A method and system for performance tuning of ultra-high temperature well cementing

Abstract

The application provides a performance optimization method and system of ultra-high temperature cementing cement, relates to the field of artificial intelligence, and comprises the following steps: acquiring performance influence characteristics and compressive strength sample data set of the ultra-high temperature cementing cement; taking the performance influence characteristics of the ultra-high temperature cementing cement as input characteristics, taking the compressive strength of the ultra-high temperature cementing cement as an output target, determining an optimal model from a set machine learning model library according to the influence coefficient of the input characteristics on the output target; determining the influence coefficient of each performance influence characteristic on the compressive strength; screening target characteristics with an influence coefficient greater than a set influence threshold, and generating an optimization scheme of the ultra-high temperature cementing cement according to the influence coefficient of each target characteristic. The application can optimize the mechanical properties of the ultra-high temperature cementing cement based on the optimization scheme, so as to ensure that the ultra-high temperature cementing cement can adapt to different application scenarios and meet the application requirements in normal temperature, high temperature, ultra-high temperature and other scenarios.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and in particular to a method and system for optimizing the performance of ultra-high temperature cementing. Background Technology

[0002] Currently, there are many factors affecting ultra-high temperature cementing, including but not limited to cement grade, water-cement ratio (also known as w / c ratio), calcium-silicon-aluminum ratio (also known as Ca / Si / Al ratio), sand addition, curing time and curing temperature, among which curing time and curing temperature include conventional curing temperature, conventional temperature curing time, high temperature curing temperature, and high temperature curing time.

[0003] The application performance and compressive strength requirements of ultra-high temperature cementing are high, making it difficult to determine the optimal design scheme for ultra-high temperature cementing under different application scenarios by combining many factors. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for performance optimization of ultra-high temperature cementing cement, which can effectively optimize the performance of ultra-high temperature cementing cement.

[0005] To address the aforementioned technical problems, this application provides a method for optimizing the performance of ultra-high temperature cementing cement, the specific technical solution of which is as follows:

[0006] Obtain the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement;

[0007] Determine the performance characteristics of ultra-high temperature cementing cement in the aforementioned compressive strength sample dataset;

[0008] Using the performance influence characteristics of ultra-high temperature cement as input features and the compressive strength of ultra-high temperature cement as output target, the optimal model is determined from a set machine learning model library based on the influence relationship between the input features and the output target.

[0009] Under the performance testing conditions of the optimal model, the influence coefficients of each of the ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing are determined.

[0010] Target features with influence coefficients greater than a set influence threshold are selected from the performance characteristics of ultra-high temperature cementing, and optimization schemes for ultra-high temperature cementing are generated based on the influence coefficients of each target feature.

[0011] Optionally, after obtaining the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement, the method further includes:

[0012] Identify the performance impact characteristics and the outliers in the compressive strength sample dataset;

[0013] After removing the abnormal data, a sample dataset of effective performance impact characteristics and compressive strength is obtained.

[0014] Optionally, identifying outliers in the performance impact characteristics and the compressive strength sample dataset includes:

[0015] Generate a marginal histogram between the input features and the output target;

[0016] Data with high dispersion in the marginal histogram are considered outliers.

[0017] Optionally, identifying outliers in the performance impact characteristics and the compressive strength sample dataset includes:

[0018] For any type of data in the compressive strength sample dataset, determine the first quartile and the third quartile;

[0019] The interquartile range is determined based on the first quartile and the third quartile.

[0020] The upper and lower limits of outliers are determined based on the interquartile range.

[0021] Abnormal data in the corresponding data type is removed based on the upper and lower limits.

[0022] Optionally, determining the optimal model from a set machine learning model library based on the influence relationship of the input features on the output target includes:

[0023] Determine the evaluation metrics for each machine learning model in the defined machine learning model library; the evaluation metrics include any combination of several of the following: coefficient of determination, mean absolute error, mean square error, and root mean square error.

[0024] The optimal model is determined by comparing the evaluation indicators using cross-validation.

[0025] Optionally, determining the influence coefficients of each of the ultra-high temperature cement performance characteristics on the compressive strength of the ultra-high temperature cement includes:

[0026] The influence coefficients of the performance characteristics of each ultra-high temperature cementing cement on the compressive strength of the ultra-high temperature cementing cement were determined using the Shapley additive interpretation analysis method.

[0027] Optionally, after determining the influence coefficients of each of the ultra-high temperature cement performance characteristics on the compressive strength of the ultra-high temperature cement, the method further includes:

[0028] Based on the influence coefficients of the performance characteristics of each ultra-high temperature cementing cement and the optimal model, a performance prediction model for the ultra-high temperature cementing cement is generated; the performance prediction model is used to predict the compressive strength of the ultra-high temperature cementing cement.

[0029] This application also provides a performance optimization system for ultra-high temperature cementing, including:

[0030] The data acquisition module is used to acquire the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement.

[0031] The feature determination module is used to determine the features affecting the performance of ultra-high temperature cementing in the compressive strength sample dataset;

[0032] The model determination module is used to take the performance influence characteristics of the ultra-high temperature cementing cement as input features, the compressive strength of the ultra-high temperature cementing cement as output target, and determine the optimal model from a set machine learning model library according to the influence relationship between the input features and the output target.

[0033] The coefficient calculation module is used to determine the influence coefficient of each of the ultra-high temperature cementing performance influence characteristics on the compressive strength of the ultra-high temperature cementing under the performance testing conditions of the optimal model.

[0034] The optimization module is used to screen the target features whose influence coefficients are greater than a set influence threshold among the performance influence features of the ultra-high temperature cementing cement, and generate an optimization scheme for the ultra-high temperature cementing cement based on the influence coefficients of each target feature.

[0035] Optionally, the model determination module includes:

[0036] The model evaluation unit is used to determine the evaluation metrics for each machine learning model in the defined machine learning model library; the evaluation metrics include any combination of several of the following: coefficient of determination, mean absolute error, mean square error, and root mean square error.

[0037] The model validation unit is used to compare the evaluation indicators using cross-validation to determine the optimal model.

[0038] Optionally, the performance tuning system may also include:

[0039] The performance prediction module is used to generate a performance prediction model for the ultra-high temperature cementing cement based on the influence coefficients of the performance characteristics of each ultra-high temperature cementing cement and the optimal model, and is used to predict the compressive strength of the ultra-high temperature cementing cement.

[0040] This application provides a method for performance optimization of ultra-high temperature cementing cement, comprising: acquiring performance influence features and compressive strength sample datasets of the ultra-high temperature cementing cement; determining the performance influence features of the ultra-high temperature cementing cement in the compressive strength sample dataset; using the performance influence features of the ultra-high temperature cementing cement as input features and the compressive strength of the ultra-high temperature cementing cement as the output target, determining the optimal model from a set machine learning model library according to the influence relationship between the input features and the output target; under the performance testing conditions of the optimal model, determining the influence coefficient of each performance influence feature of the ultra-high temperature cementing cement on the compressive strength of the ultra-high temperature cementing cement; screening target features among the performance influence features of the ultra-high temperature cementing cement whose influence coefficients are greater than a set influence threshold, and generating an optimization scheme for the ultra-high temperature cementing cement based on the influence coefficients of each target feature.

[0041] After obtaining a sample dataset of the compressive strength of ultra-high temperature cementing, this application identifies the various performance-influencing features of ultra-high temperature cementing that affect its performance. After determining the input features and output targets, it validates each machine learning model in a predefined machine learning model library to determine the optimal model. The optimal model is then applied to determine the influence coefficients of each performance-influencing feature, i.e., the ability of each feature to influence the compressive strength of the ultra-high temperature cementing. Finally, an optimization scheme is determined by comprehensively considering these features. Based on this optimization scheme, the mechanical properties of the ultra-high temperature cementing can be optimized, enabling the design of the cement slurry system. This ensures that the ultra-high temperature cementing can adapt to different application scenarios and meet application requirements in normal temperature, high temperature, and ultra-high temperature environments.

[0042] This application also provides a performance optimization system for ultra-high temperature cementing, which has the above-mentioned beneficial effects, and will not be elaborated here. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0044] Figure 1 A flowchart illustrating a method for performance optimization of ultra-high temperature cementing provided in this application embodiment;

[0045] Figure 2 This application provides a marginal histogram relating the performance characteristics and compressive strength of ultra-high temperature cementing materials.

[0046] Figure 3 This is a box plot showing the relationship between cement grade and compressive strength provided in the embodiments of this application;

[0047] Figure 4 This is a schematic diagram of the Pearson correlation coefficient and heatmap provided in the embodiments of this application;

[0048] Figure 5 A scatter plot showing the correlation between actual and predicted values ​​provided in the embodiments of this application;

[0049] Figure 6 This is a schematic diagram illustrating the feature importance analysis provided in the embodiments of this application;

[0050] Figure 7 A schematic diagram of the SHAP values ​​of the feature variables provided in the embodiments of this application;

[0051] Figure 8 This is a schematic diagram of a graphical user interface for predicting the mechanical properties of cement provided in the embodiments of this application;

[0052] Figure 9 This is a schematic diagram of a performance optimization system for ultra-high temperature cementing provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] The object information involved in this application, including but not limited to object device information, object personal information, and data, including but not limited to data used for analysis, stored data, and displayed data, are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of related data shall comply with the laws, regulations and standards of relevant countries and regions.

[0055] Figure 1 A flowchart of a method for performance optimization of ultra-high temperature cementing provided in this application embodiment, the method including:

[0056] S101: Obtain the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cementing cement;

[0057] S102: Using the performance influence characteristics of ultra-high temperature cement as input features, and the compressive strength of ultra-high temperature cement as output target, determine the optimal model from the set machine learning model library according to the influence relationship between the input features and the output target;

[0058] S103: Under the performance testing conditions of the optimal model, determine the influence coefficient of each of the ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing.

[0059] S104: Filter the target features whose influence coefficients are greater than the set influence threshold among the performance influence features of the ultra-high temperature cementing cement, and generate the optimization scheme of the ultra-high temperature cementing cement according to the influence coefficients of each target feature.

[0060] First, obtain the performance influence characteristics and compressive strength sample dataset of ultra-high temperature cement. There are no restrictions on how to obtain the compressive strength sample data; it can be obtained from databases or literature.

[0061] Subsequently, the performance characteristics of ultra-high temperature cementing cement in the compressive strength sample dataset were determined. Any characteristic that could affect the performance of ultra-high temperature cementing cement could be used as the performance characteristics of ultra-high temperature cementing cement in this application. These could include, but are not limited to, cement grade, water-cement ratio (w / c ratio), calcium-silicon-aluminum ratio (Ca / Si / Al ratio), sand addition, curing time, and curing temperature. After determining the performance characteristics of ultra-high temperature cementing cement, training and testing sets in the dataset can be further defined to facilitate statistical analysis of various machine learning models.

[0062] In one feasible implementation, after obtaining the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement, abnormal data in the performance influence characteristics and compressive strength sample dataset can be identified, thereby eliminating the abnormal data and obtaining a valid performance influence characteristics and compressive strength sample dataset. The abnormal data mentioned in this application includes data with obvious data anomalies and invalid data, etc.

[0063] Subsequently, the performance influence characteristics of ultra-high temperature cementing cement need to be used as input features, and the compressive strength of ultra-high temperature cementing cement needs to be used as output target, in order to perform feature analysis on the performance influence characteristics of ultra-high temperature cementing cement.

[0064] When performing feature analysis on the influence characteristics of ultra-high temperature cementing performance, methods such as marginal histograms, box plots, Pearson correlation coefficients, and heatmaps can be used. If a marginal histogram is used, a marginal histogram can be generated between the input features and the output target, thereby identifying data with high dispersion in the marginal histogram as outliers. See also... Figure 2 , Figure 2 This application provides a marginal histogram relating the performance characteristics and compressive strength of ultra-high temperature cementing materials. Figure 2 Taking the water-cement ratio (w / c) as an example, for other characteristics affecting the performance of ultra-high temperature cement, the following can be adopted: Figure 2 The method shown constructs a marginal histogram to identify outliers. For box plots, see [link to relevant documentation]. Figure 3 , Figure 3 The box plot between cement grade and compressive strength provided in this embodiment uses cement grade as the input feature and compressive strength as the output feature to determine the first and third quartiles; the interquartile range is determined based on the first and third quartiles; the upper and lower limits of outliers are determined based on the interquartile range; and finally, outliers in the corresponding data type are removed based on the upper and lower limits. See also... Figure 4 , Figure 4 This is a schematic diagram of the Pearson correlation coefficient and heatmap provided for an embodiment of this application.

[0065] In one feasible implementation of this application, the machine learning model library may include five algorithms: XGB (Extreme Gradient Boosting), HGBR (Histogram-based Gradient Boosting Regression), CB (Gradient Boosting Decision Tree Algorithm Optimized for Categorical Features), ANN (Artificial Neural Network), and TPE-XGB (Extreme Gradient Boosting Model Optimized by Tree-structured Parzen Estimator).

[0066] K-Fold cross-validation, as a statistical technique, is used to evaluate or estimate the performance of machine learning models. This application may employ the ten-fold cross-validation method.

[0067] Using the coefficient of determination R 2 The accuracy of the model is evaluated using the Coefficient of Determination, Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). Among these, R... 2 R is a statistical measure used to evaluate how well a model fits real-world data. MAE, MSE, and RMSE are metrics used to measure the model's predictions. Typically, R... 2 The closer the values ​​are to 1, the smaller the MAE, MSE, and RMSE, and the higher the prediction accuracy of the model.

[0068] Through R 2 A comparative analysis of the metrics MSE, MAE, and RMSE shows that the TPE-XGB model exhibits superior performance compared to the other four machine learning models. In the training set, the TPE-XGB model achieves a higher R-value.2 The value exceeded 0.95, and the MSE, MAE, and RMSE evaluation metrics were all superior to other algorithms. The test set R... 2 Reaching a score of 0.9359, closer to 1 than other algorithms, the MAE improved by 3.81%, 26.45%, 36.11%, and 46.88% compared to the other four machine learning algorithms, with improvements in RMSE of 2.90%, 21.34%, 29.92%, and 42.35%, respectively. Therefore, among the five machine learning algorithms mentioned above, the TPE-XGB model is the optimal one. Of course, under different influence characteristics of ultra-high temperature cementing performance, and with different machine learning models included in the machine learning model library, other models may also perform optimally.

[0069] See Figure 5 , Figure 5 The scatter plot illustrating the correlation between actual and predicted values ​​provided in this application embodiment demonstrates the correlation between actual and predicted values ​​for various machine learning models. The machine learning model TPE-XGB exhibits higher prediction accuracy and produces smaller errors: in the training set, the model's data points are closely aligned with the 45° ideal line, with prediction errors within 10%, indicating tight data clustering and highlighting excellent learning results. This demonstrates that the hybrid model TPE-XGB outperforms the other four machine learning models in generalization performance.

[0070] Subsequently, the influence coefficients of each of the aforementioned ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing can be determined using the Shapley Additive Explanations (SHAP) method. SHAP is a method for interpreting the output of machine learning models. It provides a global and local interpretation for complex machine learning models and is suitable for understanding and interpreting the importance and influence of model predictions.

[0071] See Figure 6 , Figure 6 This is a schematic diagram illustrating feature importance analysis provided in an embodiment of this application. Figure 6The average SAP values ​​of the characteristic variables are presented, reflecting their contribution to the prediction results. In the TPE-XGB model, cement grade, w / c, and curing time at normal temperature were identified as the most critical variables affecting the mechanical properties of cement, with average SAP values ​​of 6.55, 4.92, and 3.96, respectively. Among the chemical composition of cement, Al / Si and Ca / Si also significantly affect the mechanical properties of cement, with average SAP values ​​of 3.38 and 2.5, respectively. External factors such as normal temperature and curing time at high temperatures have relatively small effects. Overall, by adjusting these important parameters such as cement grade, w / c, Al / Si, and Ca / Si, it is possible to significantly improve cement performance.

[0072] The impact of feature variables is not only accurately quantified using the SHAP method, but the influence of feature values ​​on prediction results is also clearly demonstrated. Figure 7 The summary diagram. Figure 7 This is a schematic diagram of the SHAP values ​​of the feature variables provided in the embodiments of this application. Figure 7 In the diagram, each point corresponds to the SHAP value of a feature instance. The y-axis represents the feature, ranked from highest to lowest importance to the model's prediction. The x-axis represents the SHAP value. Furthermore, color variations indicate the magnitude of the feature values, with red areas representing high values ​​and blue areas representing low values. Fluctuations in the y-axis depict the distribution of feature values. For example, cement grade and compressive strength show a positive correlation, while w / c shows a negative correlation with compressive strength. The behavior of the input features to the machine learning model is consistent with existing knowledge, thus validating the authenticity of the TPE-XGB model.

[0073] There is no limitation on how to determine the set influence threshold. The influence threshold can be set in the same way as the calculation method of the influence coefficient, or the lowest value among the first set number of influence coefficients with higher influence coefficients in the influence characteristics of ultra-high temperature cementing performance can be taken as the influence threshold. For example, if the first three features with higher influence coefficients are taken, the third largest influence coefficient value can be taken as the influence threshold.

[0074] In other embodiments of this application, to improve the usability of machine learning models, a Python-based graphical user interface (GUI) was developed to predict the compressive strength of cement, such as... Figure 8 As shown, Figure 8This diagram illustrates a graphical user interface (GUI) for predicting the mechanical properties of cementing materials, as provided in this application embodiment. The GUI was developed using the Tkinter module in Python and incorporates the aforementioned optimal machine learning model, TPE-XGB, to enable prediction. The GUI allows users to input feature data, including the Ca / Si / Al ratio, curing time, and temperature, and then simply click the predict button to obtain predicted mechanical property data for the cement. The GUI design is simple, clear, and user-friendly, enabling non-technical users to easily operate and utilize the machine learning model. This user-friendly GUI for predicting cement compressive strength facilitates quick access to prediction results, thereby improving the usability of the machine learning model.

[0075] This embodiment, based on the optimal model determined in the above embodiments, further develops prediction software that can predict the mechanical properties and microstructures of cementing cement for normal temperature, high temperature, ultra-high temperature, and thermal recovery wells. Building upon the performance optimization method for ultra-high temperature cementing cement described above, this embodiment further uses a prediction optimization method to control the mechanical properties of ultra-high temperature cementing cement, achieving cement slurry system design to ensure that ultra-high temperature cementing cement can adapt to different application scenarios and meet application requirements in normal temperature, high temperature, and ultra-high temperature environments.

[0076] See Figure 9 , Figure 9 This is a schematic diagram of a performance tuning system for ultra-high temperature cementing provided in an embodiment of this application. This application also provides a performance tuning system for ultra-high temperature cementing, comprising:

[0077] The data acquisition module is used to acquire the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement.

[0078] The feature determination module is used to determine the features affecting the performance of ultra-high temperature cementing in the compressive strength sample dataset;

[0079] The model determination module is used to take the performance influence characteristics of the ultra-high temperature cementing cement as input features, the compressive strength of the ultra-high temperature cementing cement as output target, and determine the optimal model from a set machine learning model library according to the influence relationship between the input features and the output target.

[0080] The coefficient calculation module is used to determine the influence coefficient of each of the ultra-high temperature cementing performance influence characteristics on the compressive strength of the ultra-high temperature cementing under the performance testing conditions of the optimal model.

[0081] The optimization module is used to screen the target features whose influence coefficients are greater than a set influence threshold among the performance influence features of the ultra-high temperature cementing cement, and generate an optimization scheme for the ultra-high temperature cementing cement based on the influence coefficients of each target feature.

[0082] Based on the above embodiments, as a preferred embodiment, it further includes:

[0083] The data filtering module is used to identify abnormal data in the performance impact characteristics and the compressive strength sample dataset; and to remove the abnormal data to obtain the effective performance impact characteristics and compressive strength sample dataset.

[0084] Based on the above embodiments, as a preferred embodiment, the data filtering module includes:

[0085] A discrete data filtering unit is used to generate a marginal histogram between the input features and the output target; and to identify data with a high degree of dispersion in the marginal histogram as outlier data.

[0086] Based on the above embodiments, as a preferred embodiment, the data filtering module includes:

[0087] An abnormal data filtering unit is used to determine the first quartile and the third quartile for any type of data in the compressive strength sample dataset; determine the interquartile range based on the first quartile and the third quartile; determine the upper and lower limits of outliers based on the interquartile range; and remove abnormal data from the corresponding type of data based on the upper and lower limits.

[0088] Based on the above embodiments, as a preferred embodiment, the model determination module includes:

[0089] The model evaluation unit is used to determine the evaluation metrics for each machine learning model in the defined machine learning model library; the evaluation metrics include any combination of several of the following: coefficient of determination, mean absolute error, mean square error, and root mean square error.

[0090] The model validation unit is used to compare the evaluation indicators using cross-validation to determine the optimal model.

[0091] Based on the above embodiments, as a preferred embodiment, the coefficient calculation module is a module used to determine the influence coefficient of each of the ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing using the Shapley additive interpretation analysis method.

[0092] Based on the above embodiments, as a preferred embodiment, it further includes:

[0093] The performance prediction module is used to generate a performance prediction model for the ultra-high temperature cementing cement based on the influence coefficients of the performance characteristics of each ultra-high temperature cementing cement and the optimal model; the performance prediction model is used to predict the compressive strength of the ultra-high temperature cementing cement.

[0094] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. As the system provided in the embodiments corresponds to the method provided in the embodiments, the description is relatively simple; relevant parts can be found in the method section.

[0095] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

[0096] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for optimizing the performance of ultra-high temperature cementing, characterized in that, include: Obtain the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement; Using the performance influence characteristics of ultra-high temperature cement as input features and the compressive strength of ultra-high temperature cement as output target, the optimal model is determined from a set machine learning model library based on the influence relationship between the input features and the output target. Under the performance testing conditions of the optimal model, the influence coefficients of each of the ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing are determined; the ultra-high temperature cementing performance characteristics include cement grade, water-cement ratio, calcium-silicon-aluminum ratio, sand addition, curing time and curing temperature; Based on the influence coefficients of the performance characteristics of each ultra-high temperature cementing cement and the optimal model, a performance prediction model for the ultra-high temperature cementing cement is generated to predict the compressive strength of the ultra-high temperature cementing cement. Target features with influence coefficients greater than a set influence threshold are selected from the performance-influencing features of the ultra-high temperature cementing, and an optimization scheme for the ultra-high temperature cementing is generated based on the influence coefficients of each target feature. After obtaining the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement, the process also includes: Identify the performance impact characteristics and outliers in the compressive strength sample dataset; After removing the abnormal data, a sample dataset of effective performance impact characteristics and compressive strength is obtained; The identification of outliers in the performance impact characteristics and the compressive strength sample dataset includes: For any type of data in the compressive strength sample dataset, determine the first quartile and the third quartile; The interquartile range is determined based on the first quartile and the third quartile. The upper and lower limits of outliers are determined based on the interquartile range. Abnormal data in the corresponding data type is removed based on the upper and lower limits.

2. The method for performance optimization of ultra-high temperature cementing cement according to claim 1, characterized in that, Determining the optimal model from a set machine learning model library based on the influence relationship between the input features and the output target includes: Determine the evaluation metrics for each machine learning model in the defined machine learning model library; the evaluation metrics include any combination of several of the following: coefficient of determination, mean absolute error, mean square error, and root mean square error. The optimal model is determined by comparing the evaluation indicators using cross-validation.

3. The method for performance optimization of ultra-high temperature cementing according to claim 1, characterized in that, The influence coefficients of each of the aforementioned ultra-high temperature cementing performance characteristics on the compressive strength of the ultra-high temperature cementing include: The influence coefficients of the performance characteristics of each ultra-high temperature cementing cement on the compressive strength of the ultra-high temperature cementing cement were determined using the Shapley additive interpretation analysis method.

4. A performance optimization system for ultra-high temperature cementing, characterized in that, include: The data acquisition module is used to acquire the performance influence characteristics and compressive strength sample dataset of the ultra-high temperature cement. The feature determination module is used to determine the features affecting the performance of ultra-high temperature cementing in the compressive strength sample dataset; The model determination module is used to take the performance influence characteristics of the ultra-high temperature cementing cement as input features, the compressive strength of the ultra-high temperature cementing cement as output target, and determine the optimal model from a set machine learning model library according to the influence relationship between the input features and the output target. The coefficient calculation module is used to determine the influence coefficient of each of the ultra-high temperature cementing performance influence characteristics on the compressive strength of each of the ultra-high temperature cementing cements under the performance testing conditions of the optimal model. The performance prediction module is used to generate a performance prediction model for the ultra-high temperature cementing cement based on the influence coefficients of the performance characteristics of each ultra-high temperature cementing cement and the optimal model; the performance prediction model is used to predict the compressive strength of the ultra-high temperature cementing cement. The optimization module is used to screen the target features whose influence coefficients are greater than a set influence threshold among the performance influence features of the ultra-high temperature cementing cement, and generate an optimization scheme for the ultra-high temperature cementing cement based on the influence coefficients of each target feature. The data filtering module is used to identify abnormal data in the performance impact characteristics and the compressive strength sample dataset; After removing the abnormal data, a sample dataset of effective performance impact characteristics and compressive strength is obtained; The data filtering module includes: An abnormal data filtering unit is used to determine the first quartile and the third quartile for any type of data in the compressive strength sample dataset; determine the interquartile range based on the first quartile and the third quartile; determine the upper and lower limits of outliers based on the interquartile range; and remove abnormal data from the corresponding type of data based on the upper and lower limits.

5. The performance optimization system for ultra-high temperature cementing as described in claim 4, characterized in that, The model determination module includes: The model evaluation unit is used to determine the evaluation metrics for each machine learning model in the defined machine learning model library; the evaluation metrics include any combination of several of the following: coefficient of determination, mean absolute error, mean square error, and root mean square error. The model validation unit is used to compare the evaluation indicators using cross-validation to determine the optimal model.

Images