Method and device for predicting corrosion rate of oil casing in formate completion fluid based on Bayesian optimization strategy
By combining Bayesian optimization and XGBoost algorithm, the problem of low corrosion rate prediction accuracy in oil casing in formate completion fluid is solved, and efficient and accurate corrosion rate prediction is achieved, reducing experimental costs.
Patent Information
- Application Number
- CN202510144996.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-10
- Publication Date
- 2025-06-06
AI Technical Summary
The prior art is difficult to effectively predict the corrosion rate of oil casing in formate completion fluid, especially under complex and variable working conditions, with low prediction accuracy and high experimental cost.
Using Bayesian optimization and XGBoost algorithm methods, we intelligently select experimental points through Bayesian optimization strategy, reduce the number of experiments, and use the XGBoost algorithm to build a corrosion rate prediction model to improve prediction accuracy.
It significantly improves the prediction accuracy of the corrosion rate of oil casing in formate completion fluid, reduces the number of experiments and costs, and can more effectively deal with complex working conditions.
Smart Images
Figure CN120105360A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a corrosion rate prediction technology, in particular to the prediction of the corrosion rates of three oil casing materials, P110, 13Cr and S13Cr, in formate completion fluid in the oil and gas industry, and specifically to a method and device for predicting the corrosion rate of oil casing in formate completion fluid based on a Bayesian optimization strategy. Background Art
[0002] With the rapid development of the oil and gas industry, oil and gas pipelines are facing corrosion threats from various factors, especially in the formate completion fluid environment, the corrosion problem is more prominent. Formate liquid has strong acidity and can accelerate the corrosion of oil casing materials (such as P110SS, 13Cr, S13Cr, etc.). However, there is currently no mature systematic prediction model for the prediction of oil casing corrosion rate in formate completion fluid, and related research in this field is still blank.
[0003] Existing corrosion rate prediction methods mainly rely on experimental data and empirical formulas, especially for CO 2 and H 2 There have been some achievements in the corrosion caused by gases such as S, but these models can usually only predict for a single material or specific environmental conditions, and have low accuracy and poor generalization ability. Traditional corrosion prediction methods require a large amount of experimental data support. The experimental process is not only time-consuming and costly, but also difficult to cover all possible operating conditions. Therefore, the existing technology is difficult to meet the needs of efficient and accurate prediction of corrosion rate in formate completion fluids in actual engineering.
[0004] Machine learning and data mining methods have been initially applied in corrosion rate prediction in recent years. Some studies have shown that machine learning-based models can improve prediction accuracy through historical data training, but research on formate completion fluid corrosion is still relatively lacking, and existing models generally have problems such as insufficient accuracy and limited scope of application.
[0005] In addition, methods such as Bayesian optimization have made significant progress in optimizing experimental design and improving model accuracy. Bayesian optimization optimizes the experimental parameter space through the Gaussian process regression model, and active learning reduces the number of experiments and improves model performance by selecting the most uncertain experimental points. The successful application of these methods in other fields has shown that combining Bayesian optimization can significantly reduce the need for experiments and improve prediction accuracy in corrosion rate prediction. However, these methods have not been fully applied in the prediction of corrosion rates in formate completion fluids.
[0006] In summary, although the existing corrosion prediction methods have improved the prediction accuracy of corrosion rate to a certain extent, the systematic prediction model for the corrosion rate of oil casing in formate completion fluid is still a difficult problem to be solved. Most of the existing technologies rely on a large amount of experimental data and are difficult to cope with complex and changing working conditions. Therefore, there is an urgent need for a new and efficient prediction method that can reduce the number of experiments and improve the prediction accuracy to meet the needs of oil and gas pipeline safety management. Summary of the invention
[0007] In order to solve the above technical problems, the present invention provides a method and device for predicting the corrosion rate of oil casing in formate completion fluid based on Bayesian optimization and XGBoost algorithm. Through the Bayesian optimization strategy, the experimental points are intelligently selected and the number of experiments is reduced, so as to achieve the effect of improving the accuracy of corrosion rate prediction.
[0008] Specifically, the present invention includes the following key steps:
[0009] 1. Experimental data collection and preprocessing:
[0010] Aiming at the typical service environment of oil casing materials (P110SS, 13Cr, S13Cr), a high temperature and high pressure reactor was used to simulate the on-site working conditions (including temperature, formate density, CO 2 Partial pressure, H 2 S partial pressure, etc.), conduct corrosion simulation experiments to obtain corrosion rate data under initial conditions. Standardize the data to improve the accuracy and stability of the model.
[0011] 2. XGBoost algorithm modeling:
[0012] The corrosion rate prediction model is constructed using the XGBoost algorithm. XGBoost is a powerful gradient boosting tree algorithm that optimizes the prediction accuracy of the model by learning the relationship between the features in the experimental data and the corrosion rate labels.
[0013] 3. Bayesian Optimization and Gaussian Process Regression:
[0014] The relationship between experimental parameters and corrosion rate is modeled using a Bayesian optimization algorithm. Specifically, the corrosion rate is predicted by a Gaussian process regression model, and the mean function μ(x) and standard deviation σ(x) of the model are calculated, where σ(x) reflects the prediction uncertainty of the model for the experimental point. Based on the expected improvement (EI) criterion of Bayesian optimization, the experimental point with the largest prediction uncertainty is selected to maximize the improvement of the model.
[0015] 4. Experiment and data feedback:
[0016] According to the experimental points selected by the Bayesian optimization algorithm, corrosion simulation experiments were carried out using a high-temperature and high-pressure reactor to obtain corrosion rate values. The experimental results were fed back to the corrosion rate prediction model and the Bayesian optimization model to retrain and update the model. Through this iterative process, the model accuracy was continuously optimized and the best prediction effect was ensured under limited experimental conditions.
[0017] 5. Corrosion rate prediction:
[0018] By using the trained XGBoost algorithm corrosion rate prediction model and inputting the actual operating parameters of the oil and gas field, the corrosion rate of the oil casing is predicted, providing a scientific basis for the production of the oil and gas field.
[0019] In order to achieve the above-mentioned invention object, the technical solution provided by the present invention is as follows:
[0020] 1. Experimental data collection and preprocessing: Through the experimental data acquisition system, data of different temperatures, formate density, CO 2 Partial pressure and H 2 Corrosion rate data under S partial pressure conditions. The data is standardized, and the specific standardization formula is:
[0021]
[0022] Among them, x′ is the standardized data, x is the original data, μ is the mean, and σ is the standard deviation.
[0023] 2. XGBoost algorithm modeling: Use the XGBoost algorithm to train the corrosion rate prediction model, learn through the characteristics of training samples and corrosion rate labels, and optimize the accuracy and generalization ability of the model.
[0024] 3. Bayesian optimization and Gaussian process regression: Gaussian process regression (GP) was used as a proxy model for Bayesian optimization to establish experimental parameters (temperature, formate density, CO 2 Partial pressure, H 2 The relationship between S partial pressure) and corrosion rate. The specific Gaussian process regression model is as follows:
[0025] f(x)=μ(x)+∈(x)
[0026] Among them, f(x) is the model's predicted value of the corrosion rate, μ(x) is the mean function of the Gaussian process, ε(x) is the noise term of the Gaussian process, and x is the experimental parameter.
[0027] The expected improvement (EI) criterion of Bayesian optimization is used to select the next optimal experimental point to maximize the prediction improvement:
[0028] EI(x)=σ(x)
[0029] Among them, σ(x) is the prediction standard deviation of the Gaussian process regression model, which reflects the prediction uncertainty of the model for the experimental point x.
[0030] 4. Experiment and data feedback: Based on the experimental points selected by the Bayesian optimization algorithm, corrosion simulation experiments are carried out using a high-temperature and high-pressure reactor to obtain corrosion rate values. The experimental results are fed back to the corrosion rate prediction model and the Bayesian optimization model to retrain and update the model. Through this iterative process, the model accuracy is continuously optimized and the best prediction effect is ensured under limited experimental conditions.
[0031] 5. Corrosion rate prediction:
[0032] By using the trained XGBoost algorithm corrosion rate prediction model and inputting the actual operating parameters of the oil and gas field, the corrosion rate of the oil casing is predicted, providing a scientific basis for the production of the oil and gas field.
[0033] The present invention also provides a device for predicting the corrosion rate of oil casing in formate completion fluid based on Bayesian optimization, comprising:
[0034] a. Data acquisition module, used to collect experimental data;
[0035] b. Data preprocessing module, used to standardize or normalize the collected experimental data to ensure that the scales of different features are consistent;
[0036] c. XGBoost model module, used to build a corrosion rate prediction model based on the XGBoost algorithm;
[0037] d. A Bayesian optimization module, which is used to model the relationship between experimental parameters and corrosion rate through a Gaussian process regression model and select experimental points with maximum prediction uncertainty for experiments. The prediction standard deviation σ(x) of the Gaussian process regression model is used to measure uncertainty.
[0038] e. Model updating module, used to update the corrosion rate prediction model according to the experimental results and continuously optimize the accuracy of corrosion rate prediction;
[0039] f. Prediction output module, used to predict the corrosion rate of different oil casing materials based on the trained XGBoost model.
[0040] The present invention further provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions enable a processor to execute the steps of the oil casing corrosion rate prediction method.
[0041] The present invention integrates the advantages of Bayesian optimization and XGBoost algorithm, which can not only optimize the experimental design efficiently and reduce the number of experiments, but also significantly improve the accuracy of corrosion rate prediction. Bayesian optimization can reduce the number of experiments by selecting the most uncertain experimental points, while the XGBoost algorithm can improve the accuracy of the prediction model through efficient machine learning. Therefore, the present invention provides an efficient and accurate corrosion rate prediction method, which can provide an important reference for wellbore integrity management in oil and gas fields, and has significant economic value and practical application prospects. . BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a flow chart of a method for predicting the corrosion rate of oil casing in formate completion fluid based on Bayesian optimization strategy provided by an embodiment of this specification;
[0043] Figure 2 It is an overall flow chart of a method for predicting the corrosion rate of oil casing in formate completion fluid based on a Bayesian optimization strategy provided by an embodiment of this specification;
[0044] Figure 3 This is a comparison chart between predicted samples and actual samples provided by an embodiment of this specification. DETAILED DESCRIPTION
[0045] This embodiment further illustrates the present invention in conjunction with the accompanying drawings, but the present invention is not limited to the following embodiments. The prediction model is trained by experimental data, and the experimental points are selected in combination with Bayesian optimization, and the optimization model is continuously iterated, so as to achieve high-precision prediction of the corrosion rate of oil casing.
[0046] like Figure 1 As shown, this embodiment provides a corrosion rate prediction method based on the combination of XGBoost algorithm and Bayesian optimization algorithm. The specific steps are as follows:
[0047] Step S101: Experimental data collection and preprocessing
[0048] This step includes collecting experimental data and preprocessing it to ensure that the data can be used for model training.
[0049] 1) Collect experimental data, including temperature, pressure, concentration of corrosive media, CO 2 Partial pressure, H 2 Parameters such as S partial pressure and corresponding corrosion rate data.
[0050] 2) Standardize the collected experimental data to ensure that the scales of different features are consistent. The standardization formula is as follows:
[0051]
[0052] Among them, x′ is the standardized data, x is the original data, μ is the mean, and σ is the standard deviation.
[0053] 3) Remove outliers through data cleaning to ensure data quality.
[0054] Step S102: Constructing a corrosion rate prediction model
[0055] In this step, the XGBoost (eXtreme Gradient Boosting) algorithm is used to build a corrosion rate prediction model using experimental data.
[0056] XGBoost is a gradient boosting decision tree algorithm that can handle high-dimensional data, avoid overfitting and train efficiently.
[0057] 1) The temperature, pressure, corrosive medium concentration, CO 2 Partial pressure, H 2 Parameters such as S partial pressure are used as input features, and corrosion rate is used as the target label and input into the XGBoost model for training to learn the relationship between experimental parameters and corrosion rate.
[0058] 2) Optimize model performance by adjusting model hyperparameters (such as learning rate, maximum depth, etc.) to ensure that the prediction accuracy meets the expected standards.
[0059] Step S103: Modeling using Bayesian optimization algorithm
[0060] In this step, the relationship between the experimental parameters and the corrosion rate is modeled through a Gaussian process regression model using a Bayesian optimization algorithm.
[0061] 1) Gaussian process regression (GP) was used as a proxy model for Bayesian optimization to establish the experimental parameters (temperature, formate density, CO 2 Partial pressure, H 2 The relationship between S partial pressure) and corrosion rate. The specific Gaussian process regression model is as follows:
[0062] f(x)=μ(x)+∈(x)
[0063] Among them, f(x) is the model's predicted value of the corrosion rate, μ(x) is the mean function of the Gaussian process, ε(x) is the noise term of the Gaussian process, and x is the experimental parameter.
[0064] 2) In order to maximize the model improvement effect, the expected improvement (EI) criterion is used to select the next experimental point, that is, to select the experimental point with the maximum prediction uncertainty. The expected improvement criterion is as follows:
[0065] The expected improvement (EI) criterion of Bayesian optimization is used to select the next optimal experimental point to maximize the prediction improvement:
[0066] EI(x)=σ(x)
[0067] Among them, σ(x) is the prediction standard deviation of the Gaussian process regression model, which reflects the prediction uncertainty of the model for the experimental point x. The x when σ(x) is the maximum is the predicted value of the current best experimental point.
[0068] Step S104: Selecting experimental sites and conducting experiments
[0069] 1) Based on the Bayesian optimization algorithm, select the experimental point with the maximum uncertainty or maximum information gain and conduct the experiment. The specific steps are as follows:
[0070] Experimental sites were selected according to EI criteria;
[0071] At the selected experimental point, a high-temperature and high-pressure reactor corrosion simulation experiment is carried out to obtain the corresponding corrosion rate value;
[0072] During the experiment, control the experimental parameters, including temperature, CO 2 Partial pressure, H 2 Conditions such as S partial pressure and formate concentration;
[0073] 2) Feed the experimental results back to the XGBoost corrosion rate prediction model and Bayesian optimization model for retraining and updating.
[0074] 3) Feed the experimental results back to the XGBoost corrosion rate prediction model and Bayesian optimization model for iterative optimization. The specific steps are as follows:
[0075] Use the latest experimental data to update the XGBoost model and optimize the model parameters to make the model more predictive;
[0076] At the same time, the Gaussian process regression prediction of the Bayesian optimization model is updated to further reduce the prediction uncertainty;
[0077] 4) Through the feedback iteration process, the accuracy of the corrosion rate prediction model is continuously improved to ensure the best prediction effect under limited experimental conditions.
[0078] Step S105: Model application and prediction
[0079] After the model training is completed, the trained XGBoost corrosion rate prediction model is used to input the actual working parameters of the oil and gas field to predict the corrosion rate. The input parameters include: temperature, CO 2 Partial pressure, H 2S partial pressure, formate concentration, etc.; the output result is the predicted value of corrosion rate of oil casing.
[0080] In a specific application scenario, the method and device for constructing a prediction model for the corrosion rate of oil casing materials in formate completion fluid provided by the present invention can be effectively applied to the prediction of the corrosion rate of oil casing materials in a formate completion fluid environment. By integrating multiple algorithms, processing the experimental data of the target oil casing material, and combining temperature, formate density, CO 2 Partial pressure, H 2 The corresponding corrosion rate prediction value is obtained by using the working condition parameters such as S partial pressure, which solves the problem of insufficient accuracy of traditional corrosion rate prediction and greatly improves the prediction accuracy of corrosion rate of oil casing of different materials. The specific steps are as follows, and the overall flow chart is as follows Figure 2 As shown:
[0081] S1. Determine initial experimental parameters
[0082] According to the actual service environment of the oil field, collect relevant environmental data of the oil casing, including temperature range (for example, 60℃ to 220℃), formate density (for example, 1.25g / cm 3 Up to 1.5g / cm 3 ), CO 2 Partial pressure (e.g., 0 MPa to 10 MPa), H 2 S partial pressure (e.g., 0 kPa to 20 kPa).
[0083] To ensure the comprehensiveness and representativeness of the data, at least 25 sets of experimental data were collected as the initial data set. These data included temperature, pressure, H 2 S concentration, CO 2 The partial pressure and the corresponding corrosion rate. The following is an example of 25 sets of experimental data:
[0084] Table 1 Collected experimental parameters
[0085] Material Temperature,℃ Density of formate, g / cm3 CO2,MPa H2S, kPa Corrosion rate, mm / a P110SS 60 1.25 10 0 11.0524 P110SS 100 1.25 10 0 23.6055 P110SS 150 1.25 10 0 16.2675 P110SS 180 1.25 10 0 4.9589 P110SS 220 1.25 10 0 1.835 P110SS 60 1.5 10 0 0.9717 P110SS 100 1.5 10 0 21.0893 P110SS 150 1.5 10 0 22.8887 P110SS 180 1.5 10 0 7.5211 P110SS 220 1.5 10 0 7.6085 13Cr 100 1.25 10 0 5.0721 13Cr 150 1.25 10 0 25.6331 13Cr 180 1.25 10 0 2.7666 13Cr 220 1.25 10 0 1.3488 13Cr 60 1.5 10 0 4.1637 13Cr 100 1.5 10 0 9.5849 13Cr 150 1.5 10 0 21.977 S13Cr 100 1.5 10 0 0.1765 S13Cr 150 1.5 10 0 0.1503 S13Cr 180 1.5 10 0 4.2886 S13Cr 220 1.5 10 0 10.625 S13Cr 180 1.25 2 20 0.5613 S13Cr 180 1.25 10 20 0.9297 S13Cr 180 1.25 20 20 1.5299 S13Cr 60 1.25 2 0 0.022
[0086] S2. High temperature autoclave corrosion simulation experiment
[0087] According to the determined initial experimental parameters, a corrosion simulation experiment was carried out in a high temperature autoclave. The experimental steps include:
[0088] 1) Place the prepared oil casing sample into a high temperature and high pressure autoclave;
[0089] 2) Adding media that simulates the on-site service environment;
[0090] 3) Setting the experimental conditions, such as the experimental temperature of 150°C, the pressure of 20MPa, and the experimental time of 72 hours;
[0091] 4) After the experiment, take out the sample and obtain the corrosion rate data through weight loss method, electrochemical method and other means to form an initial data set.
[0092] S3. Data Standardization
[0093] The corrosion rate data and experimental parameters in the initial data set are standardized. The Z-score standardization method is used to process all data. The specific steps are as follows:
[0094] 1) For each data item (such as temperature, pressure, H 2 S concentration, CO 2 Concentration, etc.) and corrosion rate data to calculate their mean and standard deviation;
[0095] 2) Standardize by the following formula:
[0096]
[0097] Among them, Z 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. S4. Modeling of corrosion rate prediction model
[0098] Divide the standardized data into a training set and a validation set, for example, in a ratio of 70%:30%. Use the XGBoost algorithm for model training and set the following preliminary model parameters:
[0099] Learning rate: 0.1;
[0100] The maximum depth of the tree is 5;
[0101] During the training process, the model is verified through the validation set, and the mean square error (MSE) is used as an evaluation indicator to evaluate the prediction accuracy of the model. If the MSE of the model is greater than 0.05, Bayesian optimization is performed.
[0102] S5. Bayesian Optimization
[0103] To optimize the corrosion rate prediction model, define the objective function of Bayesian optimization, for example, minimize the mean square error of the prediction model. The Bayesian optimization algorithm is used to determine the optimal experimental parameters for the next round of high-temperature autoclave corrosion simulation experiments. The goal of Bayesian optimization is to select experimental conditions that minimize the model prediction error. For example, the optimal conditions for the next round of experiments are obtained through optimization:
[0104] Material: 13Cr;
[0105] Temperature: 180℃;
[0106] Pressure: 25MPa;
[0107] H 2S concentration: 120ppm;
[0108] CO 2 Concentration: 220ppm;
[0109] S6. Repeated experiments and model optimization
[0110] The next round of high-temperature autoclave corrosion simulation experiments was conducted based on the optimal experimental parameters determined by Bayesian optimization. Data standardization, model training, and Bayesian optimization were performed after each experiment until the prediction accuracy of the model met the requirements (for example, MSE ≤ 0.05). Through continuous optimization, the accuracy of the model was gradually improved.
[0111] S7. Corrosion rate prediction
[0112] After Bayesian optimization and multiple experimental verifications, an optimized corrosion rate prediction model was obtained. Figure 3 As shown in the figure, through this strategy, after continuous iterative training, the model prediction effect is greatly improved under a limited number of corrosion simulation test groups. Using this model, the corrosion rate of oil casing in the actual field service environment can be predicted. The actual field service environment parameters (such as temperature, pressure, medium composition, etc.) are input into the model to obtain the prediction results of the corrosion rate. For example, the annual corrosion rate of a certain oil casing under specific conditions is predicted to be 0.1mm / a.
[0113] Implementation effect: Through the above steps, the corrosion rate of oil casing in formate completion fluid can be accurately predicted, and the model can be dynamically optimized according to the experimental results, the number of experiments can be reduced, and the prediction accuracy can be improved. The method of the present invention combines Bayesian optimization with active learning strategy, and provides an efficient and accurate method for predicting oilfield corrosion rate by optimizing experimental parameters and reducing the amount of experiments.
Claims
1. A method for predicting the corrosion rate of oil casing in formate completion fluid based on Bayesian optimization strategy, comprising the following steps: S1, collect experimental parameters and perform standardization; S2. Construct a corrosion rate prediction model based on the XGBoost algorithm; S3, using a Bayesian optimization algorithm, modeling the relationship between the experimental parameters and the corrosion rate through a Gaussian process regression model, and selecting an experimental point with the maximum information gain according to a Bayesian optimization strategy; S4. According to the optimal experimental point determined in step S3, a corrosion simulation experiment is carried out using a high temperature and high pressure reactor to obtain a corrosion rate value, and the experimental results are fed back to the corrosion rate prediction model and the Bayesian optimization model, and the model is retrained and updated, and iterated; S5. The corrosion rate prediction model based on the trained XGBoost algorithm is used to input the oil and gas field field parameters to predict the corrosion rate value of the oil casing at the oil and gas field site.
2. The method for predicting the corrosion rate of oil casing according to claim 1, characterized in that: In step S1, the experimental parameters include temperature, formate density, CO2 partial pressure and H2S partial pressure.
3. The method for predicting the corrosion rate of oil casing according to claim 1 or 2, characterized in that: In step S1, the standardization process is performed according to the following formula: Among them, x′ is the standardized data, x is the original data, μ is the mean, and σ is the standard deviation.
4. The method for predicting the corrosion rate of oil casing according to any one of claims 1 to 3, characterized in that: In step S2, the corrosion rate prediction model adopts a regression task form and learns through the features of training samples and corrosion rate labels to optimize the accuracy and generalization ability of the model.
5. The method for predicting the corrosion rate of oil casing according to any one of claims 1 to 4, characterized in that: In step S3, the output of the Gaussian process regression model includes the mean function μ(x) and the prediction variance σ(x), and the specific model is expressed as: f(x)=μ(x)+∈(x) Among them, f(x) is the model's predicted value of the corrosion rate, μ(x) is the mean function of the Gaussian process, ε(x) is the noise term of the Gaussian process, and x is the experimental parameter.
6. The method for predicting the corrosion rate of oil casing according to any one of claims 1 to 5, characterized in that: In step S3, the expected improvement selection criterion of the Bayesian optimization strategy is: EI(x)=σ(x)where, σ(x) is the prediction standard deviation of the Gaussian process regression model, which reflects the prediction uncertainty of the model for the experimental point x.
7. The method for predicting the corrosion rate of oil casing according to any one of claims 1 to 6, characterized in that: In step S4, the Bayesian optimization strategy conducts the experiment by selecting the experimental point with the maximum prediction uncertainty, and the uncertainty measure is: Uncertainty(x)=σ(x) Among them, σ(x) is the standard deviation of the model for the experimental point x, reflecting the uncertainty of the prediction.
8. A device for predicting the corrosion rate of oil casing in formate completion fluid based on Bayesian optimization, comprising: a. Data acquisition module, used to collect experimental data; b. Data preprocessing module, used to standardize or normalize the collected experimental data to ensure that the scales of different features are consistent; c. XGBoost model module, used to build a corrosion rate prediction model based on the XGBoost algorithm; d. A Bayesian optimization module, which is used to model the relationship between experimental parameters and corrosion rate through a Gaussian process regression model and select experimental points with maximum prediction uncertainty for experiments. The prediction standard deviation σ(x) of the Gaussian process regression model is used to measure uncertainty. e. Model updating module, used to update the corrosion rate prediction model according to the experimental results and continuously optimize the accuracy of corrosion rate prediction; f. Prediction output module, used to predict the corrosion rate of different oil casing materials based on the trained XGBoost model.
9. A computer-readable storage medium, characterized in that: The storage medium stores computer instructions, which enable the processor to execute the steps of the method for predicting the corrosion rate of oil casing according to any one of claims 1 to 7.