Разработка алгоритмов раннего прогнозирования нестандартных ситуаций при бурении скважин (Development of algorithms for predictive alarming on non-standard situations at well drilling) тема диссертации и автореферата по ВАК РФ 00.00.00, кандидат наук Гурина Екатерина Викторовна

• Гурина Екатерина Викторовна
• кандидат наук
• 2024

Оглавление диссертации кандидат наук Гурина Екатерина Викторовна

Content

INTRODUCTION

CHAPTER 1. OVERVIEW OF DRILLING PROCESS

1.1. Phases of drilling process

1.2. Data acquisition and monitoring of drilling process

1.3. Data quality issues in mud logging data

1.4. Classification of drilling problems and accidents

1.4.1. Pipe sticking

1.4.2. Drilling bit and pipe failures

1.4.3. Integrity damage of well walls

1.4.4. Shale collars (bit balling)

1.4.5. Fluid-, oil-, gas- or water shows (kicks)

1.4.6. Mud loss (lost circulation)

1.4.7. Other failures

1.5. State-of-art solutions for drilling accident forecasting problem

1.6. SUMMARY

CHAPTER 2. DATASET COLLECTION AND PREPARATION

2.1. OVERVIEW OF AVAILABLE DATASET

2.2. Data preprocessing

2.3. Data sampling

2.4. Summary

CHAPTER 3. FORECASTING OF DRILLING ACCIDENTS

3.1. State-of-art methods for time-series comparison

3.2. Description of classification algorithms

3.3. Methods for model validation

3.4. Bag-of-features model

3.4.1. Methodology of Bag-of-features solution

3.4.2. Selection of the main hyperparameters

3.5. Interval-based feature extraction model

3.5.1. Methodology of statistical model

3.5.2. Selection of the main hyperparameters

3.6. Classification problem statement

3.7. General quality metrics

5

3.8. Robustness of Bag-of-features model

3.9. Summary

CHAPTER 4. INTERPRETABILITY OF BAG-OF-FEATURES MODEL

4.1. STATE-OF-THE ART TECHNIQUES FOR BLACK box MODEL INTERPRETABILITY

4.2. METHODOLOGY OF BAG-OF-FEATURES INTERPRETATION

4.3. Interpretation methods used in the current study

4.4. Validation procedures

4.4.1. State-of-art methods for interpretability validation

4.4.2. Assessment of interpretability quality

4.5. General interpretation quality

4.6. Consistency of explanatory model

4.7. Summary

CHAPTER 5. APPLICATION OF BAG-OF-FEATURES IN INDUSTRY

5.1. General structure of the AIDrilling system

5.2. THE RESULTS OF THE BAG-OF-FEATURES MODEL WITHIN THE AIDRILLING SYSTEM

5.3. SUMMARY

CONCLUSIONS

RECOMMENDATIONS FOR FUTURE RESEARCH

PUBLISHED PAPERS

ACKNOWLEDGMENTS

LIST OF ABBREVIATIONS

REFERENCES

APPENDIX A. BOX-PLOTS BEFORE ANOMALY TRANSFORMATION

APPENDIX B. HYPERPARAMETERS TUNING OF BAG-OF-FEATURES MODEL

APPENDIX C. TSNE FOR THE CONSISTENCY OF INTERPRETABILITY MODEL

Введение диссертации (часть автореферата) на тему «Разработка алгоритмов раннего прогнозирования нестандартных ситуаций при бурении скважин (Development of algorithms for predictive alarming on non-standard situations at well drilling)»

INTRODUCTION

A significant proportion of the investment and capital expenditures of oil and gas companies falls on the drilling of wells during the development of fields [29], as well as during the exploration phase of field development. However, drilling oil and gas wells are usually accompanied by uncertain information about the geological conditions of drilling, including the necessary information about rocks and fluids in their natural state, which is essential for successful drilling.

Drilling the first exploratory wells in the area, recognized as promising for oil and gas deposits, is carried out based on geophysical surveys and structural exploration drilling, which usually gives only an idea about deposit structure with some probability. The degree of information reliability increases with the increase in the number of wells drilled, the level of research carried out in them, and the quality of data processing.

However, regardless of the well's construction technology level and available information, unusual situations and problems (complications) inevitably happen during drilling. The share of the costs for eliminating drilling problems can be significant and is mainly determined by the complexity of drilling geological conditions. Therefore, understanding and anticipating them and their causes are necessary for overall well-cost control and reaching the target zone.

The main reasons why unusual situations occur are insufficiently well-studied geological conditions and the human factor. For example, the difference in expected and real thickness of the geological layer, physical and chemical properties of the rocks, and inaccurate lithology forecast may lead to some severe accidents, as stucks or fluid shows. in addition, drilling failures can be exacerbated by organizational reasons: low level of technology, insufficient and untimely provision of high-quality tools and materials, and low-performing discipline of the drilling crew operators.

since complications generate additional costs for materials and significantly influence the further operation of wells, one of the trends for further growth in drilling

7

productivity is to reduce non-productive time (NPT) while drilling by preventing accidents and their consequences. It is well known that eliminating an accident is much easier at an early stage of its distribution. However, one problem often entails another one, which complicates the task of eliminating it. For example, the stuck of a drilling tool can result from a slaking-off or tight pull of the drilling column, which has not been detected in advance. Therefore, it is necessary to detect all abnormal drilling behaviour, and by this, we can forecast or at least detect most of the drilling problems. Using such a scheme, the drilling process can be normalized with minimal loss of working time by timely remedial interventions.

As mentioned, interest in forecasting methods for drilling oil and gas wells is expanding while the complexity of designed wells and depths is increasing. Currently, there are several approaches for predicting complications during drilling; however, interest in developing such models remains strong. That is why developing a product that can predict accidents during drilling operations is significant for the oil and gas industry.

Aim and objectives. The dissertation's main aim is to develop an algorithm for detecting both pre-accident (precursors intervals) and drilling accident patterns for the most common types of drilling accidents using machine learning or deep learning techniques.

To achieve the stated aim following objectives have been stated:

1. Perform a comprehensive literature review of the state-of-the-art solutions used in the industry. Current objectives include an overview of the real-time drilling data used by engineers to identify accidents; drilling accident classifications; state-of-the-art approaches for the drilling accidents forecasting problem.

2. Collect a dataset of accidents and their precursors' intervals, analyse data quality issues presented in the collected dataset, and develop preprocessing module fixing the main quality problems.

3. plan and develop models that use collected data and transform it into a different feature space where the accident patterns can be distinguished from the normal drilling operations. Analyse the results obtained in different experimental settings and the robustness of applied methods.

4. plan and develop an interpretability model that uses the best-developed model that forecasts accidents and explains why the model decided that the particular accident would happen.

5. Test the developed model during real-time drilling operations using the infrastructure of the industry partner.

Methods. The research utilizes methods of machine learning including deep learning, petroleum engineering and well construction theory, methods of model agnostic meta- learning theory. The algorithms were developed in python programming language.

Novelty. This research is the first-ever comprehensive study of machine learning and deep learning techniques for forecasting abnormal events of different types during drilling in real-time. in particular, for the first time, a universal solution has been developed that is able to forecast six most common types of drilling accidents in realtime: kick, stuck, loss circulation, washout of drilling pipe, shale collar, and break of drilling pipe. For the first time, an interpretability model has been developed to explain the results of accident forecasting model in real time and visualize the logic behind the forecasting model outcome.

Practical value. The models and corresponding software allow reducing nonproductive time and failure mitigation costs during the real life well construction process. The obtained results confirm that the use of the developed algorithms during well construction can reduce non-productive time by up to 15%. In addition, the results of the presented study determine principles for telemetry data collection.

Scientific statements submitted for defense. Scientific statements submitted for defense. The following statements are submitted for defense:

1. A comprehensive model for predicting the six most common accidents during the drilling of oil and gas wells, operating in real-time based on surface telemetry data and machine learning methods

2. A model for interpreting the forecast of emergency situations during the drilling of oil and gas wells

3. Methodologies for assessing the quality of models for both accident forecasting and interpretation

Personal contribution. All the results of the dissertation were obtained personally by the applicant or with his direct involvement. In particular, the applicant performed the search and analysis of the literature related to the research topic. The applicant together with supervisor, PhD in physics and math, D.Koroteev and co-supervisor, PhD in technical sciences, K.Antipova participated in the formulation of aims and objectives of the dissertation and developed experimental methods. The results of the work were obtained personally by the author or with his direct participation. The development of a software product based on the developed models for forecasting non-standard situations and interpreting the forecast was carried out by the team of the Digital Petroleum Company with the direct participation of the author.

Publications and conferences. The statements and conclusions formulated in the dissertation have received qualified approbation at international SPE and EAGE scientific conferences. The credibility is also confirmed by the publication of research results in five articles in peer-reviewed scientific journals. The publications are fully consistent with the topic of the dissertation research and provide nearly complete coverage of the dissertation content.

Заключение диссертации по теме «Другие cпециальности», Гурина Екатерина Викторовна

CONCLUSIONS

Minimizing non-productive time during drilling by forecasting accidents is one of the most significant tasks during well construction. The current study aimed to develop a unified algorithm for forecasting and detecting pre-accident and accident patterns during the well construction process for the most common types of drilling accidents using artificial intelligence techniques.

The main research results are the following:

1. A literature review was performed. It includes description of classification, patterns and physics of the drilling accidents, different data sources available during drilling, data quality issues, SOTA solutions for drilling accident forecasting problem, time-series representations and interpretability techniques for black-box models. Based on research results obtained through summarizing and analysing existing state-of-art solutions for drilling accident forecasting, it was shown that future algorithm should be trained on the real data from different wells, shows statistically significant results, can be applied without additional model training in real-time and notifies drilling engineers about the possible accident in advance.

2. The main steps of the data collection, sampling and preprocessing steps required for further problem-solving were demonstrated. The collected the dataset includes 180 cases of real drilling accident intervals, 114 pre-accident segments and 1290 segments of normal drilling behaviour happened in open hole section. To solve data quality issues, the preprocessing module was developed.

3. Several models that are able to forecast and detect drilling accidents were proposed and developed. The ablation study of the model's primary hyperparameters was performed for each method. The best-developed solution is based on Bag-of-features approach. Model can forecast 70% of drilling accidents with a FPR equal to 30%. At the same time, it can detect 83%of drilling accidents. The main limitations of model were experimentally investigated.

4. The methodology and approach for interpreting the Bag-of-features model results were proposed. The interpretability model that uses the best-developed model that forecasts accidents was developed. It explains why the model decided that the particular accident would happen. Explanatory model has 15% precision at 70% recall and overcomes the metric values of a random baseline and multi-head attention neural network. It was shown that provided explanations are consistent with the physics of the drilling process.

5. The developed models were tested during real-time drilling operations via the integration of Bag-of-features model into AIDrilling software. The obtained research results were also confirmed with real surface telemetry

RECOMMENDATIONS FOR FUTURE RESEARCH

Drilling accident prediction is an essential task in well construction. Drilling support software allows observing the drilling parameters for multiple wells simultaneously, and artificial intelligence helps detect the drilling accident predecessor ahead of the emergency. The developed Bag-of-features and interpretation models are core parts of the AIDrilling system that can minimize non-productive time while drilling using the surface telemetry data. The system was successfully tested in real oilfields in Russia.

However, there are many things to improve. First, it is necessary to continue collecting data on various accidents retrain the model and see how its quality changes depending on the database sample size for each accident type. Moreover, when a sufficient number of cases are available, it will be possible to train generative networks, which will provide synthetic telemetry data. Such a procedure will further expand the database size for different accident types.

In addition, several additional steps in preprocessing are required to improve the preprocessing module and increase the quality of the input signals. The logical checks for the presence of several input values of data should be expanded. For example, when the WOB parameter increases, the TQA parameter should also increase with the same trend. Such tests allow for increasing the physical meaning of input signals. Moreover, it would be interesting to consider different input data quality evaluation schemes and check how the quality of forecasting and interpretability models depend on them.

Further algorithms development is also required. First, it is necessary to obtain model quality using other metrics that might indicate the ratio of true forecasted events and the particular number of false alarms per day for each accident type independently. It also might be a good strategy for the objective function during model training to assign higher penalties for false negatives of more severe drilling accidents. The different target-assigning methodologies and sampling techniques should also be considered and

compared. Since most of the patterns for each accident type are not similar, it might also be helpful to consider several independent forecasting models with different time-series representations tuned for the particular accident type to achieve better forecasting quality.

Including more physical meaning into the model using post-processing filters or additional developed features is also essential. For example, it is interesting to analyse how model quality will change if the model includes not only the telemetry parameters but also the calculated parameters using, for example, Torque and Drag models or features, showing the size or type of drilling equipment presented in the well at each time moment.

A broad area of research in the field of forecast interpretation is still uncovered. First, it is necessary to test the existing interpretation algorithm with the highest number of experts, consider different schemes of interpretation quality calculation, and develop rules for evaluating the provided feedback by the drilling engineers. Moreover, it is interesting to research the predictive model whose interpretation is as close as possible to the drilling engineer reference. Finally, considering the current explanatory model, which uses features from the Bag-of-features model, it also might be helpful to provide an explanation not only highlighting segments but also with a general description of patterns presented in a particular cluster.

As a recommendation for further AIDrilling system development, it is also necessary to check whether it is possible to increase the speed of system performance using different principles of microservices operations and decrease the memory using modern databases and different principles of data storage.

Список литературы диссертационного исследования кандидат наук Гурина Екатерина Викторовна, 2024 год

REFERENCES

1. Abanda A., Mori U., Lozano J. A. A review on distance based time series classification //Data Mining and Knowledge Discovery. - 2019. - T. 33. - №. 2. -C. 378-412.

2. Abu-Abed F. N. Application of neural network modelling tools for analysis of pre-emergency situations on drilling sites //Software Products and Systems. - 2010. -№. 3. - C. 136-139.

3. Abu-Abed F. N. Automated system for detection of pre-emergency situations at oil and gas industry facilities //Logger. - 2015. - T. 5. - №. 251. - C. 48-61.

4. Bourgoyne A. T. et al. Applied drilling engineering. - Richardson : Society of Petroleum Engineers, 1986. - T. 2. - C. 514.

5. Agarwal R. et al. Neural additive models: Interpretable machine learning with neural nets //Advances in neural information processing systems. - 2021. - T. 34.

- c. 4699-4711.

6. Ahmad M. W., Mourshed M., Rezgui Y. Trees vs Neurons: Comparison between random forest and ANN for high-resolution prediction of building energy consumption //Energy and buildings. - 2017. - T. 147. - C. 77-89.

7. Aljubran M. et al. Deep learning and time-series analysis for the early detection of lost circulation incidents during drilling operations //IEEE Access. - 2021. - T. 9.

- C. 76833-76846.

8. Alotaibi B., Aman B., Nefai M. Real-Time Drilling Models Monitoring Using Artificial Intelligence //SPE Middle East Oil and Gas Show and Conference. -OnePetro, 2019. URL: https://onepetro.org/SPEMEOS/proceedings-abstract/19MEOS/2-19MEOS/D021S001R004/218417.

9. Arnaout A. et al. Intelligent data quality control of real-time rig data //SPE Middle East Intelligent Oil and Gas Symposium. - SPE, 2013. - C. D021S006R001.

10. Artemov A., Burnaev E. Ensembles of detectors for online detection of transient changes //Eighth International Conference on Machine Vision (ICMV 2015). -SPIE, 2015. - T. 9875. - C. 413-417.

11. Bao Y., Liu Z. A fast grid search method in support vector regression forecasting time series //Intelligent Data Engineering and Automated Learning-IDEAL 2006: 7th International Conference, Burgos, Spain, September 20-23, 2006. Proceedings 7. - Springer Berlin Heidelberg, 2006. - C. 504-511..

12. Basarigin U , Bulatov A, Proselkov U,. Complications and accidents during drilling of oil and gas wells, 2000. URL: http://petrolibrary.ru/basaryigin-yu-m-texnologiya-bureniya-neftyanyix-i-gazovyix-skvazhin.html .

13. Begum N., Keogh E. Rare time series motif discovery from unbounded streams //Proceedings of the VLDB Endowment. - 2014. - T. 8. - №. 2. - C. 149-160.

14. Belkina A. C. et al. Automated optimized parameters for T-distributed stochastic neighbour embedding improve visualization and analysis of large datasets //Nature communications. - 2019. - T. 10. - №. 1. - C. 5415.

15. Bengio Y., Goodfellow I., Courville A. Deep learning. - Cambridge, MA, USA : MIT press, 2017. - T. 1. - C. 326-416.

16. Bergstra J., Yamins D., Cox D. Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vision architectures //International conference on machine learning. - PMLR, 2013. - C. 115-123.

17. Borozdin S. et al. Drilling problems forecast system based on neural network //SPE Annual Caspian Technical Conference. - SPE, 2020. - C. D023S013R008.

18. Bostrom A., Bagnall A. Binary shapelets transform for multiclass time series classification //Transactions on Large-Scale Data-and Knowledge-Centered

Systems XXXII: Special Issue on Big Data Analytics and Knowledge Discovery. - 2017. - C. 24-46.

19. Braga D. C. Field Drilling Data Cleaning and Preparation for Data Analytics Applications. - Louisiana State University and Agricultural & Mechanical College, 2019. URL: https://repository.lsu.edu/gradschool_theses/4952.

20. Buckland M., Gey F. The relationship between recall and precision //Journal of the American society for information science. - 1994. - T. 45. - №. 1. - C. 12-19.

21. Burnaev E., Erofeev P., Papanov A. Influence of resampling on accuracy of imbalanced classification //Eighth international conference on machine vision (ICMV 2015). - SPIE, 2015. - T. 9875. - C. 423-427.

22. Burns B. et al. Borg, omega, and kubernetes //Communications of the ACM. -2016. - T. 59. - №. 5. - C. 50-57.

23. Carvalho D. V., Pereira E. M., Cardoso J. S. Machine learning interpretability: A survey on methods and metrics //Electronics. - 2019. - T. 8. - №. 8. - C. 832.

24. Chawla N. V. C4. 5 and imbalanced data sets: investigating the effect of sampling method, probabilistic estimate, and decision tree structure //Proceedings of the ICML. - Toronto, ON, Canada : CIBC, 2003. - T. 3. - C. 66..

25. Cheng Y. Mean shift, mode seeking, and clustering //IEEE transactions on pattern analysis and machine intelligence. - 1995. - T. 17. - №. 8. - C. 790-799..

26. Claesen M., De Moor B. Hyperparameter search in machine learning //arXiv preprint arXiv:1502.02127. - 2015. URL: https://arxiv.org/abs/1502.02127

27. da Silva R. G. et al. Six years operating a real time drilling problem detection software in deepwater environments: Results and challenges //SPE Latin American and Caribbean Petroleum Engineering Conference. - OnePetro, 2020, DOI: 10.2118/199077-MS.

28. Day W. H. E., Edelsbrunner H. Efficient algorithms for agglomerative hierarchical clustering methods //Journal of classification. - 1984. - T. 1. - C. 7-24.

29. de Almeida Jr I. N. et al. A review of telemetry data transmission in unconventional petroleum environments focused on information density and reliability //Journal of Software Engineering and Applications. - 2015. - T. 8. - №. 09. - C. 455.

30. Deeks N. R., Halland T. WITSML changing the face of real-time //SPE Intelligent Energy International Conference and Exhibition. - SPE, 2008. - C. SPE-112016-MS.

31. Deng H. et al. A time series forest for classification and feature extraction //Information Sciences. - 2013. - T. 239. - C. 142-153.

32. Doshi-Velez F., Kim B. Towards a rigorous science of interpretable machine learning //arXiv preprint arXiv:1702.08608. - 2017. URL: https://arxiv.org/abs/1702.08608

33. ElShawi R. et al. Interpretability in healthcare: A comparative study of local machine learning interpretability techniques //Computational Intelligence. - 2021. - T. 37. - №. 4. - C. 1633-1650.

34. Ismail Fawaz H. et al. Deep learning for time series classification: a review //Data mining and knowledge discovery. - 2019. - T. 33. - №. 4. - C. 917-963.

35. Ferreira A. P. L. A. et al. Automated decision support and expert collaboration avoid stuck pipe and improve drilling operations in offshore Brazil subsalt well //Offshore Technology Conference. - OTC, 2015. - C. OTC-25838-MS.

36. Friedman J. et al. Pathwise coordinate optimization. - 2007. - T. 1. - C. 302-332.

37. Fu T. A review on time series data mining //Engineering Applications of Artificial Intelligence. - 2011. - T. 24. - №. 1. - C. 164-181.

38. Geekiyanage S. C. H., Tunkiel A., Sui D. Drilling data quality improvement and information extraction with case studies //journal of petroleum Exploration and Production. - 2021. - Т. 11. - С. 819-837.

39. Gilpin L. H. et al. Explaining explanations: An overview of interpretability of machine learning //2018 IEEE 5th International Conference on data science and advanced analytics (DSAA). - IEEE, 2018. - С. 80-89.

40. Golitsyna M. et al. Automatic Method for Anomaly Detection while Drilling //First EAGE Digitalization Conference and Exhibition. - European Association of Geoscientists & Engineers, 2020. - Т. 2020. - №. 1. - С. 1-5.

41. Goodfellow I., Bengio Y., Courville A. Deep learning. - MIT press, 2016 - C.372-376.

42. Grabocka J. et al. Learning time-series shapelets //Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining. -2014. - С. 392-401.

43. Guidotti R. et al. Local rule-based explanations of black box decision systems //arXiv preprint arXiv:1805.10820. - 2018. DOI: 10.48550/arXiv.1805.10820.

44. Guidotti R. et al. A survey of methods for explaining black box models //ACM computing surveys (CSUR). - 2018. - Т. 51. - №. 5. - С. 1-42.

45. Gurina E. et al. Making the black-box brighter: interpreting machine learning algorithm for forecasting drilling accidents //arXiv preprint arXiv:2209.02256. -2022. DOI: https://doi.org/10.48550/arXiv.2209.02256

46. Hahn D. et al. Reciprocating pulser for mud pulse telemetry : US Patent. 7417920 США. - 2008. URL: https://patents.google.com/patent/US7417920B2/en.

47. Harrell J., Brooks A. G., Morsy H. S. Method and apparatus for mud pulse telemetry in underbalanced drilling systems : US Patent. 6097310 США. - 2000 . URL: https://patents.google.com/patent/US6097310A/en.

48. Hatami N., Gavet Y., Debayle J. Bag of recurrence patterns representation for time-series classification //Pattern Analysis and Applications. - 2019. - T. 22. - C. 877887.

49. Hintze J. L., Nelson R. D. Violin plots: a box plot-density trace synergism //The American Statistician. - 1998. - T. 52. - №. 2. - C. 181-184.

50. Hirschberg S., Spiekerman G., Dones R. Severe accidents in the energy sector. -1998- C. 42-44.

51. Hung C. Y. et al. Comparing deep neural network and other machine learning algorithms for stroke prediction in a large-scale population-based electronic medical claims database //2017 39th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). - IEEE, 2017. - C. 31103113.

52. Hutin R., Tennent R. W., Kashikar S. V. New mud pulse telemetry techniques for deepwater applications and improved real-time data capabilities //SPE/IADC Drilling Conference and Exhibition. - SPE, 2001. - C. SPE-67762-MS.

53. Ismail Fawaz H. et al. Deep learning for time series classification: a review //Data mining and knowledge discovery. - 2019. - T. 33. - №. 4. - C. 917-963.

54. Kaasa G. O. et al. Simplified hydraulics model used for intelligent estimation of downhole pressure for a managed-pressure-drilling control system //SPE Drilling & Completion. - 2012. - T. 27. - №. 01. - C. 127-138.

55. Kamenskikh S.V., Fomin A.S. Complications and accidents when drilling oil and gas wells //Ukhta: USTU. - 2010 - C. 32-33.

56. Kanungo T. et al. An efficient k-means clustering algorithm: Analysis and implementation //IEEE transactions on pattern analysis and machine intelligence. - 2002. - T. 24. - №. 7. - C. 881-892.

57. Karasev D. V., Shcherbinina N. E., Karaseva T. V. Peculiarities of complications during drilling operations at great depth. - 2015 - C.19-31 .

58. Karim F. et al. Multivariate LSTM-FCNs for time series classification //Neural networks. - 2019. - T. 116. - C. 237-245.

59. Kaur H. et al. Interpreting interpretability: understanding data scientists' use of interpretability tools for machine learning //Proceedings of the 2020 CHI conference on human factors in computing systems. - 2020. - C. 1-14.

60. Kirkman M. A. et al. Wellsite Information Transfer Standard Markup Language, WITSML, an Update //SPE Annual Technical Conference and Exhibition. - SPE, 2003. - C. SPE-84066-MS.

61. Klyuchnikov N. et al. Data-driven model for the identification of the rock type at a drilling bit //Journal of Petroleum science and Engineering. - 2019. - T. 178. - C. 506-516.

62. Konik L, ComNews (2022b). NOVATEK reduced drilling accident rate with the help of AI. Available at: https://www.comnews.ru/content/219388/2022-03-23/2022-w12/novatek-snizil-avariynost-bureniya-pri-pomoschi-ai (Accessed: 05 18, 2022).

63. Kozlovskaia N., Zaytsev A. Deep ensembles for imbalanced classification //2017 16th IEEE International Conference on Machine Learning and Applications (ICMLA). - IEEE, 2017. - C. 908-913.

64. Krawczyk B., Wozniak M., Schaefer G. Cost-sensitive decision tree ensembles for effective imbalanced classification //Applied Soft Computing. - 2014. - T. 14. -C. 554-562. Larry W. L., Edward D. H. Petroleum engineering handbook //SPE. -2006. - T. 1. - C. 2006.

65. Lake L. W. et al. Petroleum Engineering Handbook, Vol. 6 //Society of Petroleum Engineers, Texas, USA. - 2006 - C. 20-32.

66. LeCun Y., Bengio Y., Hinton G. Deep learning //nature. - 2015. - T. 521. - №. 7553. - C. 436-444.

67. Li C. W. et al. Drilling mud signal processing based on wavelet //2007 International Conference on Wavelet Analysis and Pattern Recognition. - IEEE, 2007. - Т. 4. -С. 1545-1549.

68. Li H. et al. Adaptively constrained dynamic time warping for time series classification and clustering //Information Sciences. - 2020. - Т. 534. - С. 97-116.

69. Li Q. et al. Bidirectional drill string telemetry for measuring and drilling control : US Patent. 7913773 США. - 2011. URL: https://patents.google.com/patent/CN101253304A/en.

70. Liang Y. et al. Explaining the black-box model: A survey of local interpretation methods for deep neural networks //Neurocomputing. - 2021. - Т. 419. - С. 168182.

71. Lipton Z. C. The mythos of model interpretability: In machine learning, the concept of interpretability is both important and slippery //Queue. - 2018. - Т. 16. - №. 3. - С. 31-57.

72. Malhotra P. et al. TimeNet: Pre-trained deep recurrent neural network for time series classification //arXiv preprint arXiv:1706.08838. - 2017. DOI: 10.48550/arXiv.1706.08838 .

73. Man X., Chan E. P. The best way to select features? comparing mda, lime, and shap //The Journal of Financial Data Science. - 2021. - Т. 3. - №2. 1. - С. 127-139.

74. Mathis W., Thonhauser G. Mastering Real-Time Data Quality Control—How To Measure and Manage the Quality of (Rig) Sensor Data //SPE/IADC Middle East Drilling Technology Conference and Exhibition. - SPE, 2007. - С. SPE-107567-MS.

75. Milon M. H. Comparison on fourier and wavelet transformation for an ecg signal //American Journal of Engineering Research. - 2017. - Т. 6. - №. 8. - С. 01-07.

76. Mitchell R. F., Miska S. Fundamentals of drilling engineering - 2011. - С. 1-14.

77. Mohiuddin M. A. et al. Analysis of wellbore instability in vertical, directional, and horizontal wells using field data //Journal of petroleum Science and Engineering. - 2007. - T. 55. - №. 1-2. - C. 83-92.

78. Mohri M., Rostamizadeh A., Talwalkar A. Foundations of machine learning. -MIT press, 2018 - C. 68-72.

79. Molnar C. Interpretable machine learning. - 2020 - C. 14-28. URL: https://christophm.github.io/interpretable-ml-book.

80. Molnar C., Casalicchio G., Bischl B. Interpretable machine learning-a brief history, state-of-the-art and challenges //Joint European conference on machine learning and knowledge discovery in databases. - Cham : Springer International Publishing, 2020. - C. 417-431.

81. Naraghi M. E., Ezzatyar P., Jamshidi S. Prediction of drilling pipe sticking by active learning method (ALM) //Journal of Petroleum and Gas Engineering. -2013. - T. 4. - №. 07. - C. 173-183.

82. Mwachaka S. M., Wu A., Fu Q. A review of mud pulse telemetry signal impairments modeling and suppression methods //Journal of Petroleum Exploration and Production Technology. - 2019. - T. 9. - C. 779-792.

83. Namuq M. A. Simulation and modeling of pressure pulse propagation in fluids inside drill strings. - 2013- C. 12-13.

84. Newman K. R., Procter R. Analysis of hook load forces during jarring //SPE/IADC Drilling Conference and Exhibition. - SPE, 2009. - C. SPE-118435-MS.

85. Nohara Y., Wakata Y., Nakashima N. Interpreting Medical Information Using Machine Learning and Individual Conditional Expectation //MedInfo. - 2015. - T. 1073. DOI: 10.1038/s41598-021 -92442-2

86. Nori H. et al. Interpretml: A unified framework for machine learning interpretability //arXiv preprint arXiv:1909.09223. - 2019. DOI: 10.48550/arXiv.1909.09223 .

87. Pasic B., Gaurina Medimurec N., Matanovic D. Wellbore instability: causes and consequences //Rudarsko-geolosko-naftni zbornik. - 2007. - T. 19. - №. 1. - C. 87-98.

88. Pastor E., Baralis E. Explaining black box models by means of local rules //Proceedings of the 34th ACM/SIGAPP symposium on applied computing. -2019. - C. 510-517.

89. Peng Q. et al. A real-time warning system for identifying drilling accidents //SPE Annual Caspian Technical Conference and Exhibition. - OnePetro, 2014. URL: https://onepetro .org/SPECTCE/proceedings-abstract/ 14CTCE/All-14CTCE/SPE-172303-MS/212522

90. Potter K. et al. Methods for presenting statistical information: The box plot //VLUDS. - 2006. - C. 97-106.

91. Pustovoitenko L. Warning and methods to eliminate accidents in drilling: a tutorial, 1987- Nedra, 1988. - C.279.

92. Qodirov S., Shestakov A. Development of artificial neural network for predicting drill pipe sticking in real-time well drilling process //2020 Global Smart Industry Conference (GloSIC). - IEEE, 2020. - C. 139-144.

93. Rakthanmanon T., Keogh E. Fast shapelets: A scalable algorithm for discovering time series shapelets //proceedings of the 2013 SIAM International Conference on Data Mining. - Society for Industrial and Applied Mathematics, 2013. - C. 668676.

94. Ramon Y. et al. A comparison of instance-level counterfactual explanation algorithms for behavioral and textual data: SEDC, LIME-C and SHAP-C

//Advances in Data Analysis and Classification. - 2020. - T. 14. - C. 801-819.

121

95. Reckmann H. Downhole noise cancellation in mud-pulse telemetry : US Patent. 8811118. - 2014. URL: https://patents.google.com/patent/US8811118B2/en.

96. Reiber F., Vos B. E., Eide S. E. On-line torque & drag: A real-time drilling performance optimization tool //SPE/IADC Drilling Conference and Exhibition. -SPE, 1999. - C. SPE-52836-MS.

97. Ribeiro M. T., Singh S., Guestrin C. " Why should i trust you?" Explaining the predictions of any classifier //Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining. - 2016. - C. 1135-1144.

98. Ribeiro M. T., Singh S., Guestrin C. Model-agnostic interpretability of machine learning //arXiv preprint arXiv:1606.05386. - 2016. DOI: 10.48550/arXiv.1606.05386.

99. Ribeiro M. T., Singh S., Guestrin C. Anchors: High-precision model-agnostic explanations //Proceedings of the AAAI conference on artificial intelligence. -

2018. - T. 32. - №. 1. DOI: 10.1609/aaai.v32i1.11491 .

100. Romanenkova E. et al. Real-time data-driven detection of the rock-type alteration during a directional drilling //IEEE Geoscience and Remote Sensing Letters. -

2019. - T. 17. - №. 11. - C. 1861-1865.

101. Roy S., Cooper G. A. Prevention of bit balling in shales—Preliminary results //SPE drilling & completion. - 1993. - T. 8. - №. 03. - C. 195-200.

102. Sadlier A., Says I., Hanson R. Automated decision support to enhance while-drilling decision making: where does it fit within drilling automation? //SPE/IADC Drilling Conference. - OnePetro, 2013. DOI: 10.2118/163430-MS

103. Safavian S. R., Landgrebe D. A survey of decision tree classifier methodology //IEEE transactions on systems, man, and cybernetics. - 1991. - T. 21. - №. 3. -C. 660-674.

104. Sagheer A., Kotb M. Unsupervised pre-training of a deep LSTM-based stacked autoencoder for multivariate time series forecasting problems //Scientific reports.

- 2019. - T. 9. - №. 1. - C. 19038.

105. Schlumberger curve mnemonic dictionary, 2023. URL: https://www.apps.slb.com/cmd/

106. Serra J., Arcos J. L. An empirical evaluation of similarity measures for time series classification //Knowledge-Based Systems. - 2014. - T. 67. - C. 305-314.

107. Shchepetov O. A. System classification of failures in drilling //Vestnik of Astrakhan State Technical University. Series: Management, Computer Sciences and Informatics. - 2009. - №. 2. - C. 36-42.

108. Shi P. et al. Digital signal receiver for measurement while drilling system having noise cancellation: US Patent. 6741185. - 2004. URL: https://patents.google.com/patent/US6741185B2/en

109. Shor R. J., Pryor M., Van Oort E. Drill string vibration observation, modelling and prevention in the oil and gas industry //Dynamic Systems and Control Conference.

- American Society of Mechanical Engineers, 2014. - T. 46209. - C. V003T37A004.

110. Sifuzzaman M., Islam M. R., Ali M. Z. Application of wavelet transform and its advantages compared to Fourier transform. - 2009 - C. 121-134 .

111. Skrlj B. et al. Feature importance estimation with self-attention networks //arXiv preprint arXiv:2002.04464. - 2020. DOI: 10.48550/arXiv.2002.04464 .

112. Sprljan P. et al. Automation systems design and laboratory prototyping aimed at mature petroleum drilling rig retrofitting //Tehnicki vjesnik. - 2020. - T. 27. - №. 1. - C. 229-236.

113. Strumbelj E., Kononenko I. Explaining prediction models and individual predictions with feature contributions //Knowledge and information systems. -2014. - T. 41. - C. 647-665.

114. Tzanetakis G., Essl G., Cook P. Audio analysis using the discrete wavelet transform //Proc. conf. in acoustics and music theory applications. - Citeseer, 2001. - T. 66. URL: https://soundlab.cs.princeton.edu/publications/2001_amta_aadwt.pdf .

115. Vadetskii U. Drilling oil and gas wells - 2008 Nedra. URL: http://lib.dvfu.ru:8080/lib/item?id=chamo:694456&theme=FEF.

116. Van der Maaten L., Hinton G. Visualizing data using t-SNE //Journal of machine learning research. - 2008. - T. 9. - №. 11. URL: https://www.jmlr.org/papers/volume9/vandermaaten08a/vandermaaten08a.pdf

117. Vaswani A. et al. Attention is all you need //Advances in neural information processing systems. - 2017. - T. 30. URL: https://papers.nips.cc/paper_files/paper/2017/hash/3f5ee243547dee91fbd053c1c4 a845aa-Abstract.html .

118. Vellido A., Martín-Guerrero J. D., Lisboa P. J. G. Making machine learning models interpretable //ESANN. - 2012. - T. 12. - C. 163-172..

119. Visani G. et al. Statistical stability indices for LIME: Obtaining reliable explanations for machine learning models //Journal of the Operational Research Society. - 2022. - T. 73. - №. 1. - C. 91-101.

120. Vos B. E., Reiber F. The benefits of monitoring torque & drag in real time //IADC/SPE Asia Pacific Drilling Technology Conference and Exhibition?. - SPE, 2000. - C. SPE-62784-MS.

121. Wang J. et al. Bag-of-words representation for biomedical time series classification //Biomedical Signal Processing and Control. - 2013. - T. 8. - №. 6. - C. 634-644.

122. Wang X. et al. RPM: Representative Pattern Mining for Efficient Time Series Classification //EDBT. - 2016. - C. 185-196.

123. Wang Z., Yan W., Oates T. Time series classification from scratch with deep neural networks: A strong baseline //2017 International joint conference on neural networks (IJCNN). - IEEE, 2017. - C. 1578-1585.

124. Wold S., Esbensen K., Geladi P. Principal component analysis //Chemometrics and intelligent laboratory systems. - 1987. - T. 2. - №. 1-3. - C. 37-52.

125. Wu A., Hareland G., Fazaelizadeh M. Torque & drag analysis using finite element method //Modern Applied Science. - 2011. - T. 5. - №. 6. - C. 13.

126. Xiao J. et al. Attentional factorization machines: Learning the weight of feature interactions via attention networks //arXiv preprint arXiv:1708.04617. - 2017. DOI: 10.48550/arXiv.1708.04617 .

127. Yan J. Y., Yang J. G. Application of synthetic grey correlation theory on thermal point optimization for machine tool thermal error compensation //The International Journal of Advanced Manufacturing Technology. - 2009. - T. 43. - C. 1124-1132.

128. Yoon J., Jarrett D., Van der Schaar M. Time-series generative adversarial networks //Advances in neural information processing systems. - 2019. - T. 32. DOI: 10.48550/arXiv.2202.02691 .

129. Zalivin U, Vachromeevm A. Emergence situations while drilling. - 2016, Nedra. URL: https://www.iprbookshop.ru/78263.html.

130. Zhang Q., Zhu S. C. Visual interpretability for deep learning: a survey //Frontiers of information technology & electronic engineering. - 2018. - T. 19. - №. 1. - C. 27-39.

131. Zhang Z. et al. Opening the black box of neural networks: methods for interpreting neural network models in clinical applications //Annals of translational medicine. - 2018. - T. 6. - №. 11. DOI:10.21037/atm.2018.05.32.

132. Zheng Y. et al. Time series classification using multi-channels deep convolutional neural networks //International conference on web-age information management. - Cham : Springer International Publishing, 2014. - C. 298-310.

133. Zimek A., Schubert E. Outlier detection, encyclopaedia of database systems //Springer New York-2017. - 2017. - T. 10. - C. 978-1.