Well kill overflow event prediction method, system and device based on hierarchical multi-joint learning
By constructing a well control overflow event prediction method based on hierarchical multi-joint learning, and utilizing feature contribution analysis and voting mechanism, the problem of predicting overflow events in well control operations is solved, improving prediction accuracy and safety, and optimizing resource utilization and management efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CNPC BOHAI DRILLING ENG
- Filing Date
- 2024-12-02
- Publication Date
- 2026-06-02
AI Technical Summary
In the petroleum industry, the risk of overflow events caused by changes in formation conditions, operational errors, or equipment failures during well control operations can lead to equipment damage, production interruption, safety accidents, and environmental pollution. Existing technologies are difficult to effectively predict and prevent these events.
A hierarchical multi-joint learning approach is adopted to construct multiple base models for predicting wellbore overflow events. An ensemble model is then built through feature contribution analysis and a voting mechanism to improve prediction accuracy and robustness.
It improves the accuracy of well control overflow prediction, reduces the risk of equipment damage and safety accidents, optimizes resource utilization, and enhances safety and management efficiency.
Smart Images
Figure CN122129223A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of overflow events, and specifically relates to a method, system and device for predicting well control overflow events based on hierarchical multi-joint learning. Background Technology
[0002] In the oil industry, well control is a crucial operation used to control bottomhole pressure, maintain wellbore stability, and ensure production safety. Properly executed well control is essential for the smooth operation and safe production of an oilfield. However, due to various factors such as changes in formation conditions, operational errors, or equipment failures, the risk of well-control overflows always exists. Overflows not only cause equipment damage and production interruptions but can also trigger serious safety accidents and environmental pollution, severely impacting the entire oilfield. Therefore, effectively predicting well-control overflow events is crucial to helping reduce accident risks and ensure production safety.
[0003] Based on this, the present invention proposes a method, system and device for predicting well control overflow events based on hierarchical multi-joint learning. Summary of the Invention
[0004] To address the aforementioned problems in existing technologies, namely the risk of overflow events during well control operations due to factors such as changes in formation conditions, operational errors, or equipment failures, which may lead to equipment damage, production interruption, safety accidents, and environmental pollution, this invention provides a well control overflow event prediction method, system, and equipment based on hierarchical multi-joint learning.
[0005] In a first aspect, the present invention proposes a method for predicting wellbore overflow events based on hierarchical multi-joint learning, the method comprising:
[0006] Obtain the dataset and divide it into a training set and a test set;
[0007] The training set is used to train multiple well control overflow event prediction base models to obtain a trained well control overflow event prediction base model.
[0008] The test set is input into multiple trained well control overflow event prediction base models to obtain prediction results. The accuracy of each trained well control overflow event prediction base model is calculated based on the prediction results and label data.
[0009] Calculate the average of multiple accuracies, and select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, and select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution to take the union of the features to obtain the final feature set.
[0010] The retained models are integrated based on the final feature set to construct a voting-based ensemble model. The data to be predicted is input into the ensemble model to obtain the prediction results of the well control overflow event.
[0011] In some preferred embodiments, the accuracy ACC of each trained well control blowout event prediction base model is... i The calculation method is as follows:
[0012]
[0013] Among them, Y test For the labeled data in the test set, X test For the feature data in the test set, Y test Let j be the j-th label data in the test set, and X be... test Let j be the j-th feature data in the test set, and M be the feature data. i Let be the i-th base model, and 1 be the indicator function, which takes the value 1 when the prediction matches the actual result, and 0 otherwise.
[0014] In some preferred embodiments, the cumulative contribution C i,j The calculation method is as follows:
[0015]
[0016] Among them, Importance i,k This represents the contribution of the k-th feature in the i-th base model.
[0017] In some preferred embodiments, a set F of features whose cumulative contribution reaches or exceeds a preset percentage of the total contribution is selected. i The method is as follows:
[0018]
[0019] in, Let |X| represent the total contribution of all features in the i-th base model, and let |X| represent the number of features. i，j Let represent the j-th feature in the i-th base model, and 'a' represent the preset percentage.
[0020] In some preferred embodiments, the ensemble model E(X) is constructed as follows:
[0021]
[0022] Where Vote is the voting mechanism function. Let X be the k-th retained base model, and let X be the input data.
[0023] In some preferred embodiments, the prediction results include category data, which are obtained by:
[0024]
[0025] Where P(c|X) is the probability that the ensemble model predicts that the input data X belongs to class c, Pi(c|X) is the probability that the i-th base model predicts that the input data X belongs to class c, and n is the number of base models;
[0026] The category with the highest predicted probability is selected as the final predicted category data.
[0027] In some preferred embodiments, the base model is XGBoost, LightGBM, GBM, CatBoost, and Random Forest.
[0028] In another aspect, this invention proposes a well control blowout event prediction system based on hierarchical multi-joint learning. The system, based on a hierarchical multi-joint learning method for predicting well control blowout events, includes:
[0029] A dataset construction module is configured to acquire a dataset and divide the dataset into a training set and a test set.
[0030] The base model construction module is configured to train multiple well control overflow event prediction base models based on the training set to obtain a trained well control overflow event prediction base model.
[0031] The accuracy calculation module is configured to input the test set into multiple trained well control overflow event prediction base models, obtain prediction results, and calculate the accuracy of each trained well control overflow event prediction base model based on the prediction results and label data.
[0032] The feature set acquisition module is configured to calculate the average of multiple accuracies, select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution, and take the union of the feature sets to obtain the final feature set.
[0033] The prediction result acquisition module is configured to integrate the retained model based on the final feature set, construct a voting-based integrated model, and input the data to be predicted into the integrated model to obtain the prediction result of the well control overflow event.
[0034] A third aspect of the present invention provides an electronic device comprising:
[0035] At least one processor; and
[0036] A memory communicatively connected to at least one of the processors; wherein,
[0037] The memory stores instructions that can be executed by the processor to implement the above-described hierarchical multi-joint learning-based well control overflow event prediction method.
[0038] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer instructions for execution by the computer to implement the above-described hierarchical multi-joint learning-based well control overflow event prediction method.
[0039] The beneficial effects of this invention are:
[0040] Improving prediction accuracy: By constructing multiple base models for predicting wellbore spill events and combining the prediction results of each model, an ensemble model can be built using a voting mechanism, which can effectively improve prediction accuracy. This method can fully utilize the advantages of different models and reduce the overfitting or underfitting problems that may exist in a single model.
[0041] Enhancing model robustness: This invention analyzes the contribution of features and selects key features to construct the final feature set, which helps reduce the impact of noisy features on the model and improves the model's robustness and generalization ability.
[0042] Optimize resource utilization: By accurately predicting potential overflow events during well control, measures can be taken in advance to avoid or mitigate such events, thereby saving costs, improving resource utilization, and avoiding unnecessary equipment damage and production interruptions.
[0043] Enhancing safety: Effective prediction of overflow events can help identify potential safety hazards in a timely manner, take preventive measures, reduce safety accidents caused by overflow events, and protect the lives of oilfield workers and the environment.
[0044] Facilitating Decision Making: The forecast results provide oilfield managers with a scientific basis, which helps to formulate more reasonable and effective well control operation plans and emergency plans, thereby improving management efficiency and decision-making level. Attached Figure Description
[0045] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:
[0046] Figure 1 This is a flowchart illustrating a well control overflow event prediction method based on hierarchical multi-joint learning according to the present invention.
[0047] Figure 2 This is a schematic diagram of the structure of a computer system used to implement the methods, systems, and devices of this application. Detailed Implementation
[0048] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.
[0049] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0050] This invention provides a method for predicting wellbore blowout events based on hierarchical multi-joint learning, the method comprising:
[0051] Obtain the dataset and divide it into a training set and a test set;
[0052] The training set is used to train multiple well control overflow event prediction base models to obtain a trained well control overflow event prediction base model.
[0053] The test set is input into multiple trained well control overflow event prediction base models to obtain prediction results. The accuracy of each trained well control overflow event prediction base model is calculated based on the prediction results and label data.
[0054] Calculate the average of multiple accuracies, and select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, and select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution to take the union of the features to obtain the final feature set.
[0055] The retained models are integrated based on the final feature set to construct a voting-based ensemble model. The data to be predicted is input into the ensemble model to obtain the prediction results of the well control overflow event.
[0056] To more clearly illustrate the hierarchical multi-joint learning-based well control overflow event prediction method of this invention, the following will be combined with... Figure 1 The steps in the embodiments of the present invention are described in detail below:
[0057] Obtain the dataset and divide it into a training set and a test set;
[0058] In this embodiment, the total dataset is defined as a dataset containing 9 wells, and is defined as follows:
[0059]
[0060] Where Xi represents the feature vector, Yi represents the label, and indicates whether an overflow event occurred. The dataset is randomly divided into a training set and a test set:
[0061]
[0062] The training set is used to train multiple well control overflow event prediction base models to obtain a trained well control overflow event prediction base model.
[0063] Five base models were selected: M1, M2, M3, M4, and M5, corresponding to XGBoost, LightGBM, GBM, CatBoost, and Random Forest, respectively. Each model Mi was trained using the training set to obtain the trained model.
[0064]
[0065] The test set is input into multiple trained well control overflow event prediction base models to obtain prediction results. The accuracy of each trained well control overflow event prediction base model is calculated based on the prediction results and label data.
[0066] In this embodiment, the accuracy (ACC) of each trained well control overflow event prediction base model is... i The calculation method is as follows:
[0067]
[0068] Among them, Y test For the labeled data in the test set, X test For the feature data in the test set, Y test Let j be the j-th label data in the test set, and X be... test Let j be the j-th feature data in the test set, and M be the feature data. i Let be the i-th base model, and 1 be the indicator function, which takes the value 1 when the prediction matches the actual result, and 0 otherwise.
[0069] Calculate the average of multiple accuracies, and select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, and select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution to take the union of the features to obtain the final feature set.
[0070] Calculate the mean accuracy (mean_ACC) for all models:
[0071] Discard models with accuracy below mean_ACC, and retain the set of models with accuracy greater than or equal to mean_ACC:
[0072] {Mi∣ACCi≥mean_ACC}.
[0073] For each retained model Mi, calculate its feature contribution and sort them from highest to lowest. Let the sorted feature set be denoted as . The corresponding contribution level is Importance i,j Define the cumulative contribution C. i,j for:
[0074]
[0075] Among them, Importance i,k This represents the contribution of the k-th feature in the i-th base model.
[0076] Then select the feature set Fi that satisfies the condition that the cumulative contribution reaches or exceeds 80% of the total contribution:
[0077]
[0078] in, Let |X| represent the total contribution of all features in the i-th base model, and let |X| represent the number of features. i，j Let represent the j-th feature in the i-th base model, and 'a' represent the preset percentage.
[0079] In this embodiment, 'a' is preferably 80%.
[0080] That is, in the sorted feature set, the selection starts from the feature with the highest contribution and continues until the cumulative contribution reaches or exceeds 80%.
[0081] Final feature set
[0082] The final feature set F is obtained by taking the union of the top 80% contribution feature sets of all retained models.
[0083] F = ∪ i F i .
[0084] The retained models are integrated based on the final feature set to construct a voting-based ensemble model. The data to be predicted is input into the ensemble model to obtain the prediction results of the well control overflow event.
[0085] The ensemble model E(X) is constructed as follows:
[0086]
[0087] Where Vote is the voting mechanism function. Let X be the k-th retained base model, and let X be the input data.
[0088] Soft voting mechanism formula:
[0089] For a given input data X, the predicted probability of the ensemble model E is:
[0090]
[0091] in:
[0092] P(c|X) is the probability that the ensemble model predicts the input data belongs to category c;
[0093] n is the number of base models;
[0094] Pi(clX) is the probability that the input data belongs to category c predicted by the i-th base model.
[0095] Final prediction category The category with the highest prediction probability:
[0096]
[0097] Evaluate the prediction accuracy (ACC) of the ensemble model E on the test set. E :
[0098]
[0099] ACC E The superior accuracy compared to the highest accuracy of all individual base models indicates that the integrated model performs better in predicting well control overflow events.
[0100] The process for using this method in practice is as follows:
[0101] We obtained a total of 9 wells and divided the dataset. The training dataset contains data from 6 wells, while the test dataset contains data from the other 3 wells. Initially, the total dataset had 77 features, as shown in Table 1.
[0102] Table 1:
[0103] train test Number of wells 6 3 Data volume 334722 83399
[0104] Five base models were selected: M1, M2, M3, M4, and M5, corresponding to XGBoost, LightGBM, GBM, CatBoost, and Random Forest, respectively. Each model Mi was trained using the training set to obtain the trained model. The accuracy results for each model are shown in Table 2.
[0105] Table 2:
[0106] Model XGBoost LightGBM GBM CatBoost Random Forest accuracy 0.857 0.737 0.828 0.849 0.849
[0107] The average accuracy of these five models was obtained:
[0108] mean_ACC = 0.824
[0109] Then, base models with below-average accuracy (in this case, LightGBM) are removed.
[0110] For each model that is retained, a feature importance score is calculated, and the set of features that contribute the most to the model is selected.
[0111] Take the union of the feature sets of each model as the final total feature.
[0112] The overall characteristic here is:
[0113] {'bit_depth', 'inlet_temperature', 'rotary_speed', 'inlet_conductance', 'outlet_conductance', 'outlet_temperature', 'inlet_density', 'drill_time' ,'casing_pressure','propane','standpipe_pressure','standard_well_depth','pool_capacity6','pool_capacity5','pump_speed1','bottoms_up_depth'}
[0114] The corresponding Chinese values for these features are as follows: {Drill bit position, inlet temperature, rotary table speed, inlet conductivity, outlet conductivity, outlet temperature, inlet density, drilling time, casing pressure, propane, stand pressure, standard well depth, total pool volume 6, total pool volume 5, pump stroke count 1, late arrival well depth}.
[0115] Using ensemble learning (based on the four base models mentioned above) to train the data, the accuracy on the test set was 0.895, which is higher than that of each of the previous base models.
[0116] Comparative experiment:
[0117] 1. Without model elimination, conduct the complete experiment using all models:
[0118] With 24 features, the final accuracy was 0.8471.
[0119] 2. Ablation experiments on characteristic proportions, not just 80% of the characteristics, as shown in Table 3:
[0120] Table 3:
[0121] Feature proportions of each model Final number of features Accuracy obtained from ensemble learning 0.5 5 0.8366 0.55 6 0.8658 0.6 8 0.8461 0.65 10 0.7924 0.7 12 0.8886 0.75 14 0.8876 0.8 16 0.8956 0.85 18 0.8830 0.9 23 0.8733 0.95 27 0.8871 1 77 0.8808
[0122] As shown above, we reduced the number of features from 77 to 16 and improved the accuracy.
[0123] Although the steps in the above embodiments are described in the above order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not need to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple variations are all within the protection scope of this invention.
[0124] A second embodiment of the present invention provides a well control blowout event prediction system based on hierarchical multi-joint learning, which is based on the well control blowout event prediction method based on hierarchical multi-joint learning described in the first embodiment. The system includes:
[0125] A dataset construction module is configured to acquire a dataset and divide the dataset into a training set and a test set.
[0126] The base model construction module is configured to train multiple well control overflow event prediction base models based on the training set to obtain a trained well control overflow event prediction base model.
[0127] The accuracy calculation module is configured to input the test set into multiple trained well control overflow event prediction base models, obtain prediction results, and calculate the accuracy of each trained well control overflow event prediction base model based on the prediction results and label data.
[0128] The feature set acquisition module is configured to calculate the average of multiple accuracies, select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution, and take the union of the feature sets to obtain the final feature set.
[0129] The prediction result acquisition module is configured to integrate the retained models based on the final feature set, construct a voting-based ensemble model, and input the data to be predicted into the ensemble model to obtain the prediction result of the well control overflow event. Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related explanations of the system described above can be found in the corresponding processes of the foregoing method embodiments, and will not be repeated here.
[0130] It should be noted that the hierarchical multi-joint learning-based well control overflow event prediction system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the various modules or steps and are not considered as an improper limitation of the present invention.
[0131] An electronic device according to a third embodiment of the present invention includes:
[0132] At least one processor; and
[0133] A memory communicatively connected to at least one of the processors; wherein,
[0134] The memory stores instructions that can be executed by the processor to implement the above-described hierarchical multi-joint learning-based well control overflow event prediction method.
[0135] A fourth embodiment of the present invention provides a computer-readable storage medium storing computer instructions, which are executed by the computer to implement the above-described hierarchical multi-joint learning-based well control overflow event prediction method.
[0136] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the storage device and processing device described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0137] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. The programs corresponding to the software modules and method steps can be placed in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. To clearly illustrate the interchangeability of electronic hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the invention.
[0138] The following is for reference. Figure 2 It shows a schematic diagram of the structure of a computer system for implementing the methods, systems, and devices of this application. Figure 2 The server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.
[0139] like Figure 2 As shown, the computer system includes a Central Processing Unit (CPU) 201, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 202 or programs loaded from storage section 208 into Random Access Memory (RAM) 203. The RAM 203 also stores various programs and data required for system operation. The CPU 201, ROM 202, and RAM 203 are interconnected via a bus 204. An Input / Output (I / O) interface 205 is also connected to the bus 204.
[0140] The following components are connected to I / O interface 205: an input section 206 including a keyboard, mouse, etc.; an output section 207 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 208 including a hard disk, etc.; and a communication section 209 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 209 performs communication processing via a network such as the Internet. A drive 210 is also connected to I / O interface 205 as needed. Removable media 211, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 210 as needed so that computer programs read from them can be installed into storage section 208 as needed.
[0141] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 209, and / or installed from removable medium 211. When the computer program is executed by central processing unit (CPU) 201, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.
[0142] Computer program code for performing the operations of this application can be written in one or more programming languages or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, and C++, and conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0143] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0144] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.
[0145] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.
[0146] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.
Claims
1. A method for predicting well control overflow events based on hierarchical multi-joint learning, characterized in that, The method includes: Obtain the dataset and divide it into a training set and a test set; The training set is used to train multiple well control overflow event prediction base models to obtain a trained well control overflow event prediction base model. The test set is input into multiple trained well control overflow event prediction base models to obtain prediction results. The accuracy of each trained well control overflow event prediction base model is calculated based on the prediction results and label data. Calculate the average of multiple accuracies, and select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, and select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution to take the union of the features to obtain the final feature set. The retained models are integrated based on the final feature set to construct a voting-based ensemble model. The data to be predicted is input into the ensemble model to obtain the prediction results of the well control overflow event.
2. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 1, characterized in that, Accuracy (ACC) of each trained base model for predicting wellbore blowout events i The calculation method is as follows: Among them, Y test For the labeled data in the test set, X test For the feature data in the test set, Y test Let j be the j-th label data in the test set, and X be... test Let j be the j-th feature data in the test set, and M be the feature data. i Let be the i-th base model, and 1 be the indicator function, which takes the value 1 when the prediction matches the actual result, and 0 otherwise.
3. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 1, characterized in that, The cumulative contribution C i,j The calculation method is as follows: Among them, Importance i,k This represents the contribution of the k-th feature in the i-th base model.
4. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 3, characterized in that, Select a set of features F whose cumulative contribution reaches or exceeds a preset percentage of the total contribution. i The method is as follows: in, Let |X| represent the total contribution of all features in the i-th base model, and let |X| represent the number of features. i，j Let represent the j-th feature in the i-th base model, and 'a' represent the preset percentage.
5. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 1, characterized in that, The ensemble model E(X) is constructed as follows: Where Vote is the voting mechanism function. Let X be the k-th retained base model, and let X be the input data.
6. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 1, characterized in that, The prediction results include category data, which is obtained as follows: Where P(c|X) is the probability that the ensemble model predicts that the input data X belongs to class c, Pi(c|X) is the probability that the i-th base model predicts that the input data X belongs to class c, and n is the number of base models; The category with the highest predicted probability is selected as the final predicted category data.
7. The well control overflow event prediction method based on hierarchical multi-joint learning according to claim 1, characterized in that, The base models are XGBoost, LightGBM, GBM, CatBoost, and Random Forest.
8. A well control overflow event prediction system based on hierarchical multi-joint learning, based on the well control overflow event prediction method based on hierarchical multi-joint learning as described in any one of claims 1-7, characterized in that, The system includes: A dataset construction module is configured to acquire a dataset and divide the dataset into a training set and a test set. The base model construction module is configured to train multiple well control overflow event prediction base models based on the training set to obtain a trained well control overflow event prediction base model. The accuracy calculation module is configured to input the test set into multiple trained well control overflow event prediction base models, obtain prediction results, and calculate the accuracy of each trained well control overflow event prediction base model based on the prediction results and label data. The feature set acquisition module is configured to calculate the average of multiple accuracies, select the trained well control overflow event prediction base models with accuracies higher than the average as retained models; based on each retained model, calculate the contribution of each feature in the dataset, sort them in descending order, select the feature set whose cumulative contribution reaches or exceeds the preset percentage of the total contribution, and take the union of the feature sets to obtain the final feature set. The prediction result acquisition module is configured to integrate the retained model based on the final feature set, construct a voting-based integrated model, and input the data to be predicted into the integrated model to obtain the prediction result of the well control overflow event.
9. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor to implement the well control overflow event prediction method based on hierarchical multi-joint learning as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are executed by the computer to implement the well control overflow event prediction method based on hierarchical multi-joint learning as described in any one of claims 1-7.