Machine learning-based leakage channel size prediction model correction method
By using Spearman correlation analysis and the Adaboost algorithm to correct the fracture width model during drilling, the problem of large calculation errors in fracture width in existing technologies is solved, and the accuracy and reliability of leakage depth prediction are improved.
Patent Information
- Application Number
- CN202510892094.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-30
- Publication Date
- 2025-11-28
AI Technical Summary
Existing crack width calculation models have significant errors in practical applications, resulting in poor accuracy and reliability in predicting leakage depth.
By acquiring basic drilling parameters, Spearman correlation analysis was used to select parameters with correlation coefficients greater than a preset threshold, a dataset was constructed, and the Adaboost algorithm was used to train the model to obtain a fracture width correction model, which corrected the basic predicted fracture width.
By combining drilling field data and physical model predictions, the model for predicting the size of the leakage channel was corrected using machine learning algorithms. This improved the accuracy and reliability of the leakage depth prediction and demonstrated strong versatility and adaptability.
Smart Images
Figure CN121031263A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer technology, and in particular to a method for correcting a leak channel size prediction model based on machine learning. Background Technology
[0002] Fracture width affects drilling fluid loss. Larger fractures can lead to significant fluid loss, wasting fluid, increasing costs, and potentially causing wellbore instability. Accurately predicting fracture width allows for the selection of appropriate drilling fluid types and performance parameters to minimize loss.
[0003] While existing crack width calculation models are theoretically feasible, they suffer from significant errors in practical applications, resulting in poor accuracy and reliability in predicting leakage depth. Summary of the Invention
[0004] In view of this, the present invention provides a method for correcting the missing channel size prediction model based on machine learning to solve the above problems.
[0005] This invention provides a method for correcting a leakage channel size prediction model based on machine learning, comprising: acquiring drilling baseline parameters related to fracture width; inputting the drilling baseline parameters into the acquired leakage channel size prediction model to obtain a baseline predicted fracture width; calculating and analyzing the correlation between the drilling baseline parameters using Spearman correlation analysis to obtain analysis results; selecting parameters with correlation coefficients greater than a preset threshold based on the analysis results to construct a dataset; training the model based on the dataset and the Adaboost algorithm to obtain a fracture width correction model; and correcting the baseline predicted fracture width using the fracture width correction model to obtain a corrected fracture width.
[0006] In another implementation of the present invention, the basic drilling parameters include drilling pressure differential, wellbore radius, elastic compressibility coefficient, drilling fluid density, and leakage time.
[0007] In another implementation of the present invention, the leakage channel size prediction model is expressed as:
[0008]
[0009] Where w is the calculated fracture width; Q(t) is the leakage rate; Δp is the drilling pressure differential; r w Where is the wellbore radius; π is pi, taken as 3.14; ρ is the drilling fluid density; C t t is the elastic compressibility coefficient; μ is the drilling fluid viscosity; t is the leakage time.
[0010] In another implementation of the present invention, the dataset is further divided into training data and test data in an 8:2 ratio.
[0011] In another implementation of the present invention, the step of training the model based on the dataset and the Adaboost algorithm to obtain the crack width correction model includes: determining a weak predictor and its corresponding data, and determining the initial weights of each group of training data according to the correlation analysis results; iteratively calculating the weak predictor using the training data, updating the weights of the training data according to the results of each iteration, assigning higher weights to data with poor prediction performance so that these data receive more attention in the next iteration; calculating a sequence of prediction functions by repeatedly iterating the weak predictor, with each prediction function having a corresponding weight, and the better the prediction function, the greater its weight; after a preset number of iterations, obtaining the function of the strong predictor through the weighting of the weak prediction functions, which serves as the crack width correction model.
[0012] In another implementation of the present invention, the method further includes: evaluating the crack width correction model using a coefficient of determination to obtain an evaluation result.
[0013] In another implementation of the present invention, the determination coefficient is expressed as:
[0014]
[0015] Among them, y i This represents the true value of the i-th sample. This represents the predicted value of the i-th sample.
[0016] In another aspect, the present invention provides a machine learning-based system for correcting a leak channel size prediction model, comprising: a data acquisition module for acquiring basic drilling parameters related to fracture width; a basic result prediction module for inputting the basic drilling parameters into the acquired leak channel size prediction model to obtain a basic predicted fracture width; a correlation analysis module for calculating and analyzing the correlation between the basic drilling parameters using the Spearman correlation analysis method to obtain analysis results; a dataset construction module for selecting parameters with correlation coefficients greater than a preset threshold based on the analysis results to construct a dataset; a correction model training module for training the model based on the dataset and the Adaboost algorithm to obtain a fracture width correction model; and a correction result output module for correcting the basic predicted fracture width using the fracture width correction model to obtain a corrected fracture width.
[0017] In another aspect, the present invention provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of a machine learning-based method for correcting a leaky channel size prediction model as described in any of the preceding claims.
[0018] In another aspect, the present invention provides a computer storage medium, characterized in that the computer storage medium stores a computer program, which, when executed by a processor, implements the steps of a machine learning-based method for correcting a missing channel size prediction model as described in any of the preceding claims.
[0019] The machine learning-based method for correcting the leakage channel size prediction model of the present invention collects a large amount of drilling field data and actual fracture width data, combines the prediction results of the physical model, and uses machine learning algorithms to correct the prediction results of the leakage channel size prediction model. It fully considers the complexity and uncertainty of geological conditions, thereby improving the accuracy and reliability of leakage depth prediction. It is applicable to different leakage channel size prediction models and has strong versatility and adaptability. Attached Figure Description
[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. By reading the detailed description of the embodiments below, the advantages and benefits of the solutions will become clear to those skilled in the art. The accompanying drawings are only for illustrating preferred embodiments and are not intended to limit the present invention. In the accompanying drawings:
[0021] Figure 1 This is a schematic diagram of a machine learning-based method for correcting a missing channel size prediction model, according to an embodiment of the present invention.
[0022] Figure 2 This is a schematic diagram showing the results of correlation calculation and analysis for one embodiment of the present invention.
[0023] Figure 3 This is a schematic diagram of the Adaboost algorithm flow according to an embodiment of the present invention.
[0024] Figure 4 This is a schematic diagram of the scatter plot fitting of the true and predicted values of the Bertuzzi correction model according to an embodiment of the present invention.
[0025] Figure 5 This is a schematic diagram of the residuals of the Bertuzzi correction model according to an embodiment of the present invention. Detailed Implementation
[0026] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and thoroughly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art should fall within the protection scope of the present invention.
[0027] Figure 1 A schematic diagram of a machine learning-based method for correcting a prediction model of lost channel size is provided in an embodiment of the present invention, as shown below. Figure 1 As shown, this embodiment mainly includes:
[0028] S101. Obtain drilling base parameters related to fracture width.
[0029] S102. Input the drilling basic parameters into the obtained leakage channel size prediction model to obtain the basic predicted fracture width.
[0030] S103. The correlation between the drilling basic parameters is calculated and analyzed using the Spearman correlation analysis method to obtain the analysis results.
[0031] S104. Based on the analysis results, parameters with correlation coefficients greater than a preset threshold are selected to construct a dataset.
[0032] S105. Based on the dataset and the Adaboost algorithm, train the model to obtain the crack width correction model.
[0033] S106. The basic predicted crack width is corrected by the crack width correction model to obtain the corrected crack width.
[0034] The machine learning-based method for correcting the leakage channel size prediction model of the present invention collects a large amount of drilling field data and actual fracture width data, combines the prediction results of the physical model, and uses machine learning algorithms to correct the prediction results of the leakage channel size prediction model. It fully considers the complexity and uncertainty of geological conditions, thereby improving the accuracy and reliability of leakage depth prediction. It is applicable to different leakage channel size prediction models and has strong versatility and adaptability.
[0035] In another implementation of the present invention, the basic drilling parameters include drilling pressure differential, wellbore radius, elastic compressibility coefficient, drilling fluid density, and leakage time.
[0036] In another implementation of the present invention, the leakage channel size prediction model can use the Bertuzzits model, expressed as:
[0037]
[0038] Where w is the calculated crack width, mm; Q(t) is the leakage rate, m. 3 / s; Δp is the drilling pressure differential, MPa; r w ρ is the wellbore radius, in meters; π is pi, typically taken as 3.14; ρ is the drilling fluid density, in kilograms per cubic meter of water. 3 C t The elastic compressibility coefficient, Pa⁻¹, is typically taken as 4 × 10⁻¹⁰. -4 Pa -1 μ is the drilling fluid viscosity, Pa·s; t is the leakage time, s.
[0039] For example, some of the collected basic data and crack width data calculated based on the physical model are shown in Table 1.
[0040] Table 1
[0041]
[0042] like Figure 2 As shown, to investigate the influence of different factors on fracture width and identify the main influencing factors, Spearman correlation analysis was used to calculate and analyze the correlation between all parameters. Due to formation compression C t Take a fixed value of 0.0004, P i A fixed value of 3.14 is chosen, therefore the formation is compressed by C. t and P i After removing these two features, Spearman correlation analysis is used to calculate and analyze the correlation between all parameters.
[0043] It can be seen that the leakage rate Q(t), pressure difference Δp, and wellbore radius r w Fluid density ρ, elastic compressibility C t Features such as drilling fluid viscosity (μ) and leakage time (t) all have a certain impact on fracture width error. Among them, drilling fluid viscosity and leakage rate have the strongest correlation, with correlation coefficients of -0.88 and -0.87, respectively. The next most significant correlations are the cumulative leakage amount and drilling fluid density, with correlation coefficients of -0.74 and -0.63, respectively. The correlation between wellbore radius and leakage time during leakage is relatively small, at 0.028 and 0.031, respectively.
[0044] In another implementation of the present invention, the dataset is further divided into training data and test data in an 8:2 ratio.
[0045] In another implementation of the present invention, the model training based on the dataset and the Adaboost algorithm to obtain the crack width correction model is as follows: Figure 3 As shown, the process includes: determining a weak predictor and its corresponding data, and determining the initial weights of each training data set based on the correlation analysis results; iteratively calculating the weak predictor using the training data, updating the weights of the training data according to the results of each iteration, assigning higher weights to data with poor prediction performance so that these data receive more attention in the next iteration; calculating a sequence of prediction functions through repeated iterations of the weak predictor, with each prediction function having a corresponding weight, and the better the prediction function, the greater its weight; and after a predetermined number of iterations, obtaining the function of the strong predictor through the weighted sum of the weak prediction functions, which serves as the crack width correction model.
[0046] For example, the AdaBoost algorithm is an iterative algorithm, belonging to the adaptive boosting algorithm category. Its core idea is to generate effective predictions by merging the outputs of multiple "weak" predictors. A crack width correction model based on AdBoost is established, corresponding to the number of weak learners, and optimized using a grid search method.
[0047] In another implementation of the present invention, the method further includes: evaluating the crack width correction model using a coefficient of determination to obtain an evaluation result.
[0048] In another implementation of the present invention, the determination coefficient is expressed as:
[0049]
[0050] Among them, y i This represents the true value of the i-th sample. This represents the predicted value of the i-th sample (the predicted value obtained through a linear regression model).
[0051] For example, the coefficient of determination R 2 The coefficient of determination R represents the ratio of the regression sum of squares to the total sum of squares. 2 The closer the value is to 1, the more accurate the model is and the more significant the regression effect is.
[0052] In another implementation of the present invention, the R2 value of the model is the largest, which is 0.862, when the number of weak learners is 100 on the test set.
[0053] like Figure 4 and Figure 5 The figures show the fitting results and residual plots of the Bertuzzi model based on AdBoost on the test set. Figure 4 As can be seen, the predicted values of the modified model are quite close to the actual values, indicating that the model fits well and the prediction accuracy of the modified model has been improved. Figure 5As can be seen, the residuals of the model are randomly distributed, indicating that the model performs well. Therefore, the Bertuzzi model based on AdBoost correction has good applicability.
[0054] Another aspect of the present invention provides a machine learning-based model correction system for predicting the size of lost channels, comprising:
[0055] Data acquisition module: Acquires drilling baseline parameters related to fracture width.
[0056] Basic Result Prediction Module: Input the drilling basic parameters into the acquired leakage channel size prediction model to obtain the basic predicted fracture width.
[0057] Correlation analysis module: The Spearman correlation analysis method is used to calculate and analyze the correlation between the drilling basic parameters and obtain the analysis results.
[0058] Dataset construction module: Based on the analysis results, parameters with correlation coefficients greater than a preset threshold are selected to construct the dataset.
[0059] Correction model training module: Based on the dataset and the Adaboost algorithm, the model is trained to obtain the crack width correction model.
[0060] Correction result output module: Corrects the basic predicted crack width using the crack width correction model to obtain the corrected crack width.
[0061] The machine learning-based leakage channel size prediction model correction system of the present invention collects a large amount of drilling field data and actual fracture width data, combines the prediction results of the physical model, and uses machine learning algorithms to correct the prediction results of the leakage channel size prediction model. It fully considers the complexity and uncertainty of geological conditions, thereby improving the accuracy and reliability of leakage depth prediction. It can be applied to different leakage channel size prediction models and has strong versatility and adaptability.
[0062] In another aspect of the present invention, the electronic device includes: a processor, a memory, and a communication bus and a communication interface.
[0063] in:
[0064] The processor, memory, and communication interface communicate with each other via a communication bus.
[0065] A communication interface is used to communicate with other electronic devices or servers.
[0066] The processor is used to execute programs, specifically, to execute any of the steps of the machine learning-based missing channel size prediction model correction method in the above embodiments.
[0067] Specifically, the program may include program code, which includes computer operation instructions.
[0068] The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs.
[0069] Memory is used to store programs. Memory may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive.
[0070] Specifically, the program can be used to cause the processor to execute the steps of any of the machine learning-based missing channel size prediction model correction methods described in the embodiments. The specific implementation of each step in the program can be found in the corresponding descriptions of the steps and units executed by any of the machine learning-based missing channel size prediction model correction methods described above, and will not be repeated here. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the devices and modules described above can be referred to the corresponding process descriptions in the foregoing method embodiments.
[0071] An exemplary embodiment of this application also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform the methods of various embodiments of this application.
[0072] The methods described above according to embodiments of the present invention can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code originally stored on a remote recording medium or a non-transitory machine-readable medium and subsequently stored on a local recording medium, downloaded via a network. Thus, the methods described herein can be processed by software stored on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components (e.g., RAM, ROM, flash memory, etc.) capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods described herein. Furthermore, when a general-purpose computer accesses code used to implement the methods shown herein, the execution of the code transforms the general-purpose computer into a dedicated computer for executing the methods shown herein.
[0073] Specific embodiments of the present invention have now been described. Other embodiments are within the scope of the appended claims. In some cases, the actions described in the claims can be performed in a different order and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result.
[0074] It should be noted that all directional indications (such as up, down, left, right, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship between the components in a certain specific order (as shown in the figure). If the specific order changes, the directional indication will also change accordingly.
[0075] In the description of this invention, the terms "first" and "second" are used only for convenience in describing different components or names, and should not be construed as indicating or implying a sequential relationship, relative importance, or implicitly specifying the number of technical features indicated. Thus, a feature defined with "first" and "second" may explicitly or implicitly include at least one of that feature.
[0076] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
[0077] It should be noted that although specific embodiments of the present invention have been described in detail with reference to the accompanying drawings, this should not be construed as limiting the scope of protection of the present invention. Various modifications and variations that can be made by those skilled in the art without inventive effort within the scope described in the claims still fall within the scope of protection of the present invention.
[0078] The examples of the embodiments of the present invention are intended to concisely illustrate the technical features of the embodiments of the present invention, so that those skilled in the art can intuitively understand the technical features of the embodiments of the present invention, and are not intended to be an improper limitation of the embodiments of the present invention.
[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for correcting a leak channel size prediction model based on machine learning, characterized in that, include: Obtain drilling base parameters related to fracture width; The drilling basic parameters are input into the obtained leakage channel size prediction model to obtain the basic predicted fracture width; The correlation between the drilling basic parameters was calculated and analyzed using the Spearman correlation analysis method, and the analysis results were obtained. Based on the analysis results, parameters with correlation coefficients greater than a preset threshold are selected to construct a dataset; Based on the dataset and the Adaboost algorithm, a crack width correction model is obtained through model training. The predicted crack width is corrected by using the crack width correction model to obtain the corrected crack width.
2. The method according to claim 1, characterized in that, The basic drilling parameters include drilling pressure differential, wellbore radius, elastic compressibility coefficient, drilling fluid density, and leakage time.
3. The method according to claim 2, characterized in that, The leakage channel size prediction model is expressed as follows: Where w is the calculated fracture width; Q(t) is the leakage rate; Δp is the drilling pressure differential; r w Where is the wellbore radius; π is pi, taken as 3.14; ρ is the drilling fluid density; C t t is the elastic compressibility coefficient; μ is the drilling fluid viscosity; t is the leakage time.
4. The method according to claim 1, characterized in that, Also includes: The dataset is divided into training and testing data in an 8:2 ratio.
5. The method according to claim 4, characterized in that, The process of training the model based on the dataset and the Adaboost algorithm to obtain the crack width correction model includes: Weak predictors and corresponding data are identified, and the initial weights for each group of training data are determined based on the correlation analysis results. The weak predictor is iteratively calculated using the training data, and the weights of the training data are updated according to the results of each iteration. Data with poor prediction performance are given higher weights so that these data will receive more attention in the next iteration. The prediction function sequence is calculated by iteratively calculating the weak predictor. Each prediction function has a corresponding weight, and the better the prediction function, the greater the weight. After several preset iterations, the function of the strong predictor is obtained by weighting the weak predictor function, which serves as the crack width correction model.
6. The method according to claim 1, characterized in that, Also includes: The crack width correction model was evaluated using the coefficient of determination, and the evaluation results were obtained.
7. The method according to claim 6, characterized in that, The determination coefficient is expressed as: Among them, y i This represents the true value of the i-th sample. This represents the predicted value of the i-th sample.
8. A machine learning-based model correction system for predicting the size of lost channels, characterized in that, include: Data acquisition module: Acquires drilling basic parameters related to fracture width; Basic Result Prediction Module: Input the drilling basic parameters into the acquired leakage channel size prediction model to obtain the basic predicted fracture width; Correlation analysis module: The correlation between the drilling basic parameters is calculated and analyzed using the Spearman correlation analysis method to obtain the analysis results; Dataset construction module: Based on the analysis results, parameters with correlation coefficients greater than a preset threshold are selected to construct the dataset; Correction model training module: Based on the dataset and Adaboost algorithm, the model is trained to obtain the crack width correction model; Correction result output module: Corrects the basic predicted crack width using the crack width correction model to obtain the corrected crack width.
9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the machine learning-based leak channel size prediction model correction method as described in any one of claims 1 to 7.
10. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the steps in the machine learning-based leak channel size prediction model correction method as described in any one of claims 1 to 7.