Soft soil parameter value analysis and correction method and system based on machine learning
Through a machine learning-based method, geological engineering data is used for preprocessing and nonlinear correlation analysis, combined with mixed integer nonlinear programming and Bayesian network, the accuracy, reliability and applicability of soft soil parameter analysis is solved. The problem is to achieve more efficient soft soil parameter prediction and model adaptability.
Patent Information
- Application Number
- CN202410540036.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-30
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing soft soil parameter analysis methods have problems such as low analysis accuracy, low prediction reliability, low applicability, and inability to effectively deal with outliers that may be encountered in actual projects.
Using a machine learning-based approach, data preprocessing is performed by collecting geological engineering data, identifying and utilizing nonlinear correlations in soft soil behavior, and training models using a mixed integer nonlinear programming method, combining nonlinear kernel functions and Bayesian networks , process discrete and continuous decision variables in the model, dynamically adjust model parameters to adapt to environmental changes.
It improves the accuracy and reliability of soft soil parameter analysis, enhances the applicability and accuracy of the model, can better handle outliers and environmental changes, and improves data processing capabilities in engineering applications.
Smart Images

Figure CN118917163A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of soft soil parameter analysis, and in particular to a method and system for analyzing and correcting soft soil parameter values based on machine learning. Background Art
[0002] In the field of modern civil engineering, the accurate analysis and correction of soft soil parameters are of great significance, because these parameters directly affect the stability and safety of engineering structures. With the rapid development of computing technology, machine learning technology has been widely used in geological engineering data analysis to improve the accuracy and efficiency of parameter prediction. Traditional soil analysis methods rely on laboratory tests and empirical formulas. Although these methods have provided reliable support in the past few decades, they have limitations in processing large amounts of data and identifying complex nonlinear relationships. In addition, with the expansion of the scale of engineering projects and the complexity of environmental conditions, higher requirements are placed on the accuracy and adaptability of soil parameter analysis methods. This has promoted the research and application of advanced data analysis techniques, especially the exploration of machine learning models in soil behavior analysis, because they can learn and predict soil behavior from a large amount of geological data.
[0003] Although machine learning has shown excellent data processing and prediction capabilities in many fields, existing technologies still face some key challenges in the analysis and correction of soft soil parameter values. First, existing machine learning applications generally fail to fully utilize the nonlinear characteristics of soft soil behavior, which limits the accuracy and generalization ability of the model under complex geological conditions. Second, traditional machine learning methods often ignore the processing of outliers that may be encountered in actual engineering during model training, which may lead to unstable prediction results or large errors. In addition, many methods lack effective parameter optimization mechanisms and cannot dynamically adjust the model to adapt to environmental changes or new data inputs. Summary of the invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] Therefore, the technical problem solved by the present invention is that the existing soft soil parameter analysis method has the problems of low analysis accuracy, low prediction reliability, low applicability, and inability to handle the problem of abnormal value processing that may be encountered in actual engineering.
[0006] To solve the above technical problems, the present invention provides the following technical solutions: a method for analyzing and correcting soft soil parameter values based on machine learning, comprising collecting geological engineering data; performing data preprocessing based on the geological engineering data and identifying and utilizing the nonlinear correlation in soft soil behavior; and training a model through a mixed integer nonlinear programming method to perform soft soil parameter analysis and correction.
[0007] As a preferred solution of the soft soil parameter value analysis and correction method based on machine learning described in the present invention, wherein: the acquisition of geological engineering data includes the acquisition of soil physical and chemical properties, soil environmental data, engineering data, laboratory test data and geological survey data;
[0008] The soil physical and chemical properties include density, water content, particle size distribution, plasticity index and liquid limit, organic matter content, chemical composition, shear strength and compressibility;
[0009] The soil environmental data include groundwater level, temperature and humidity records and historical settlement data;
[0010] The engineering data includes load history, engineering structure type and construction process records;
[0011] The laboratory test data include standard penetration test results, compression tests, and shear tests;
[0012] The geological survey data include stratigraphic distribution and geological age.
[0013] As a preferred solution of the soft soil parameter value analysis and correction method based on machine learning described in the present invention, wherein: the data preprocessing based on geological engineering data and identifying and utilizing the nonlinear correlation in the soft soil behavior includes weighted conversion of the data using information entropy and redistributing the weight according to the amount of information of each data point;
[0014] The conversion is based on the entropy process to represent the measure of uncertainty and randomness of information;
[0015] Output the information entropy of each sample point, adjust the weight of each sample point according to the entropy value, and preprocess the data set, which is expressed as:
[0016]
[0017] Where x represents the original input data point, x″ represents the transformed data point, n represents the total number of data points, and x i represents the value of the i-th data point, x j represents the value of the jth data point used in the summation process;
[0018] By preprocessing the data set, the amount of concentrated information determines the contribution of each point to the whole.
[0019] As a preferred solution of the soft soil parameter value analysis and correction method based on machine learning described in the present invention, wherein: the data preprocessing based on geological engineering data and identifying and utilizing the nonlinear correlation in the soft soil behavior includes mapping the original data to a high-dimensional feature space through a nonlinear kernel function, the kernel function used includes a combination of a polynomial and a radial basis kernel function, selecting the kernel parameters of the polynomial degree and the γ parameter of the RBF, and converting the data to a high-dimensional space through a kernel technology, which is expressed as:
[0020]
[0021] Among them, K(x,x′) represents the kernel function, which is used to output the similarity between data points x and x′, x, x′ represent the two data points compared in the kernel function, and θ i represents the weight parameter of the i-th dimension, p i represents the power of the polynomial kernel of the i-th dimension, γ i represents the scale parameter of the radial basis function RBF kernel of the i-th dimension, d represents the number of dimensional features of the data, and x i , x′ i represents the value of data points x and x′ in the i-th dimension,
[0022] The kernel function method is used to map the data into a high-dimensional space and capture the nonlinear data structure to predict the behavior of soft soil.
[0023] As a preferred solution of the soft soil parameter value analysis and correction method based on machine learning described in the present invention, the mixed integer nonlinear programming method is used to train the model for soft soil parameter analysis and correction, including combining linear or nonlinear optimization with integer programming technology, processing discrete and continuous decision variables in the model at the same time, optimizing the structure and parameter selection of the model, adapting to different soil types and environmental conditions, defining the optimization objectives and constraints of the model, and using the algorithm to output the MINLP problem and model parameters and structure:
[0024]
[0025]
[0026] Where w represents the weight vector, b represents the bias term, ξ represents the slack variable vector, which allows data points to violate boundary conditions, C represents the regularization parameter, which controls the trade-off between model complexity and error, n represents the total number of data points, and ξ i represents the slack variable of the i-th data point, z indicates whether the data point is considered an outlier, and z i represents the binary state of the i-th data point, m represents the total number of data points, which is equal to n, k represents the maximum number of outliers allowed, and y i represents the label of the i-th data point, x iRepresents the feature vector of the i-th data point;
[0027] The model is trained through an optimization framework to handle classification and regression problems. During the model training process, the selection of variables and the constraints of the model are handled simultaneously to eliminate data interference and optimize the prediction of soft soil parameters.
[0028] As a preferred solution of the soft soil parameter value analysis and correction method based on machine learning described in the present invention, the training model for soft soil parameter analysis and correction by mixed integer nonlinear programming method includes describing the dynamic changes of model parameters over time by stochastic differential equations, introducing random processes to simulate the uncertainty and variability of field conditions, adjusting and optimizing parameters in real time, and dynamically adjusting model parameters by stochastic differential equations according to real-time monitoring data to adapt to environmental and data conditions, which is expressed as:
[0029] dθ t =f(θ t ,t)dt+g(θ t ,t)dW t
[0030] Where dθ t represents the change of model parameter θ at time t, θ t represents the model parameter vector at time t, f(θ t ,t) represents the drift function, describing the deterministic change of parameters over time, g(θ t ,t) represents the diffusion term function, describing the random change of parameters over time, dW t represents an incremental Wiener process, simulating random noise;
[0031] Stochastic differential equations are used to describe dynamic systems with random terms, enabling the model to respond to random changes in continuous time, allowing real-time adjustment of model parameters according to parameter changes in dynamic soft soil environments.
[0032] As a preferred solution of the method for analyzing and correcting soft soil parameter values based on machine learning described in the present invention, the training model for soft soil parameter analysis and correction by mixed integer nonlinear programming method includes using Bayesian network for probabilistic reasoning and decision-making process, expressing conditional dependency and joint probability distribution between variables, providing structured and probabilistic methods to deal with uncertainty in decision-making, continuously updating and improving the model based on evidence, constructing a Bayesian network model, inputting data, reasoning through the network, and outputting a decision path, which is expressed as:
[0033] Pr(θ t+1 ∣θt)=∫Pr(θ t+1 ∣θ t ,d)Pr(d|θ t )dd
[0034] Among them, Pr(θ t+1 ∣θ t ) represents the current parameter θt, and at the next time point t+1 the parameter θ t+1 The conditional probability, θ t+1 ,θ t represents the parameter vector of two consecutive time points, d represents the decision factor, the external or internal factor that affects the parameter change, Pr(d|θ t ) represents the probability of decision factor d given the current parameter θt;
[0035] The decision-making process is modeled using Bayesian networks to handle decision problems, make decisions based on the probabilistic outputs provided by the model, and continuously adjust the model in response to data and feedback to continuously improve soft soil parameters.
[0036] Another object of the present invention is to provide a soft soil parameter value analysis and correction system based on machine learning, which can solve the problems of low accuracy and prediction reliability of soft soil parameter analysis in current soft soil parameter analysis methods by preprocessing data based on geological engineering data and identifying and utilizing nonlinear correlations in soft soil behavior.
[0037] As a preferred solution of the soft soil parameter value analysis and correction system based on machine learning described in the present invention, it includes: an initialization module, a data preprocessing module, and a model training, verification and application module; the initialization module is used to collect geological engineering data; the data preprocessing module is used to weight and standardize the data through information entropy transformation; the model training, verification and application module is used to construct, verify and apply a machine learning model to perform soft soil parameter analysis and correction.
[0038] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of a soft soil parameter value analysis and correction method based on machine learning.
[0039] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a soft soil parameter value analysis and correction method based on machine learning.
[0040] The beneficial effects of the present invention are as follows: the soft soil parameter value analysis and correction method based on machine learning provided by the present invention collects geological engineering data, enhances the accuracy and reliability of the model in practical applications, preprocesses the data based on the geological engineering data and identifies and utilizes the nonlinear correlation in the soft soil behavior, improves the accuracy of the soft soil parameter analysis and the reliability of the prediction, and trains the model through the mixed integer nonlinear programming method to perform soft soil parameter analysis and correction, thereby enhancing the accuracy of the soft soil parameter prediction and the applicability of the model. The present invention achieves better results in terms of prediction accuracy, model applicability and reliability. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0042] Figure 1 An overall flow chart of a method for analyzing and correcting soft soil parameter values based on machine learning provided for the first embodiment of the present invention.
[0043] Figure 2 An overall flow chart of a soft soil parameter value analysis and correction system based on machine learning provided for the third embodiment of the present invention. DETAILED DESCRIPTION
[0044] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0045] Example 1
[0046] Reference Figure 1 , which is an embodiment of the present invention, provides a method for analyzing and correcting soft soil parameter values based on machine learning, comprising:
[0047] S1: Collect geological engineering data.
[0048] Going further, collecting geotechnical data, the following are the types of data that need to be collected:
[0049] 1. Soil physical and chemical properties
[0050] Density: The dry or wet density of the soil.
[0051] Moisture Content: The percentage of water content in a soil sample.
[0052] Particle size distribution: the proportion of soil particles of different sizes in a sample.
[0053] Plasticity index and liquid limit: used to describe the plastic properties of soil.
[0054] Organic matter content: the proportion of organic matter in the soil.
[0055] Chemical composition: such as calcium carbonate content, salt content, etc.
[0056] Shear strength: The ability of soil to resist shear deformation.
[0057] Compression coefficient: An indicator of the change in volume of soil under pressure.
[0058] 2. Soil environment data
[0059] Groundwater level: affects soil moisture and chemistry.
[0060] Temperature and humidity records: Environmental factors that influence soil behavior.
[0061] Historical settlement data: reflects the behavior of soil under heavy loads.
[0062] 3. Engineering related data
[0063] Loading history: The historical maximum loads and types of loads the soil has been subjected to.
[0064] Types of engineering structures: buildings, roads, bridges, etc., structures that may affect soil properties.
[0065] Construction process records: including soil treatment methods and change records during construction.
[0066] 4. Laboratory test data
[0067] Standard Penetration Test (SPT) results: reflect soil density and bearing capacity.
[0068] Compression test: Provides data such as compression modulus and preload.
[0069] Shear test: direct shear or triaxial shear test results.
[0070] 5. Geological survey data
[0071] Stratigraphic distribution: depth and distribution of soil layers.
[0072] Geological age: The geological historical context in which soils were formed.
[0073] Data Collection and Processing Methods
[0074] Field investigation: Soil sampling and preliminary testing carried out directly on site.
[0075] Laboratory Testing: Detailed physical and chemical testing is performed on the collected samples.
[0076] Historical data analysis: Study previous engineering cases and scientific research data.
[0077] Monitoring equipment: Install soil pressure gauges, water level gauges and other equipment for long-term monitoring.
[0078] S2: Data preprocessing based on geo-engineering data and identification and utilization of nonlinear correlations in soft soil behavior.
[0079] Furthermore, information entropy is used to perform a weighted transformation on the data, redistributing the weight according to the amount of information in each data point;
[0080] The conversion is based on the entropy process to represent the measure of uncertainty and randomness of information;
[0081] Output the information entropy of each sample point, adjust the weight of each sample point according to the entropy value, and preprocess the data set, which is expressed as:
[0082]
[0083] Where x represents the original input data point, x″ represents the transformed data point, n represents the total number of data points, and x i represents the value of the i-th data point, x j represents the value of the jth data point used in the summation process;
[0084] By preprocessing the data set, the amount of concentrated information determines the contribution of each point to the whole.
[0085] The information entropy is used to perform a weighted transformation on the data, redistributing the weight according to the amount of information in each data point. This transformation is based on the definition of entropy value, which is a measure of uncertainty and randomness of information.
[0086] By improving the homogeneity and expressiveness of data, the impact of data skew and noise is reduced, thus providing more accurate and consistent input for model training and analysis.
[0087] The data set is preprocessed by calculating the information entropy of each sample point and then adjusting the weight of each sample point according to its entropy value.
[0088] It should be noted that the original data is mapped to a high-dimensional feature space through a nonlinear kernel function. The kernel function used includes a combination of a polynomial and a radial basis kernel function. The kernel parameters of the polynomial degree and the γ parameter of the RBF are selected. The data is converted to a high-dimensional space through kernel technology, which is expressed as:
[0089]
[0090] Among them, K(x,x′) represents the kernel function, which is used to output the similarity between data points x and x′, x, x′ represent the two data points compared in the kernel function, and θ i represents the weight parameter of the i-th dimension, p i represents the power of the polynomial kernel of the i-th dimension, γ i represents the scale parameter of the radial basis function RBF kernel of the i-th dimension, d represents the number of dimensional features of the data, and x i , x′ i represents the value of data points x and x′ in the i-th dimension,
[0091] The kernel function method is used to map the data into a high-dimensional space and capture the nonlinear data structure to predict the behavior of soft soil.
[0092] The original data is mapped to a high-dimensional feature space through a nonlinear kernel function. The kernel function used includes a combination of polynomial and radial basis kernel functions, which enables the model to reveal the intrinsic structure and complex relationships of the data.
[0093] Enhanced nonlinear processing capabilities of the model enable it to identify and exploit complex patterns and correlations hidden in the data, which is particularly critical for soft soil parameters in complex geological structures.
[0094] Select appropriate kernel parameters (such as polynomial degree and γ parameter of RBF) and use kernel technology to transform data into a high-dimensional space for subsequent model learning.
[0095] S3: Soft soil parameter analysis and correction are performed by training the model using mixed integer nonlinear programming method.
[0096] Furthermore, we combine linear or nonlinear optimization with integer programming techniques, process both discrete and continuous decision variables in the model, optimize the model structure and parameter selection, adapt to different soil and environmental conditions, define the optimization objectives and constraints of the model, and use the algorithm to output the MINLP problem and model parameters and structure:
[0097]
[0098]
[0099] Where w represents the weight vector, b represents the bias term, ξ represents the slack variable vector, which allows data points to violate boundary conditions, C represents the regularization parameter, which controls the trade-off between model complexity and error, n represents the total number of data points, and ξ i represents the slack variable of the i-th data point, z indicates whether the data point is considered an outlier, and z irepresents the binary state of the i-th data point, m represents the total number of data points, which is equal to n, k represents the maximum number of outliers allowed, and y i represents the label of the i-th data point, x i Represents the feature vector of the i-th data point;
[0100] The model is trained through an optimization framework to handle classification and regression problems. During the model training process, the selection of variables and the constraints of the model are handled simultaneously to eliminate data interference and optimize the prediction of soft soil parameters.
[0101] Combining linear or nonlinear optimization with integer programming techniques allows for simultaneous processing of both discrete and continuous decision variables in the model, making it suitable for complex optimization problems.
[0102] Optimize the model structure and parameter selection to ensure the optimality of the model in multiple dimensions (accuracy, complexity, and anti-interference ability) and adapt to different soil types and environmental conditions.
[0103] Define the optimization objectives and constraints of the model, use the algorithm to solve the MINLP problem, and find the optimal model parameters and structure.
[0104] It should be noted that the stochastic differential equations are used to describe the dynamic changes of model parameters over time, random processes are introduced to simulate the uncertainty and variability of field conditions, and parameters are adjusted and optimized in real time. According to real-time monitoring data, the model parameters are dynamically adjusted using stochastic differential equations to adapt to environmental and data conditions, which can be expressed as:
[0105] dθ t =f(θ t ,t)dt+g(θ t ,t)dW t
[0106] Where dθ t represents the change of model parameter θ at time t, θ t represents the model parameter vector at time t, f(θ t ,t) represents the drift function, describing the deterministic change of parameters over time, g(θ t ,t) represents the diffusion term function, describing the random change of parameters over time, dW t represents an incremental Wiener process, simulating random noise;
[0107] Stochastic differential equations are used to describe dynamic systems with random terms, enabling the model to respond to random changes in continuous time, allowing real-time adjustment of model parameters according to parameter changes in dynamic soft soil environments.
[0108] The temporal dynamics of model parameters are described by stochastic differential equations, and random processes are introduced to simulate the uncertainty and variability of field conditions.
[0109] This enables the model to adjust and optimize its parameters in real time, improving the adaptability and accuracy of predictions, especially for soft soil environments with large changes in geological conditions.
[0110] Based on real-time monitoring data, stochastic differential equations are used to dynamically adjust model parameters to adapt to new environmental and data conditions.
[0111] It should also be noted that the Bayesian network is used for probabilistic reasoning and decision-making processes, expressing the conditional dependencies and joint probability distributions between variables, providing a structured and probabilistic approach to deal with uncertainty in decision-making, and continuously updating and improving the model based on evidence. The Bayesian network model is constructed, data is input, reasoning is performed through the network, and the decision path is output, which is expressed as:
[0112] Pr(θ t+1 ∣θ t )=∫Pr(θ t+1 ∣θ t ,d)Pr(d|θ t )dd
[0113] Among them, Pr(θ t+1 ∣θ t ) represents the current parameter θt, and at the next time point t+1 the parameter θ t+1 The conditional probability, θ t+1 ,θ t represents the parameter vector of two consecutive time points, d represents the decision factor, the external or internal factor that affects the parameter change, Pr(d|θ t ) represents the probability of decision factor d given the current parameter θt;
[0114] The decision-making process is modeled using Bayesian networks to handle decision problems, make decisions based on the probabilistic outputs provided by the model, and continuously adjust the model in response to data and feedback to continuously improve soft soil parameters.
[0115] Bayesian networks are used for probabilistic reasoning and decision-making. The model can express conditional dependencies and joint probability distributions among multiple variables.
[0116] Providing a structured and probabilistic approach to dealing with uncertainty in decision making, allowing models to be continually updated and improved based on new evidence.
[0117] A Bayesian network model is established, relevant data and prior knowledge are input, reasoning is performed through the network, and the optimal and most likely decision path is output.
[0118] Example 2
[0119] An embodiment of the present invention provides a method for analyzing and correcting soft soil parameter values based on machine learning. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.
[0120] First, the experiment was divided into two groups: the experimental group adopted the method of the present invention, and the control group used the existing geological data processing technology to perform soft soil parameter analysis.
[0121] First, geotechnical data from the same location were collected for the two experiments, including soil physical and chemical properties (such as density, water content, particle size distribution, etc.), soil environmental data (such as groundwater level and temperature), engineering data (including load history and construction process records), laboratory test data (standard penetration test, compression test, shear test results) and geological survey data (stratum distribution and geological age). These data were standardized to ensure the consistency and comparability of the input data.
[0122] For the experimental group, advanced data preprocessing was performed based on the collected geotechnical data, including weighted transformation of the data using information entropy. The quality of the dataset was optimized by calculating the information entropy of each data point and adjusting the weight of each sample point according to its entropy value. Furthermore, a nonlinear kernel function (a combination of polynomial and radial basis function) was used to map the data into a high-dimensional feature space to capture the nonlinear correlations in the soft soil behavior.
[0123] The model was then trained using a mixed integer nonlinear programming (MINLP) approach that simultaneously processed both discrete and continuous decision variables in the model, optimizing the model structure and parameter selection. In addition, stochastic differential equations were applied to the experimental group model during training to dynamically adjust model parameters to accommodate the uncertainty and variability of field conditions.
[0124] The control group used the traditional linear regression model to process the same data set without performing information entropy weighted transformation or using nonlinear kernel mapping, and only relied on basic statistical methods to predict and analyze soft soil parameters.
[0125] Table 1 Experimental results
[0126]
[0127] From the above table data, it can be clearly seen that the prediction error of the experimental group after each stage of processing is significantly reduced compared with the control group. The experimental group has reduced the prediction error from 12% of the control group to 8% through information entropy weighted data preprocessing, which proves that the weighted transformation of data through information entropy optimizes data quality and enhances the model's ability to capture key information in the data. After further mapping to high-dimensional space through nonlinear kernel functions, the error is further reduced to 6%, demonstrating the importance of identifying nonlinear relationships for improving the accuracy of model predictions. Finally, after optimizing the model through the mixed integer nonlinear programming method, the error is reduced to the lowest 4%. The optimization at this stage not only considers the structure and parameter selection of the model, but also improves the adaptability and robustness of the model by processing discrete and continuous variables.
[0128] In summary, the present invention achieves better results in terms of prediction accuracy, model applicability and reliability.
[0129] Example 3
[0130] Reference Figure 2 , which is an embodiment of the present invention, provides a soft soil parameter value analysis and correction system based on machine learning, including an initialization module, a data preprocessing module, and a model training, verification and application module.
[0131] The initialization module is used to collect geological engineering data, the data preprocessing module is used to weight and standardize the data through information entropy transformation, and the model training, verification and application module is used to build, verify and apply machine learning models for soft soil parameter analysis and correction.
[0132] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0133] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0134] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.
[0135] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.
[0136] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for analyzing and correcting soft soil parameter values based on machine learning, characterized in that: include: Collect geological engineering data; Data preprocessing based on geo-engineering data and identification and utilization of nonlinear correlations in soft soil behavior; The model is trained by mixed integer nonlinear programming method for soft soil parameter analysis and correction.
2. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 1, characterized in that: The collecting of geological engineering data includes collecting soil physical and chemical properties, soil environmental data, engineering data, laboratory test data and geological survey data; The soil physical and chemical properties include density, water content, particle size distribution, plasticity index and liquid limit, organic matter content, chemical composition, shear strength and compressibility; The soil environmental data include groundwater level, temperature and humidity records and historical settlement data; The engineering data includes load history, engineering structure type and construction process records; The laboratory test data include standard penetration test results, compression tests, and shear tests; The geological survey data include stratigraphic distribution and geological age.
3. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 2, characterized in that: The data preprocessing based on the geological engineering data and identifying and utilizing the nonlinear correlation in the soft soil behavior includes weighting the data using information entropy and redistributing the weight according to the information content of each data point; The conversion is based on the entropy process to represent the measure of uncertainty and randomness of information; Output the information entropy of each sample point, adjust the weight of each sample point according to the entropy value, and preprocess the data set, which is expressed as: Where x represents the original input data point, x″ represents the transformed data point, n represents the total number of data points, and x i represents the value of the i-th data point, x j represents the value of the jth data point used in the summation process; By preprocessing the data set, the amount of concentrated information determines the contribution of each point to the whole.
4. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 3, characterized in that: The data preprocessing based on geological engineering data and identifying and utilizing the nonlinear correlation in the soft soil behavior include mapping the original data to a high-dimensional feature space through a nonlinear kernel function, the kernel function used includes a combination of a polynomial and a radial basis kernel function, selecting the kernel parameters of the polynomial degree and the γ parameter of the RBF, and converting the data to a high-dimensional space through a kernel technique, which is expressed as: Among them, K(x,x′) represents the kernel function, which is used to output the similarity between data points x and x′, x, x′ represent the two data points compared in the kernel function, and θ i represents the weight parameter of the i-th dimension, p i represents the power of the polynomial kernel of the i-th dimension, γ i represents the scale parameter of the radial basis function RBF kernel of the i-th dimension, d represents the number of dimensional features of the data, and x i , x′ i represents the value of data points x and x′ in the i-th dimension, The kernel function method is used to map the data into a high-dimensional space and capture the nonlinear data structure to predict the behavior of soft soil.
5. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 4, characterized in that: The mixed integer nonlinear programming method is used to train the model for soft soil parameter analysis and correction, which includes combining linear or nonlinear optimization with integer programming techniques, processing discrete and continuous decision variables in the model at the same time, optimizing the structure and parameter selection of the model, adapting to different soil types and environmental conditions, defining the optimization objectives and constraints of the model, and using the algorithm to output the MINLP problem and model parameters and structure: Where w represents the weight vector, b represents the bias term, ξ represents the slack variable vector, which allows data points to violate boundary conditions, C represents the regularization parameter, which controls the trade-off between model complexity and error, n represents the total number of data points, and ξ i represents the slack variable of the i-th data point, z indicates whether the data point is considered an outlier, and z i represents the binary state of the i-th data point, m represents the total number of data points, which is equal to n, k represents the maximum number of outliers allowed, and y i represents the label of the i-th data point, x i Represents the feature vector of the i-th data point; The model is trained through an optimization framework to handle classification and regression problems. During the model training process, the selection of variables and the constraints of the model are handled simultaneously to eliminate data interference and optimize the prediction of soft soil parameters.
6. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 5, characterized in that: The mixed integer nonlinear programming method is used to train the model for soft soil parameter analysis and correction, including describing the dynamic changes of model parameters over time by stochastic differential equations, introducing random processes to simulate the uncertainty and variability of field conditions, adjusting and optimizing parameters in real time, and dynamically adjusting model parameters by stochastic differential equations according to real-time monitoring data to adapt to environmental and data conditions, which is expressed as: dth t =f(θ t ,t)dt+g(θ t ,t)dW t Where dθ t represents the change of model parameter θ at time t, θ t represents the model parameter vector at time t, f(θ t ,t) represents the drift function, describing the deterministic change of parameters over time, g(θ t ,t) represents the diffusion term function, describing the random change of parameters over time, dW t represents an incremental Wiener process, simulating random noise; Stochastic differential equations are used to describe dynamic systems with random terms, enabling the model to respond to random changes in continuous time, allowing real-time adjustment of model parameters according to parameter changes in dynamic soft soil environments.
7. The method for analyzing and correcting soft soil parameter values based on machine learning according to claim 6, characterized in that: The mixed integer nonlinear programming method is used to train the model for soft soil parameter analysis and correction, including using Bayesian network for probabilistic reasoning and decision-making process, expressing the conditional dependency and joint probability distribution between variables, providing a structured and probabilistic method to deal with uncertainty in decision-making, continuously updating and improving the model based on evidence, building a Bayesian network model, inputting data, reasoning through the network, and outputting a decision path, which is expressed as: Pr(θ t+1 |θ t )=∫Pr(θ t+1 |θ t ,d)Pr(d|θ t )dd Among them, Pr(θ t+1 ∣θ t ) represents the current parameter θt, and at the next time point t+1 the parameter θ t+1 The conditional probability, θ t+1 ,θ t represents the parameter vector of two consecutive time points, d represents the decision factor, the external or internal factor that affects the parameter change, Pr(d|θ t ) represents the probability of decision factor d given the current parameter θt; The decision-making process is modeled using Bayesian networks to handle decision problems, make decisions based on the probabilistic outputs provided by the model, and continuously adjust the model in response to data and feedback to continuously improve soft soil parameters.
8. A system using the soft soil parameter value analysis and correction method based on machine learning as claimed in any one of claims 1 to 7, characterized in that: Includes initialization module, data preprocessing module, model training, verification and application module; The initialization module is used to collect geological engineering data; The data preprocessing module is used to weight and standardize the data through information entropy transformation; The model training, verification and application module is used to build, verify and apply machine learning models to perform soft soil parameter analysis and correction.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the soft soil parameter value analysis and correction method based on machine learning described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the soft soil parameter value analysis and correction method based on machine learning described in any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Sandy soil liquefaction discrimination method based on CPTU and weighted nonlinear similarity
CN120974283A
