Cement paste strength prediction method and system based on integration algorithm

Abstract

The invention discloses a cement paste strength prediction method based on an integration algorithm. The method comprises the following steps: collecting original cement paste data; performing feature extraction on the original cement paste data to obtain to-be-predicted cement paste data; and sequentially inputting the to-be-predicted cement paste data into a trained convolutional neural network model, an XGBoost model and a random forest model to obtain the compressive strength of the cement paste test block with the standard size, so as to predict the cement paste strength according to the compressive strength of the cement paste test block with the standard size. According to the cement paste strength prediction method based on the integration algorithm, the strength of the cement paste is predicted by using three models, and respective advantages, the powerful capability of CNN in image and signal processing, the high efficiency of XGBoost in structured data processing and the robustness of a random forest in classification and regression problems can be utilized, so that the strength of the cement paste is predicted. Therefore, the prediction accuracy and robustness are improved.

Description

Technical Field

[0001] The present invention belongs to the field of cement slurry data processing, and in particular relates to a cement slurry strength prediction method and system based on an integrated algorithm. Background Art

[0002] As one of the important means of oil production, cementing technology is closely related to the quality of the entire wellbore. In cementing engineering operations, due to the complex influencing conditions of many aspects, the design of cement slurry system is a key issue, and the choice of cement slurry has a direct impact on the quality of cementing.

[0003] Cement slurry is a new type of composite material with a large amount of chemical admixtures and auxiliary cementitious materials. The cement slurry used in seawater is designed to meet the special requirements of the marine environment. Factors such as the chloride ion concentration in seawater, the curing age, and the amount of fly ash used will affect the mechanical properties and pore structure of the cement slurry, which is slightly different from the materials used for land cementing. The strength of the cement slurry used in marine cementing has a highly nonlinear relationship with each component. How to effectively reveal the relationship between the cement slurry components and strength is of great significance to the modification of cement slurry and the optimization of the mix design.

[0004] In terms of concrete strength prediction, convolutional neural networks can make predictions by learning the complex nonlinear relationship between the ratio of input features such as cement, slag, aggregates, etc. and the output. A study mentioned that the compressive strength of different types of concrete can be predicted by using only the CNN model, and the reliability of the model was verified by experiments. There are some shortcomings in this study: the limitation of the data set, only 380 data sets are not enough to cover all factors that may affect the compressive strength of concrete. The uncertainty of generalization ability and the robustness of the model affect the accuracy of the prediction. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a cement slurry strength prediction method and system based on an integrated algorithm. The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0006] A cement slurry strength prediction method based on an integrated algorithm, comprising:

[0007] Collect raw cement slurry data;

[0008] Extracting features from the original cement slurry data to obtain cement slurry data to be predicted;

[0009] The cement slurry data to be predicted is input into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

[0010] In a specific embodiment, the training process of the convolutional neural network model is:

[0011] Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples;

[0012] Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process;

[0013] When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy;

[0014] The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

[0015] In a specific implementation, the training process of the XGBoost model is:

[0016] Preprocess the input data, including normalization and missing value processing.

[0017] Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters;

[0018] The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations;

[0019] After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

[0020] In a specific implementation, the training process of the random forest model is:

[0021] Preprocess the input data, including normalization and missing value filling;

[0022] Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method;

[0023] Construct multiple decision trees as weak learners;

[0024] The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently.

[0025] The prediction results of multiple decision trees are combined to obtain the final prediction result.

[0026] In a specific embodiment, the training weights of the convolutional neural network model, the XGBoost model, and the random forest model are 0.25, 0.35, and 0.4, respectively.

[0027] In a specific embodiment, feature extraction is performed on the original cement slurry data to obtain cement slurry data to be predicted, and the method includes:

[0028] removing duplicate data, invalid data and abnormal data from the original cement slurry data;

[0029] Filling the missing data in the original cement slurry data;

[0030] Normalize or standardize the data;

[0031] Use Random Forest or XGBoost models for feature importance assessment;

[0032] According to the feature importance ranking, the features that contribute more to the prediction of cement slurry strength are selected for subsequent modeling.

[0033] The present invention also provides a cement slurry strength prediction system based on an integrated algorithm, comprising:

[0034] Data acquisition module, used to collect raw cement slurry data;

[0035] A data processing module, used for extracting features from the original cement slurry data to obtain cement slurry data to be predicted;

[0036] The strength prediction module is used to input the cement slurry data to be predicted into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

[0037] In a specific embodiment, the training process of the convolutional neural network model is:

[0038] Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples;

[0039] Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process;

[0040] When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy;

[0041] The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

[0042] In a specific implementation, the training process of the XGBoost model is:

[0043] Preprocess the input data, including normalization and missing value processing.

[0044] Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters;

[0045] The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations;

[0046] After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

[0047] In a specific implementation, the training process of the random forest model is:

[0048] Preprocess the input data, including normalization and missing value filling;

[0049] Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method;

[0050] Construct multiple decision trees as weak learners;

[0051] The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently.

[0052] The prediction results of multiple decision trees are combined to obtain the final prediction result.

[0053] Beneficial effects of the present invention:

[0054] 1. The cement slurry strength prediction method based on an integrated algorithm of the present invention predicts the strength of cement slurry by using three models, which can take advantage of their respective advantages, the powerful ability of CNN in image and signal processing, the high efficiency of XGBoost in processing structured data and the robustness of random forest in classification and regression problems, thereby improving the accuracy and robustness of the prediction.

[0055] 2. The integrated model of the cement slurry strength prediction method based on the integrated algorithm of the present invention can improve the prediction performance, reduce the risk of overfitting, and reduce the prediction error by integrating the prediction results of multiple models. The influence of cement slurry characteristics on strength can be predicted through the convolution kernel weight, thereby improving the performance and generalization ability of the overall model.

[0056] 3. The cement slurry strength prediction method based on the integrated algorithm of the present invention trains three models separately, stacks them and then trains a model regression, which provides higher flexibility and can adapt to different data characteristics and prediction tasks.

[0057] 4. The cement slurry strength prediction method based on an integrated algorithm of the present invention optimizes computational efficiency. CNN requires a large amount of computing resources. By combining it with XGBoost and random forest, a lighter model can be designed, or the amount of computation can be reduced while maintaining accuracy.

[0058] 5. The cement slurry strength prediction method based on an integrated algorithm of the present invention enables cement slurry engineers to use their prior knowledge on the interpretable characteristics of cement slurry formulations, thereby promoting the understanding of the impact of construction site components on the development of cement slurry strength and better serving the actual construction needs of cement slurry strength design.

[0059] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic flow chart of a cement slurry strength prediction method based on an integrated algorithm provided in an embodiment of the present invention;

[0061] Figure 2 It is a module block diagram of a cement slurry strength prediction system based on an integrated algorithm provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0063] Embodiment 1

[0064] See also Figure 1 , Figure 1 The present invention provides a flow chart of a cement slurry strength prediction method based on an integrated algorithm, including:

[0065] S1. Collecting original cement slurry data;

[0066] S2, extracting features from the original cement slurry data to obtain cement slurry data to be predicted;

[0067] S3. Input the cement slurry data to be predicted into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

[0068] In a specific embodiment, the training process of the convolutional neural network model is:

[0069] Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples;

[0070] Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process;

[0071] When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy;

[0072] The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

[0073] In a specific implementation, the training process of the XGBoost model is:

[0074] Preprocess the input data, including normalization and missing value processing.

[0075] Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters;

[0076] The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations;

[0077] After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

[0078] In a specific implementation, the training process of the random forest model is:

[0079] Preprocess the input data, including normalization and missing value filling;

[0080] Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method;

[0081] Construct multiple decision trees as weak learners;

[0082] The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently.

[0083] The prediction results of multiple decision trees are combined to obtain the final prediction result.

[0084] In a specific embodiment, the training weights of the convolutional neural network model, the XGBoost model, and the random forest model are 0.25, 0.35, and 0.4, respectively.

[0085] In a specific embodiment, feature extraction is performed on the original cement slurry data to obtain cement slurry data to be predicted, and the method includes:

[0086] removing duplicate data, invalid data and abnormal data from the original cement slurry data;

[0087] Filling the missing data in the original cement slurry data;

[0088] Normalize or standardize the data;

[0089] Use Random Forest or XGBoost models for feature importance assessment;

[0090] According to the feature importance ranking, the features that contribute more to the prediction of cement slurry strength are selected for subsequent modeling.

[0091] See also Figure 2 This embodiment also provides a cement slurry strength prediction system based on an integrated algorithm, including:

[0092] Data acquisition module, used to collect raw cement slurry data;

[0093] A data processing module, used for extracting features from the original cement slurry data to obtain cement slurry data to be predicted;

[0094] The strength prediction module is used to input the cement slurry data to be predicted into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

[0095] In a specific embodiment, the training process of the convolutional neural network model is:

[0096] Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples;

[0097] Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process;

[0098] When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy;

[0099] The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

[0100] In a specific implementation, the training process of the XGBoost model is:

[0101] Preprocess the input data, including normalization and missing value processing.

[0102] Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters;

[0103] The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations;

[0104] After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

[0105] In a specific implementation, the training process of the random forest model is:

[0106] Preprocess the input data, including normalization and missing value filling;

[0107] Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method;

[0108] Construct multiple decision trees as weak learners;

[0109] The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently.

[0110] The prediction results of multiple decision trees are combined to obtain the final prediction result.

[0111] The cement slurry strength prediction method based on an integrated algorithm in this embodiment predicts the strength of cement slurry by using three models, which can take advantage of their respective advantages, the powerful ability of CNN in image and signal processing, the high efficiency of XGBoost in processing structured data, and the robustness of random forest in classification and regression problems, thereby improving the accuracy and robustness of the prediction.

[0112] The integrated model of the cement slurry strength prediction method based on the integrated algorithm of this embodiment can improve the prediction performance, reduce the risk of overfitting, and reduce the prediction error by integrating the prediction results of multiple models. The influence of cement slurry characteristics on strength can be predicted through the convolution kernel weight, thereby improving the performance and generalization ability of the overall model.

[0113] In the cement slurry strength prediction method based on an integrated algorithm of this embodiment, three models are trained separately, and then stacked to train a model regression, which provides higher flexibility and can adapt to different data characteristics and prediction tasks.

[0114] The cement slurry strength prediction method based on an integrated algorithm in this embodiment optimizes computational efficiency. CNN requires a large amount of computing resources. By combining it with XGBoost and random forest, a lighter model can be designed, or the amount of computation can be reduced while maintaining accuracy.

[0115] An integrated algorithm-based cement slurry strength prediction method of this embodiment enables cement slurry engineers to use their prior knowledge on the interpretable characteristics of cement slurry formulations, thereby facilitating the understanding of the impact of construction site components on cement slurry strength development and better serving the actual construction needs of cement slurry strength design.

[0116] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification.

[0117] Although the present application is described herein in conjunction with various embodiments, in the process of implementing the claimed application, those skilled in the art may understand and implement other variations of the disclosed embodiments by viewing the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "one" or "an" does not exclude multiple situations. A single processor or other unit may implement several functions listed in a claim. Certain measures are recorded in different dependent claims, but this does not mean that these measures cannot be combined to produce good results.

[0118] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A cement slurry strength prediction method based on an integrated algorithm, characterized in that: include: Collect raw cement slurry data; Extracting features from the original cement slurry data to obtain cement slurry data to be predicted; The cement slurry data to be predicted is input into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

2. The cement slurry strength prediction method based on integrated algorithm according to claim 1 is characterized in that: The training process of the convolutional neural network model is as follows: Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples; Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process; When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy; The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

3. The cement slurry strength prediction method based on integrated algorithm according to claim 1 is characterized in that: The training process of the XGBoost model is: Preprocess the input data, including normalization and missing value processing. Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters; The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations; After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

4. The cement slurry strength prediction method based on integrated algorithm according to claim 1, characterized in that: The training process of the random forest model is: Preprocess the input data, including normalization and missing value filling; Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method; Construct multiple decision trees as weak learners; The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently. The prediction results of multiple decision trees are combined to obtain the final prediction result.

5. The cement slurry strength prediction method based on integrated algorithm according to claim 1, characterized in that: The training weights of the convolutional neural network model, the XGBoost model and the random forest model are 0.25, 0.35 and 0.4 respectively.

6. The cement slurry strength prediction method based on integrated algorithm according to claim 1, characterized in that: Feature extraction is performed on the original cement slurry data to obtain cement slurry data to be predicted, and the method includes: removing duplicate data, invalid data and abnormal data from the original cement slurry data; Filling the missing data in the original cement slurry data; Normalize or standardize the data; Use Random Forest or XGBoost models for feature importance assessment; According to the feature importance ranking, the features that contribute more to the prediction of cement slurry strength are selected for subsequent modeling.

7. A cement slurry strength prediction system based on an integrated algorithm, characterized in that: include: Data acquisition module, used to collect raw cement slurry data; A data processing module, used for extracting features from the original cement slurry data to obtain cement slurry data to be predicted; The strength prediction module is used to input the cement slurry data to be predicted into the trained convolutional neural network model, XGBoost model and random forest model in sequence to obtain the compressive strength of the cement slurry standard size test block, so as to predict the cement slurry strength according to the compressive strength of the cement slurry standard size test block.

8. The cement slurry strength prediction system based on integrated algorithm according to claim 7, characterized in that: The training process of the convolutional neural network model is as follows: Obtain a training data set and a test data set, the training data set contains 80% of the data samples, and the test data set contains 20% of the data samples; Initialize the convolutional neural network, train the convolutional neural network using the training data set, and determine the error during the training process; When the number of iterations of the convolutional neural network is set, the test data set is input into the convolutional neural network for testing to obtain the test accuracy; The convolutional neural network is trained repeatedly using the training data set until the error of the convolutional neural network is lower than a preset first threshold and the test accuracy is higher than a preset second threshold, thereby obtaining a target convolutional neural network.

9. The cement slurry strength prediction system based on integrated algorithm according to claim 8, characterized in that: The training process of the XGBoost model is: Preprocess the input data, including normalization and missing value processing. Set the parameters of the XGBoost model, which include general parameters, task parameters, and boost parameters; The preprocessed data is input into the XGBoost model for training, and iterates multiple times until the model converges or reaches the preset number of iterations; After training is completed, the importance of each feature is evaluated for feature selection and model optimization.

10. The cement slurry strength prediction system based on integrated algorithm according to claim 7, characterized in that: The training process of the random forest model is: Preprocess the input data, including normalization and missing value filling; Set the parameters of random forest, including the number of decision trees, maximum depth, and feature selection method; Construct multiple decision trees as weak learners; The preprocessed data is input into the random forest model for training, and each decision tree learns and predicts independently. The prediction results of multiple decision trees are combined to obtain the final prediction result.

Images