Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm
Abstract Oil well productivity capacity is an important parameter in oilfield development, which is of great significance for efficient development. Traditional oil well productivity capacity prediction methods have a series of problems, such as limited application scope, large prediction errors, di...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2024-10-01
|
| Series: | Journal of Petroleum Exploration and Production Technology |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s13202-024-01873-w |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841559794222104576 |
|---|---|
| author | Kuiqian Ma Chunxin Wu Yige Huang Pengfei Mu Peng Shi |
| author_facet | Kuiqian Ma Chunxin Wu Yige Huang Pengfei Mu Peng Shi |
| author_sort | Kuiqian Ma |
| collection | DOAJ |
| description | Abstract Oil well productivity capacity is an important parameter in oilfield development, which is of great significance for efficient development. Traditional oil well productivity capacity prediction methods have a series of problems, such as limited application scope, large prediction errors, difficulty in characterizing changes under the influence of multiple factors. Aiming at these problems, a well productivity prediction method based on machine learning algorithm was proposed. Taking Bohai X oilfield as the research object, 12 factors affecting oil well productivity capacity were selected from three aspects: geology, engineering, and production. The degree of each factor influence on oil well productivity capacity was analyzed by using the mean decrease impurity (MDI) method, the feature parameters were sequentially excluded, and redundant features that do not affect the prediction accuracy of the model were removed. And then support vector machine (SVM) optimized by improved whale optimization algorithm (IWOA) was used to establish prediction model for oil well productivity capacity. The results show that the main control factors of oil well productivity capacity are: permeability, porosity, effective thickness, pressure draw-down, perforation thickness, fracturing sand addition amount, resistivity, oil saturation, sand addition strength and shale content. The model based on SVM optimized by the improved whale algorithm have an average error of 9.3%, while the model based on SVM optimized by grid search and whale algorithm have bigger errors, which are 21.7% and 15.7% respectively. Residual sum of squares (R2) values for SVM optimized by grid search optimization, whale algorithm and improved whale algorithm are 0.372, 0.939 and 0.941 respectively. The model based on SVM optimized by the improved whale algorithm has higher accuracy in predicting oil well productivity capacity. Compared with existing literature, the MDI method was used to optimize the factors affecting oil well productivity, and IWOA was used to improve the accuracy of oil well productivity capacity prediction. The research results can provide reference for the well productivity capacity prediction. |
| format | Article |
| id | doaj-art-31ba65a3caed4fb4aae0066f1bdc966f |
| institution | Kabale University |
| issn | 2190-0558 2190-0566 |
| language | English |
| publishDate | 2024-10-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Journal of Petroleum Exploration and Production Technology |
| spelling | doaj-art-31ba65a3caed4fb4aae0066f1bdc966f2025-01-05T12:09:21ZengSpringerOpenJournal of Petroleum Exploration and Production Technology2190-05582190-05662024-10-0114123251326010.1007/s13202-024-01873-wOil well productivity capacity prediction based on support vector machine optimized by improved whale algorithmKuiqian Ma0Chunxin Wu1Yige Huang2Pengfei Mu3Peng Shi4Tianjin Bohai Branch of CNOOC (China) LimitedTianjin Bohai Branch of CNOOC (China) LimitedTianjin Bohai Branch of CNOOC (China) LimitedTianjin Bohai Branch of CNOOC (China) LimitedTianjin Bohai Branch of CNOOC (China) LimitedAbstract Oil well productivity capacity is an important parameter in oilfield development, which is of great significance for efficient development. Traditional oil well productivity capacity prediction methods have a series of problems, such as limited application scope, large prediction errors, difficulty in characterizing changes under the influence of multiple factors. Aiming at these problems, a well productivity prediction method based on machine learning algorithm was proposed. Taking Bohai X oilfield as the research object, 12 factors affecting oil well productivity capacity were selected from three aspects: geology, engineering, and production. The degree of each factor influence on oil well productivity capacity was analyzed by using the mean decrease impurity (MDI) method, the feature parameters were sequentially excluded, and redundant features that do not affect the prediction accuracy of the model were removed. And then support vector machine (SVM) optimized by improved whale optimization algorithm (IWOA) was used to establish prediction model for oil well productivity capacity. The results show that the main control factors of oil well productivity capacity are: permeability, porosity, effective thickness, pressure draw-down, perforation thickness, fracturing sand addition amount, resistivity, oil saturation, sand addition strength and shale content. The model based on SVM optimized by the improved whale algorithm have an average error of 9.3%, while the model based on SVM optimized by grid search and whale algorithm have bigger errors, which are 21.7% and 15.7% respectively. Residual sum of squares (R2) values for SVM optimized by grid search optimization, whale algorithm and improved whale algorithm are 0.372, 0.939 and 0.941 respectively. The model based on SVM optimized by the improved whale algorithm has higher accuracy in predicting oil well productivity capacity. Compared with existing literature, the MDI method was used to optimize the factors affecting oil well productivity, and IWOA was used to improve the accuracy of oil well productivity capacity prediction. The research results can provide reference for the well productivity capacity prediction.https://doi.org/10.1007/s13202-024-01873-wWhale algorithmSupport vector machineMachine learningProductivity capacity predictionFeature selection |
| spellingShingle | Kuiqian Ma Chunxin Wu Yige Huang Pengfei Mu Peng Shi Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm Journal of Petroleum Exploration and Production Technology Whale algorithm Support vector machine Machine learning Productivity capacity prediction Feature selection |
| title | Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| title_full | Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| title_fullStr | Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| title_full_unstemmed | Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| title_short | Oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| title_sort | oil well productivity capacity prediction based on support vector machine optimized by improved whale algorithm |
| topic | Whale algorithm Support vector machine Machine learning Productivity capacity prediction Feature selection |
| url | https://doi.org/10.1007/s13202-024-01873-w |
| work_keys_str_mv | AT kuiqianma oilwellproductivitycapacitypredictionbasedonsupportvectormachineoptimizedbyimprovedwhalealgorithm AT chunxinwu oilwellproductivitycapacitypredictionbasedonsupportvectormachineoptimizedbyimprovedwhalealgorithm AT yigehuang oilwellproductivitycapacitypredictionbasedonsupportvectormachineoptimizedbyimprovedwhalealgorithm AT pengfeimu oilwellproductivitycapacitypredictionbasedonsupportvectormachineoptimizedbyimprovedwhalealgorithm AT pengshi oilwellproductivitycapacitypredictionbasedonsupportvectormachineoptimizedbyimprovedwhalealgorithm |