are determined by adopting a proper statistical method (such as confidence interval estimation or a quantitive method). Therefore, not only is a single tunnel deformation predicted value obtained, but also prediction fluctuation caused by model uncertainty and data randomness is reflected, and finally a tunnel deformation regression prediction result which contains specific deformation values and is provided with a reliability confidence interval is generated.
In some more specific embodiments, the tunnel deformation response prediction data includes 50 prediction results, and a mean value calculated by the 50 prediction results is output as a final prediction result, so as to obtain the tunnel deformation corresponding to the random field data. Therefore, the effective machine learning model takes the place of the traditional large-scale random finite element numerical calculation, and the calculation efficiency is effectively improved. Thereafter, the upper and lower limit values of the tunnel distortion of 50 groups are calculated as confidence intervals of the prediction result by using (98% quantile, 2% quantile), (90% quantile, 10% quantile), (80% quantile, 20% quantile) and (70% quantile, 30% quantile), respectively, so that the tunnel distortion prediction result with the confidence intervals is obtained. 200 random field data in the test set are input as a model and sequentially input into 50 basic learners, and a final prediction result and corresponding upper and lower limit values can be obtained for each random field data according to the strategy, so that the machine learning efficient agent of random finite elements and intelligent tunnel deformation prediction with confidence intervals are realized.
The tunnel deformation regression prediction result generation method has the same advantages as the tunnel deformation regression prediction model training method compared with the prior art, and is not described in detail herein.
As shown in fig. 4, an electronic device 400 provided by an embodiment of the present invention includes a memory 410 and a processor 420, where the memory 410 is configured to store a computer program, and the processor 420 is configured to implement the tunnel deformation regression prediction model training method or the tunnel deformation regression prediction result generating method described above when executing the computer program.
Alternatively stated, an electronic device 400 comprises a memory 410 and a processor 420 coupled to the memory 410, the memory 410 being configured to store a computer program, the processor 420 being configured to, when executing the computer program, perform the following:
Obtaining random field sample data through an initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics;
based on a central point method, correspondingly assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model;
Obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set, wherein the tunnel deformation data set comprises at least one group of random field sample data and corresponding tunnel deformation response data;
And repeatedly sampling the tunnel deformation data set based on a Bagging algorithm, and training a machine learning model to obtain a tunnel deformation regression prediction model.
The embodiment of the invention provides a computer readable storage medium, wherein a computer program is stored on the storage medium, and when the computer program is executed by a processor, the tunnel deformation regression prediction model training method or the tunnel deformation regression prediction result generating method are realized.
Alternatively, a non-transitory computer readable storage medium having a computer program stored thereon, which when executed by a processor, causes the processor to:
Obtaining random field sample data through an initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics;
based on a central point method, correspondingly assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model;
Obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set, wherein the tunnel deformation data set comprises at least one group of random field sample data and corresponding tunnel deformation response data;
And repeatedly sampling the tunnel deformation data set based on a Bagging algorithm, and training a machine learning model to obtain a tunnel deformation regression prediction model.
An electronic device 400 that may be a server or a client of the present invention will now be described as an example of a hardware device that may be applied to aspects of the present invention. Electronic device 400 is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Electronic device 400 may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
The electronic device 400 includes a computing unit that can perform various suitable actions and processes according to a computer program stored in a Read Only Memory (ROM) or a computer program loaded from a storage unit into a Random Access Memory (RAM). In the RAM, various programs and data required for the operation of the device may also be stored. The computing unit, ROM and RAM are connected to each other by a bus. An input/output (I/O) interface is also connected to the bus.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a random-access Memory (Random Access Memory, RAM), or the like.
Although the invention is disclosed above, the scope of the invention is not limited thereto. Various changes and modifications may be made by one skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications will fall within the scope of the invention.
Claims (10)
1. The tunnel deformation regression prediction model training method is characterized by comprising the following steps of:
Obtaining random field sample data through an initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics;
based on a central point method, correspondingly assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain a target three-dimensional tunnel numerical model;
Obtaining tunnel deformation response data through the target three-dimensional tunnel numerical model, returning to the step of obtaining random field sample data through the initial three-dimensional tunnel numerical model and parameter statistical distribution characteristics, and repeating the steps until the preset sample acquisition times are reached to obtain a tunnel deformation data set, wherein the tunnel deformation data set comprises at least one group of random field sample data and corresponding tunnel deformation response data;
And repeatedly sampling the tunnel deformation data set based on a Bagging algorithm, and training a machine learning model to obtain a tunnel deformation regression prediction model.
2. The method according to claim 1, wherein before the random field sample data is obtained by the initial three-dimensional tunnel numerical model and the parameter statistical distribution feature, further comprising:
Acquiring three-dimensional tunnel modeling data, wherein the three-dimensional tunnel modeling data comprises the geometric dimension, grid data, load conditions, boundary conditions, material properties and Shi Gongbu of a tunnel;
Inputting the three-dimensional tunnel modeling data into a numerical simulation model to obtain the initial three-dimensional tunnel numerical model;
and obtaining the parameter statistical distribution characteristics through the initial three-dimensional tunnel numerical model based on a space statistical analysis algorithm.
3. The method for training a regression prediction model for tunnel deformation according to claim 1, wherein the obtaining the random field sample data from the initial three-dimensional tunnel numerical model and the parameter statistical distribution features comprises:
obtaining the size of a random field mathematical model according to the initial three-dimensional tunnel numerical model;
based on a Cholesky matrix decomposition method, obtaining a lower triangular matrix through the size of the random field mathematical model and the parameter statistical distribution characteristics;
obtaining a random number sequence vector by using a pseudo-random number generator, wherein the random number sequence vector obeys standard normal distribution;
and obtaining the random field sample data through the lower triangular matrix and the random number sequence vector.
4. The method of claim 3, wherein said obtaining a lower triangular matrix from said random field mathematical model dimensions and said parameter statistical distribution characteristics comprises:
Obtaining correlation coefficients of all space points in the random field mathematical model size according to the parameter statistical distribution characteristics, wherein the parameter statistical distribution characteristics comprise autocorrelation distances of the space points;
Wherein, the correlation coefficient is:
,
Wherein, the The relative distances of the two space points in the directions of the x axis, the y axis and the z axis are respectively、、The correlation coefficient at the time of the time,、、The autocorrelation distances of the spatial points in the directions of the x axis, the y axis and the z axis are respectively;
obtaining a correlation matrix through all the correlation coefficients;
Wherein, the correlation matrix is:
,
Wherein, the For the purpose of the correlation matrix,For the correlation coefficients of the i-th and j-th said spatial points, i=1, 2, & gt, n, j=1, 2, & gt, n, n being the number of spatial points;
Decomposing the correlation matrix to obtain the lower triangular matrix;
Wherein, the lower triangular matrix is:
,
Wherein, the For the lower triangular matrix to be described,Is the transpose of the lower triangular matrix.
5. The method for training a regression prediction model for tunnel deformation according to claim 1, wherein the assigning the random field sample data to the initial three-dimensional tunnel numerical model to obtain the target three-dimensional tunnel numerical model comprises:
obtaining random field center coordinates of each grid cell according to the random field sample data;
obtaining the central coordinates of the numerical model of each grid unit according to the initial three-dimensional tunnel numerical model;
And judging the position relation between the central coordinates of each random field and the central coordinates of the numerical model, and assigning the data of the grid cells in the random field sample data to the corresponding grid cells in the initial three-dimensional tunnel numerical model according to the judging result to obtain the target three-dimensional tunnel numerical model.
6. The method of claim 1, wherein the tunnel deformation regression prediction model comprises at least one sub-model, wherein the resampling the tunnel deformation dataset trains a machine learning model to obtain the tunnel deformation regression prediction model, comprising:
Carrying out data preprocessing on the tunnel deformation data set to obtain a processed tunnel deformation data set;
Randomly extracting samples from the processed tunnel deformation data set to obtain a target data set;
Training the machine learning model according to the target data set to obtain the sub-model;
And returning to the step of obtaining a target data set by randomly extracting samples from the tunnel deformation data set, repeating the steps until the preset training times are reached, and taking all the submodels as the tunnel deformation regression prediction model.
7. The method of claim 6, wherein training the machine learning model from the target dataset to obtain the sub-model comprises:
Inputting the target data set into the machine learning model for training until the model converges to obtain an initial training model;
and testing the model accuracy of the initial training model, and obtaining the sub-model when the model accuracy reaches an accuracy threshold.
8. The tunnel deformation regression prediction result generation method is characterized by comprising the following steps of:
Acquiring target random field sample data of a tunnel to be predicted;
Inputting the target random field sample data into a tunnel deformation regression prediction model obtained by the tunnel deformation regression prediction model training method according to any one of claims 1 to 7 to obtain tunnel deformation response prediction data;
And carrying out confidence interval estimation on the tunnel deformation response prediction data to obtain a tunnel deformation regression prediction result of the tunnel to be predicted.
9. An electronic device comprising a memory and a processor;
the memory is used for storing a computer program;
the processor is configured to implement the tunnel deformation regression prediction model training method according to any one of claims 1 to 7 or the tunnel deformation regression prediction result generation method according to claim 8 when executing the computer program.
10. A computer-readable storage medium, characterized in that the storage medium has stored thereon a computer program which, when executed by a processor, implements the tunnel deformation regression prediction model training method according to any one of claims 1 to 7 or the tunnel deformation regression prediction result generation method according to claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510796888.2A CN120316887B (en) | 2025-06-16 | 2025-06-16 | Tunnel deformation regression prediction model training method, result generation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510796888.2A CN120316887B (en) | 2025-06-16 | 2025-06-16 | Tunnel deformation regression prediction model training method, result generation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN120316887A true CN120316887A (en) | 2025-07-15 |
| CN120316887B CN120316887B (en) | 2025-10-14 |
Family
ID=96331635
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510796888.2A Active CN120316887B (en) | 2025-06-16 | 2025-06-16 | Tunnel deformation regression prediction model training method, result generation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN120316887B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200182718A1 (en) * | 2016-11-12 | 2020-06-11 | Soochow University | Method for calculating earth pressure load on a tunnel |
| CN112395667A (en) * | 2020-11-10 | 2021-02-23 | 华中科技大学 | Random field parameter generation method and device for shield tunnel soil body |
| CN115374570A (en) * | 2022-04-29 | 2022-11-22 | 北京城建设计发展集团股份有限公司 | Multi-source weighted training set construction method for deformation prediction of engineering tunnel crossing |
| US20230144184A1 (en) * | 2021-11-11 | 2023-05-11 | Shandong University | Advanced geological prediction method and system based on perception while drilling |
| CN118132985A (en) * | 2024-03-04 | 2024-06-04 | 中铁四局集团有限公司 | A method for predicting large deformation of soft rock tunnels based on an improved Bayes model |
| CN118446058A (en) * | 2024-05-20 | 2024-08-06 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | A tunnel section deformation prediction method, device, terminal and storage medium |
| CN119293897A (en) * | 2024-08-23 | 2025-01-10 | 武汉市市政建设集团有限公司 | Deformation analysis method of shield tunneling through existing shield tunnel based on random field theory |
| CN119692178A (en) * | 2024-12-02 | 2025-03-25 | 中国二十二冶集团有限公司 | Method for determining the construction design parameters of isolation piles between new foundation pit and existing tunnel |
| CN119830387A (en) * | 2024-11-27 | 2025-04-15 | 长沙理工大学 | Training method, device, equipment and medium for tunnel bearing capacity prediction model |
-
2025
- 2025-06-16 CN CN202510796888.2A patent/CN120316887B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200182718A1 (en) * | 2016-11-12 | 2020-06-11 | Soochow University | Method for calculating earth pressure load on a tunnel |
| CN112395667A (en) * | 2020-11-10 | 2021-02-23 | 华中科技大学 | Random field parameter generation method and device for shield tunnel soil body |
| US20230144184A1 (en) * | 2021-11-11 | 2023-05-11 | Shandong University | Advanced geological prediction method and system based on perception while drilling |
| CN115374570A (en) * | 2022-04-29 | 2022-11-22 | 北京城建设计发展集团股份有限公司 | Multi-source weighted training set construction method for deformation prediction of engineering tunnel crossing |
| CN118132985A (en) * | 2024-03-04 | 2024-06-04 | 中铁四局集团有限公司 | A method for predicting large deformation of soft rock tunnels based on an improved Bayes model |
| CN118446058A (en) * | 2024-05-20 | 2024-08-06 | 哈尔滨工业大学(深圳)(哈尔滨工业大学深圳科技创新研究院) | A tunnel section deformation prediction method, device, terminal and storage medium |
| CN119293897A (en) * | 2024-08-23 | 2025-01-10 | 武汉市市政建设集团有限公司 | Deformation analysis method of shield tunneling through existing shield tunnel based on random field theory |
| CN119830387A (en) * | 2024-11-27 | 2025-04-15 | 长沙理工大学 | Training method, device, equipment and medium for tunnel bearing capacity prediction model |
| CN119692178A (en) * | 2024-12-02 | 2025-03-25 | 中国二十二冶集团有限公司 | Method for determining the construction design parameters of isolation piles between new foundation pit and existing tunnel |
Also Published As
| Publication number | Publication date |
|---|---|
| CN120316887B (en) | 2025-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Verification of stochastic seismic analysis method and seismic performance evaluation based on multi-indices for high CFRDs | |
| CN110837669B (en) | Dynamic Construction Method of Landslide Uncertainty Model Based on Multi-source Heterogeneous Data Fusion | |
| Beiki et al. | Genetic programming approach for estimating the deformation modulus of rock mass using sensitivity analysis by neural network | |
| Zhou et al. | Seismic fragility analysis of high concrete faced rockfill dams based on plastic failure with support vector machine | |
| CN112035939B (en) | Rock-soil body parameter random field modeling method for double-side-wall pilot tunnel | |
| CN112163328A (en) | Geotechnical particle material constitutive modeling method based on deep learning and data driving | |
| CN112395667B (en) | Random field parameter generation method and device for shield tunnel soil body | |
| CN117763895A (en) | A method and system for dynamic response sensitivity analysis of concrete gravity dams | |
| Li et al. | Soil seismic response modeling of KiK-net downhole array sites with CNN and LSTM networks | |
| Fu et al. | Slope stability analysis based on big data and convolutional neural network | |
| CN107609265B (en) | Finite element simulation method and system for formation stress field based on ant tracking | |
| CN116911148A (en) | A seismic damage assessment method and assessment system for building groups in sedimentary basins | |
| CN118330720B (en) | A method for identifying earthquake liquefaction considering the terrain effect of river valley sites | |
| Chen et al. | Method for rapidly generating urban damage scenarios under non-uniform ground motion input based on matching algorithms and time history analyses | |
| CN119203729A (en) | A method and device for intelligent prediction of pile foundation bearing capacity | |
| CN120180537A (en) | A full-process modeling method, device and storage medium for foundation pit | |
| Van Nguyen et al. | Deep learning application for nonlinear seismic ground response prediction based on centrifuge test and numerical analysis | |
| CN117235510A (en) | Joint roughness prediction method and joint roughness prediction model training method | |
| Zhao et al. | Determining geomechanical parameters and a deformation uncertainty analysis of the Longtan Hydropower Station slope, China | |
| CN120316887B (en) | Tunnel deformation regression prediction model training method, result generation method and device | |
| CN114117753A (en) | Vulnerability-based probabilistic earthquake slope slip risk analysis method and device | |
| Preston et al. | Models of earthquake faults with long-range stress transfer | |
| CN120257737A (en) | Parametric rapid modeling method, system, equipment and medium for foundation pit finite element model | |
| Zheng et al. | Machine learning based damage state identification: A novel perspective on fragility analysis for nuclear power plants considering structural uncertainties | |
| Pan et al. | Dynamic analysis of nonlinear civil engineering structures using artificial neural network with adaptive training |
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 |