Cage type anti-floating anchor rod applicable condition judgment method and system thereof

Abstract

The application relates to the field of civil construction, and discloses a cage type anti-floating anchor rod applicable condition judgment method and system, the method comprises the following steps: collecting field geological data; learning the field geological data based on a support vector machine algorithm, establishing a geological property identification model; according to the geological property identification model, evaluating the applicability of the field geological data, and outputting anchor rod application suggestions; using a deep neural network, predicting anchor rod design parameters according to the field geological data and the anchor rod application suggestions; using a genetic algorithm to optimize the anchor rod design parameters, and obtaining a design scheme. Through the geological property identification model of the support vector machine, the application can identify the adaptability of the anchor rod under a complex geological environment; compared with a traditional geomechanics model, the application overcomes the shortcomings caused by material heterogeneity and simplification of an underground water flow field, avoids limited prediction accuracy, significantly reduces the design misjudgment rate through more accurate geological identification, and improves the scientificity and reliability of the design.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of civil construction, in particular to a cage type anti-floating anchor rod applicable condition judgment method and system. BACKGROUND

[0002] In complex geological environments, it is crucial to ensure the stability of structures such as foundation pits and bridge piers. Traditional geomechanical models play a key role in this process. They predict the bearing capacity and stability of anchor rods by calculating factors such as soil pressure distribution and hydrodynamic effects. These models have played an important role in promoting the safety and economy of structural design. The existing technical solutions corresponding to the present technical solution: traditional geomechanical models mainly include finite element method and boundary element method. These methods establish mathematical models based on the physical and mechanical properties of rock and soil, and simulate and predict the stress and deformation mode of anchor rods.

[0003] However, such models are often based on idealized assumptions, such as ignoring the anisotropy and heterogeneity of materials, simplifying complex groundwater flow fields, resulting in limited prediction accuracy, especially in complex geological conditions, which may result in large errors, thereby affecting the safety and reliability of engineering design. SUMMARY

[0004] In view of the shortcomings of the prior art, the present application provides a cage type anti-floating anchor rod applicable condition judgment method and system, which solves the problem that the model is based on idealized assumptions, resulting in limited prediction accuracy, and in complex working conditions, large errors may occur, thereby affecting the safety and reliability of engineering design.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a cage type anti-floating anchor rod applicable condition judgment method, comprising the following steps:

[0006] Collecting field geological data;

[0007] Learning the field geological data based on a support vector machine algorithm to establish a geological property identification model;

[0008] According to the geological property identification model, the applicability of the field geological data is evaluated, and an anchor rod application suggestion is output;

[0009] Using a deep neural network, the anchor rod design parameters are estimated according to the field geological data and the anchor rod application suggestion;

[0010] Using a genetic algorithm to optimize the anchor rod design parameters to obtain a design scheme.

[0011] Preferably, the field geological data includes rock density, permeability coefficient, water content, and soil particle size distribution.

[0012] Preferably, the deep neural network comprises an input layer, a hidden layer and an output layer, the input layer is used to receive field geological data, and the output layer is used to estimate the anchor rod design parameters.

[0013] Preferably, the anchor rod design parameters include the length, diameter and spacing of the anchor rod.

[0014] Preferably, the genetic algorithm optimizes the anchor rod design parameters by simulating a natural evolution process based on the anchor rod design parameters.

[0015] Preferably, the natural evolution process comprises:

[0016] Selection: selecting design parameters with high fitness values as parents;

[0017] Crossover: combining parent design parameters into new child design parameters through crossover to generate several design candidates;

[0018] Mutation: mutating the several design candidates of the children, and the mutation comprises randomly changing the design parameters;

[0019] Evaluation: calculating the fitness values of all design candidates after mutation, and determining whether the design candidates meet the design target according to the fitness values;

[0020] Termination condition: when the design candidates meet the requirements, terminating the evolution process, and outputting the design candidate as the optimal design scheme.

[0021] Preferably, the deep neural network further comprises a probabilistic risk assessment for generating a risk assessment report based on seismic load, groundwater level fluctuation and seasonal temperature change.

[0022] A cage anti-floating anchor rod applicable condition judgment system comprises:

[0023] A data acquisition module is used to collect field geological data;

[0024] A geological property identification module classifies the field geological data based on a support vector machine algorithm and outputs an anchor rod application suggestion; a design parameter optimization module optimizes the anchor rod design parameters in the anchor rod application suggestion in combination with a deep neural network and a genetic algorithm; and a risk assessment module generates a risk probability atlas and adjusts design redundancy by using a probabilistic risk assessment method based on the anchor rod design parameters optimized by the design parameter optimization module.

[0025] A user interaction module is used to provide an interactive interface and display the design scheme and the anchor rod design parameters in real time.

[0026] The present application provides a cage anti-floating anchor rod applicable condition judgment method and system. The present application has the following beneficial effects:

[0027] 1、The present application can accurately identify the adaptability of anchor rods in complex geological environments by using a geological property intelligent identification model based on a support vector machine algorithm. Compared with traditional geomechanical models, the present technology overcomes the shortcomings of ignoring material heterogeneity and simplifying the groundwater flow field, and avoids the problem of limited prediction accuracy. Through more accurate identification of geological properties, the misdiagnosis rate during the engineering design stage is significantly reduced, and the scientificity and reliability of the design scheme are improved.

[0028] 2、The present application can greatly shorten the design cycle, improve work efficiency, and reduce labor and time input by using an integrated deep neural network and genetic algorithm for automatic optimization of design parameters. In addition, automatic optimization also effectively reduces material waste, reduces project costs, and improves overall economic efficiency.

[0029] 3、The present application analyzes uncertain factors, including seismic load and groundwater level fluctuations, during the design process using a probabilistic risk assessment mechanism, generates a risk probability map, and adjusts the design redundancy. Compared with existing technologies that lack comprehensive risk assessment schemes, the present technology significantly enhances the safety redundancy of the design scheme. Even under extreme working conditions, the design scheme can ensure the stability of the anchor rod, reduce potential risk hazards, and ensure the safety and long-term feasibility of the engineering project. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is a method flowchart of the present application;

[0031] Figure 2 is a system architecture diagram of the present application. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0033] In order to better understand the present application, the above content will be described in detail below in combination with specific embodiments.

[0034] Embodiment 1: Please refer to the attached Figure 1 The present application provides a cage anti-floating anchor rod applicable condition determination method, comprising the following steps:

[0035] Collecting field geological data;

[0036] In this embodiment, the acquisition quality of geological data directly affects the accuracy of subsequent geological characteristic identification, anchor rod design parameter estimation and optimization results. Therefore, the present application comprehensively considers key parameters in different geological environments in the data acquisition stage to ensure the comprehensiveness and representativeness of the geological data, and provides reliable input basis for subsequent calculation.

[0037] The collection of geological data mainly includes but is not limited to the following parameters:

[0038] Generally, rock density, permeability coefficient, water content and soil particle size distribution are mainly collected. These parameters can be obtained by field drilling sampling, geophysical measurement, permeability experiment and the like to ensure the authenticity and effectiveness of the data.

[0039] In one possible implementation, the rock density refers to the mass of unit volume of rock-soil body. This parameter is usually determined by field sampling combined with indoor experiment, for example, measured by using Archimedes method or density bottle method. For heterogeneous stratum, the rock density change at different depths also needs to be considered, and the data is processed by weighted average to more accurately reflect the geological conditions.

[0040] Specifically, the permeability coefficient reflects the seepage capacity of underground water in rock-soil medium, and the determination of the permeability coefficient can adopt field pumping test or laboratory permeability test. Generally, in homogeneous soil layer, the permeability coefficient can be calculated by Darcy's law:

[0041] q = k · A · i;

[0042] Wherein, q is the seepage flow; A is the seepage cross-sectional area; i is the hydraulic slope; k is the permeability coefficient;

[0043] In some embodiments, for layered heterogeneous soil layer, the equivalent permeability coefficient method can be used for calculation, wherein the vertical permeability coefficient k v and the horizontal permeability coefficient k h are calculated as follows:

[0044]

[0045] Wherein, k i is the permeability coefficient of the i-th layer of soil; H i is the thickness of the i-th layer of soil; n is the sample number.

[0046] As an option, the water content represents the ratio of the mass of water in the soil to the mass of dry soil, and the determination of the water content generally adopts the drying method, that is, the soil sampled in the field is dried to constant weight, and the mass difference before and after evaporation of water is calculated to determine the water content w:

[0047]

[0048] wherein m w is the mass of water in the soil; m d is the mass of the dried soil.

[0049] In some embodiments, to improve the accuracy of the water content data, in-situ real-time monitoring can be combined with time domain reflectometry (TDR) or resistivity measurement methods.

[0050] In addition, the soil particle size distribution refers to the gradation of soil particles, which is generally determined by sieving or sedimentation. Specifically, in coarse-grained soils (such as sand, gravel), sieving is used to separate soil particles through different aperture screens, and the content of each size fraction is determined. In fine-grained soils (such as silt, clay), specific gravity meter method or laser particle size analysis method can be used to determine the distribution of particle size. The soil particle size distribution has an important influence on the uplift performance of the anchor, so it needs to be measured and statistically analyzed.

[0051] In one possible implementation, to improve the accuracy of geological data collection, in-situ testing can be used, such as standard penetration test (SPT) for evaluating the tightness of the soil layer, cone penetration test (CPT) for measuring the shear strength parameters of the soil, seismic wave testing method (such as Rayleigh wave testing) for determining the elastic modulus of the soil, and groundwater level monitor for monitoring the long-term groundwater level trend.

[0052] Moreover, the collection of geological data can be combined with remote sensing technology, such as geological radar (GPR) or satellite remote sensing imaging, for extracting geological features in a large range, improving the timeliness and coverage of data collection.

[0053] Therefore, by collecting in-situ geological data in multiple ways, the accuracy and completeness of the input data are ensured, thereby providing a reliable basis for subsequent identification of geological characteristics, optimization of design parameters, and risk assessment.

[0054] Based on the support vector machine algorithm, the in-situ geological data is learned to establish a geological characteristic identification model;

[0055] In this embodiment, the learning of in-situ geological data is the core link of establishing the geological characteristic identification model. The accuracy of the model directly affects the evaluation quality of the applicability conditions of the cage anti-floating anchor, so in the data processing and learning process, the diversity of geological data and its influence on the adaptability of the anchor should be fully considered. By using the support vector machine (SVM) algorithm, the collected geological data is classified and feature extracted, which can establish an efficient and stable geological characteristic identification model, thereby providing a basis for subsequent design parameter estimation and optimization.

[0056] In this embodiment, the support vector machine algorithm is used to classify and learn key geological data such as rock density, permeability coefficient, water content, and soil particle size distribution, and to establish an adaptability determination model. Generally, SVM is suitable for data classification in high-dimensional space and can construct an optimal decision boundary between nonlinear geological data characteristics to improve classification accuracy.

[0057] In one possible implementation, the collected geological data is first standardized to eliminate the influence between different data scales. The standardization method can use mean normalization or Z-score normalization, where the Z-score normalization calculation is as follows:

[0058]

[0059] where X ′ is the standardized data; X is the original data; μ is the mean of the data; and σ is the standard deviation of the data.

[0060] As an option, in the geological data preprocessing stage, the principal component analysis (PCA) method can also be used to reduce the dimension of the data to reduce redundant information and improve computational efficiency. PCA mainly realizes dimension reduction of high-dimensional data by solving the eigenvectors of the covariance matrix, and the calculation formula is as follows:

[0061]

[0062] where C is the covariance matrix; X i is the i-th data sample; μ is the total number of samples; and (X i - μ) T is the transpose of the decentered data.

[0063] In some embodiments, SVM uses a radial basis kernel function (RBF-Kernel) for nonlinear classification to improve the accuracy of the adaptability determination model. The expression of the radial basis kernel function is as follows:

[0064] K(x i ,x j ) = exp(-γ∥x i -x j ∥ 2 );

[0065] where K(x i ,x j ) is the kernel function value between samples x i and x j ; γ is the parameter of the kernel function; ∥x i -x j ∥ 2 is the square of the Euclidean distance; and represents the sample xi and x j The distance between the two points is used to measure the difference or similarity between the samples.

[0066] Specifically, the SVM training process finds the optimal hyperplane to maximize the data classification margin by solving the following optimization problem:

[0067]

[0068] y i (w·x i +b)≥1;

[0069] where w is the hyperplane normal vector, b is the bias term, y i is the class label of the sample.

[0070] In one possible implementation, after training is complete, when classifying new geological data, the decision function is used:

[0071] f(x)=sign(w·x+b);

[0072] If f(x)>0, the geological condition is suitable for cage anti-floating anchor; otherwise, the condition is not suitable.

[0073] As a further optimization method, the cross-validation method can be used to optimize the SVM model to determine the optimal hyperparameters (such as kernel parameter gamma). Generally, K-fold cross-validation can be used to divide the training data set into K parts for training and testing to evaluate the generalization ability of the model.

[0074] To improve the robustness of the geological property recognition model, ensemble learning methods such as multiple-SVM can also be used, that is, multiple SVM models are used to independently train different types of geological features, and the applicability of the anchor is determined through a voting mechanism. This method can effectively improve the accuracy of applicability determination in complex geological environments.

[0075] Therefore, the geological property recognition model based on the support vector machine algorithm realizes the classification learning of the field geological data, and combines feature engineering, kernel function optimization and model integration technology to improve the accuracy and stability of the anchor applicability evaluation, providing data basis for subsequent design parameter estimation and optimization.

[0076] According to the geological property recognition model, the applicability of the field geological data is evaluated, and the anchor application suggestion is output;

[0077] In this embodiment, based on the field geological data collected in the previous step, the geological property identification model can evaluate the applicability of the data, thereby outputting appropriate cage anti-floating anchor application suggestions. Through this process, the present application can accurately determine whether the cage anti-floating anchor is suitable for the specific construction environment in a complex geological environment, thereby effectively reducing design errors and unnecessary cost waste in the design stage.

[0078] Specifically, the role of the geological property identification model is to learn and analyze the collected geological data and evaluate the applicability of the cage anti-floating anchor under different geological characteristics. The model uses a support vector machine (SVM) algorithm to map the geological data to a higher-dimensional feature space and determine the applicability by calculating the similarity between different feature vectors. Specifically, SVM constructs a decision boundary by finding the maximum classification interval to determine whether each geological sample is suitable for the cage anti-floating anchor.

[0079] Generally, the output of the model is a binary classification result indicating whether the measured geological conditions meet the use standards of the cage anti-floating anchor. The decision value obtained by calculation can determine whether the geological conditions are suitable for using the cage anti-floating anchor.

[0080] In some embodiments, the support vector machine uses a radial basis kernel function (RBF kernel function) for nonlinear classification to improve the accuracy of the model under complex geological conditions. The kernel function calculates the similarity between samples to help the model better handle nonlinear problems and improve the accuracy of applicability determination.

[0081] To further enhance the predictive ability of the model, especially for cases with complex geological characteristics, the present application can also combine multiple different kernel functions to improve the adaptability of the model. In this way, the support vector machine can more accurately divide the applicable and inapplicable geological conditions in high-dimensional space, further improving the accuracy of the cage anti-floating anchor application suggestions.

[0082] In some embodiments, for different geological conditions such as rock type, soil density, permeability coefficient, and other parameters, the model can automatically adjust the classification boundary of the support vector machine by learning historical geological data, thereby accurately determining which conditions are suitable for the use of the cage anti-floating anchor. Therefore, the model not only improves the accuracy of prediction, but also reduces design errors caused by human factors.

[0083] In addition, the output of the geological property identification model is not limited to suitability judgment, but can further output specific design recommendations based on the evaluation results. According to the results of suitability evaluation, the system can automatically suggest appropriate cage anti-floating anchor design scheme. For example, if the geological conditions meet the use standard, the system can output cage anti-floating anchor with standard length, diameter and spacing; if not applicable, the system will suggest to use different anti-floating design scheme or adjust the design parameters.

[0084] Therefore, the geological property identification model based on support vector machine algorithm can accurately evaluate the suitability of anchor in complex geological environment, avoiding the prediction error of traditional methods in complex environment. Through this evaluation process, the invention realizes more efficient and accurate cage anti-floating anchor design suggestion, significantly improving the safety and economy of engineering design.

[0085] Using deep neural network, the anchor design parameters are estimated according to the field geological data and anchor application suggestions;

[0086] In this embodiment, deep neural network (DNN) is used to estimate the design parameters of anchor. Through learning the relationship between a large amount of historical geological data and anchor design parameters, the model can automatically extract key features from complex geological characteristics and give preliminary design suggestions. The core goal of this process is to convert geological data into preliminary parameters of anchor design, thereby providing strong support for subsequent optimization and adjustment.

[0087] Specifically, deep neural network (DNN) is used as the core component of the model, which processes and transforms the input data step by step through multiple levels of neuron nodes. In this model, the input data includes various geological properties collected in the field, such as rock density, permeability coefficient, water content, soil particle size distribution, etc. Specifically, DNN receives these geological data through the input layer, and gradually extracts the potential features in the data through several hidden layers, finally generates the preliminary design parameters of anchor in the output layer, such as length, diameter, spacing, etc.

[0088] Generally, the structure of deep neural network is composed of multiple levels of neuron nodes. Each layer will perform weighted summation on the output of the previous layer and perform nonlinear transformation through activation function. Specifically, the output z (l) It can be calculated by the following formula:

[0089] z (l) = W (l) ·a (l-1) + b (l) ;

[0090] Where W (l) is the weight matrix of the l-th layer; a (l-1) is the activation output of the l-1 layer; b(l) The bias vector of the first layer.

[0091] Through layer-by-layer calculation, the neural network converts the input geological data into preliminary parameters of the anchor design. These parameters are summarized through the output layer to obtain the preliminary design results of the anchor.

[0092] As an option, the activation function of the deep neural network can select the ReLU (Rectified-Linear-Unit) function, which can effectively solve the gradient disappearance problem and improve the training efficiency. The calculation formula of ReLU is as follows: (l) a = max(0, z (l) );

[0093] where a (l) is the activation value; through this function, the model can effectively model the nonlinear features in the input data.

[0094] In some embodiments, to improve the accuracy and robustness of the model, the training process of the network uses the backpropagation algorithm to optimize the model parameters by minimizing the loss function. The loss function usually uses methods such as mean square error (MSE) or cross-entropy (Cross-Entropy), and the specific choice varies according to the nature of the task and the target. The expression of the mean square error loss function Z is as follows:

[0095]

[0096] where, is the model prediction value; y a is the actual value; and n is the number of samples.

[0097] Specifically, in the application of the present application, the deep neural network can automatically predict the preliminary design parameters of the anchor according to the field geological data. This process greatly reduces the involvement of manual work and can fully consider various complex geological features, laying a solid foundation for subsequent optimization and adjustment. The model continuously optimizes its prediction ability through multiple training and verification, and finally outputs design parameters that meet the bearing capacity requirements.

[0098] In one possible implementation, the present application can also combine other optimization algorithms such as genetic algorithm (GA) or particle swarm optimization (PSO) to further optimize the preliminary design results. Through these optimization algorithms, the model can find the optimal solution within the range of design parameters, thereby effectively reducing costs and improving design efficiency.

[0099] Therefore, through the learning and estimation of field geological data by the deep neural network, the application can quickly and accurately calculate the preliminary design parameters of the cage anti-floating anchor rod, providing strong technical support for subsequent optimization and adjustment, while reducing the error of manual intervention and improving the accuracy and safety of engineering design.

[0100] The genetic algorithm is used to optimize the design parameters of the anchor rod to obtain a design scheme.

[0101] In this embodiment, the deep neural network (DNN) provides basic parameters for the preliminary design of the cage anti-floating anchor rod. These preliminary design parameters include the length, diameter and spacing of the anchor rod, and other important design elements. In order to further optimize the design and ensure the economy and safety of the design scheme, a genetic algorithm (GA) is used to optimize these preliminary design parameters to obtain the final optimal design scheme.

[0102] Specifically, the genetic algorithm (GA) optimizes the design parameters by simulating the process of species evolution in nature. The algorithm includes steps such as selection, crossover, mutation and evaluation, aiming to gradually approach the global optimal solution by continuously iterating optimization. The core idea of the genetic algorithm is to gradually approach the global optimal solution by selecting high-quality individuals and generating new individuals.

[0103] Specifically, the initial population of the genetic algorithm is composed of preliminary design parameters generated by the deep neural network. Each individual represents a potential design scheme, including parameters such as the length, diameter and spacing of the anchor rod. Generally, the selection operation is completed by calculating the fitness of each individual. The fitness f opt (L,D,S) is calculated by the following objective function: f opt (L,D,S) = α·C mat (L,D,S) + β·C labor (L,D,S) - θ·Strength(L,D,S);

[0104] Where L represents the length of the anchor rod; D represents the diameter of the anchor rod; S represents the spacing of the anchor rod; C mat is the material consumption cost; C labor is the labor cost; Strength represents the bearing capacity of the anchor rod; α, β, θ are weight coefficients used to control the relative importance between different objectives.

[0105] And in the selection process, individuals with higher fitness are preferentially selected to generate the next generation of individuals. Next, new design schemes are generated through crossover operations. These new design schemes inherit some of the excellent features of the parent individuals, and through mutation operations, a certain randomness is introduced to increase the diversity of the solution space and prevent the algorithm from falling into a local optimal solution.

[0106] Meanwhile, crossover operations typically employ single-point or multi-point crossover. Specifically, one possible implementation involves dividing the parent generation's design parameters into two parts, exchanging information at the crossover point, and generating a new child generation design scheme. Mutation operations, on the other hand, involve making minor, random modifications to a portion of the design parameters to increase the breadth of the search space. The probability and range of mutation are usually controlled within the algorithm parameters to ensure that the algorithm does not converge prematurely to an undesirable solution.

[0107] For each generation of individuals, the algorithm selects the optimal design as the basis for the next generation by evaluating its fitness function. As the number of iterations increases, high-quality designs gradually become dominant, eventually converging to the optimal solution. The final optimal design will satisfy all constraints, including requirements for load-bearing capacity, safety, and economy.

[0108] In some embodiments of the present invention, to improve the efficiency and accuracy of the genetic algorithm, a local search algorithm may be combined to refine the optimal solution. The local search algorithm can further optimize the accuracy of the solution based on the genetic algorithm. For example, simulated annealing or particle swarm optimization can be used to refine the results of the genetic algorithm to ensure the final optimization of the design scheme.

[0109] Specifically, the optimization process of the genetic algorithm combines geological characteristic identification models with preliminary design parameters estimated by deep neural networks, making the optimization more closely aligned with actual working conditions. Through this multi-level, multi-step optimization mechanism, this invention can further reduce material consumption and engineering costs, improve design efficiency, and reduce unnecessary waste, while ensuring that the anchor bolt design meets the bearing capacity requirements.

[0110] Therefore, by combining deep neural networks and genetic algorithms, the design parameters of cage-type anti-buoyancy anchor bolts were efficiently optimized. In this process, the genetic algorithm effectively improved the accuracy and economy of the design scheme, ensuring the safety and cost-effectiveness of the project. Simultaneously, this optimization process is automated, reducing manual intervention and improving the reliability and efficiency of the design.

[0111] Example 2: Please refer to the appendix Figure 2 This embodiment provides a system for determining the applicable conditions of cage-type anti-buoyancy anchor bolts, including:

[0112] The data acquisition module is used to collect on-site geological data;

[0113] The geological characteristic identification module classifies on-site geological data based on the support vector machine algorithm and outputs application suggestions for cage-type anti-buoyancy anchor bolts;

[0114] The design parameter optimization module combines deep neural networks and genetic algorithms to optimize the anchor bolt design parameters in the application suggestions.

[0115] a risk assessment module that generates a risk probability map using a probabilistic risk assessment method based on the optimized anchor design parameters from the design parameter optimization module and adjusts the design redundancy;

[0116] a user interaction module that provides an interactive interface to display the design scheme and anchor design parameters in real-time.

[0117] In this embodiment, for the collected field geological data in each operation, a storage module is introduced to save and manage all historical data, establishing an efficient geological condition evaluation system. The introduction of the storage module not only ensures that the geological data of each project is effectively archived, but also improves design efficiency and accuracy through data reuse and historical data comparison in subsequent projects.

[0118] Specifically, the main function of the storage module is to record the geological data of each engineering project in real time and store it together with the corresponding design scheme and optimization results. Specifically, all field measured geological data, including rock density, permeability coefficient, water content, soil particle size distribution, and other information, will be saved in a database. Each data record will be associated with the project number, geological characteristics, anchor design parameters, and other information to ensure that the data can be clearly traced and managed between projects.

[0119] In general, the storage module uses a relational database (such as MySQL, PostgreSQL, etc.) or a non-relational database (such as MongoDB, NoSQL, etc.) for data storage and management. Through the table structure design in the database, geological data is associated with project details. For example, the table structure in the database may include the following fields:

[0120] Project Number: A unique identifier for each engineering project;

[0121] Geological Data: Rock density, permeability coefficient, water content, soil particle size distribution, etc.

[0122] Design Parameters: Including the length, diameter, and spacing of the anchor rod for preliminary design data;

[0123] Optimization Results: The final design scheme optimized by the genetic algorithm;

[0124] Risk Assessment Results: Risk map based on the probabilistic risk assessment method and design redundancy adjustment suggestions.

[0125] As an alternative, to improve the efficiency of query and data processing of the storage module, historical data can also be indexed and classified. Through the classification index of different project categories and geological conditions, users can quickly find similar historical cases to the geological conditions of the new project when encountering a new project, thereby providing a reference and basis for the design scheme of the new project.

[0126] In a possible implementation, the system can quickly retrieve the matching historical geological data and the corresponding cage anti-floating anchor design scheme from the storage module according to the new engineering project input field measured geological data. In this way, the design of the new project can not only be customized according to the current field data, but also can combine the successful design scheme and optimization result in the historical data to improve the accuracy and efficiency of the design. For example, if the geological conditions of a new engineering project are highly similar to those of a previous project, the system can quickly extract the historical data of the project and provide a preliminary design scheme based on the data.

[0127] Specifically, the system first standardizes and normalizes the field collected geological data through the data preprocessing module to ensure consistent data format. Then, the suitability evaluation is performed using the support vector machine (SVM) based geological property recognition model to determine whether the geological conditions are suitable for cage anti-floating anchor. If suitable, the system will provide the most matching design parameters based on the stored data, and combine deep neural network (DNN) to perform preliminary parameter estimation of anchor design. Finally, the storage module extracts the optimization scheme of the historical project as a reference and optimizes the design of the new project.

[0128] As a further technical extension, the storage module can also integrate machine learning and artificial intelligence technology, gradually "learn" the design optimization experience of each engineering project, and feed it back to the model for continuous improvement. For example, the system can automatically track feedback from historical projects, collect effect data after engineering implementation, and re-evaluate and optimize the design scheme. Through continuous learning process, the accuracy of the design scheme can be continuously improved, and the project risk can be further reduced.

[0129] Therefore, by adding the storage module, the present application not only realizes an efficient geological condition evaluation system, but also accumulates and reuses historical project data, further improves the design efficiency and accuracy, and reduces the design cost and risk. In a new project, only the field measured geological data needs to be input, and the system can quickly output the appropriate cage anti-floating anchor application suggestion, making the engineering design process more intelligent and automated.

[0130] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for determining the applicable conditions of a cage-type anti-buoyancy anchor bolt, characterized in that, Includes the following steps: Collect on-site geological data; A geological characteristic identification model is established by learning from field geological data based on the support vector machine algorithm. The field geological data includes rock density, permeability coefficient, water content, and soil particle size distribution. Based on the geological characteristic identification model, the applicability of the on-site geological data is assessed, and recommendations for anchor bolt application are output. Using deep neural networks, anchor bolt design parameters are estimated based on on-site geological data and anchor bolt application recommendations. A genetic algorithm was used to optimize the anchor bolt design parameters to obtain a design scheme; The deep neural network includes an input layer, a hidden layer, and an output layer. The input layer is used to receive on-site geological data, and the output layer is used to estimate anchor bolt design parameters. The anchor bolt design parameters include the length, diameter, and spacing of the anchor bolts; The genetic algorithm is based on anchor bolt design parameters and optimizes the anchor bolt design parameters by simulating a natural evolutionary process; The natural evolutionary process includes: Selection: Select design parameters with high fitness values ​​as the parent generation; Crossover: By crossover, parent design parameters are combined into new child design parameters to generate several design candidate schemes; Mutation: Mutating several design candidate schemes of the offspring, the mutation including randomly changing the design parameters; Evaluation: Calculate the fitness values ​​of all design candidate solutions after mutation, and determine whether the design candidate solutions meet the design objectives based on the fitness values; Termination condition: When the design candidate solution meets the requirements, the evolution process is terminated, and the optimal design solution is output. The method for learning from field geological data using the Support Vector Machine (SVM) algorithm to establish a geological data learning model includes: using the SVM algorithm to classify key geological data and establish an adaptive judgment model; standardizing the collected geological data to eliminate the influence between different data scales; using radial basis kernel functions for nonlinear classification to improve the accuracy of the adaptive judgment model; finding the optimal hyperplane by solving an optimization problem during the training process to maximize the data classification margin; using a decision function when classifying new geological data after training; and combining ensemble learning methods to independently train different types of geological features and comprehensively judge the applicability of anchor bolts through a voting mechanism. The output of the geological characteristic identification model is a binary classification result, indicating whether the measured geological conditions meet the usage standards of cage-type anti-buoyancy anchors. The output of the geological characteristic identification model is not limited to applicability judgment, but can also further output specific design suggestions based on the evaluation results. Based on the applicability evaluation results, the system can automatically suggest a suitable cage-type anti-buoyancy anchor design scheme.

2. The method for determining the applicable conditions of a cage-type anti-buoyancy anchor bolt according to claim 1, characterized in that, The deep neural network further includes probabilistic risk assessment, used to generate risk assessment reports based on seismic loads, groundwater level fluctuations, and seasonal temperature changes.

3. A system for determining the applicability of cage-type anti-buoyancy anchors, based on a method for determining the applicability of cage-type anti-buoyancy anchors according to any one of claims 1-2, characterized in that, include: The data acquisition module is used to collect on-site geological data; The geological characteristic identification module classifies on-site geological data based on the support vector machine algorithm and outputs anchor bolt application suggestions; The design parameter optimization module combines deep neural networks and genetic algorithms to optimize the anchor bolt design parameters in the anchor bolt application recommendations. The risk assessment module, based on the anchor bolt design parameters optimized by the design parameter optimization module, uses a probabilistic risk assessment method to generate a risk probability map and adjust the design redundancy. The user interaction module provides an interactive interface that displays design schemes and anchor bolt design parameters in real time.

Images