Drilling Rate Optimization by Automatic Lithology Prediction Using Hybrid Machine Learning
Publish Year: 1398
نوع سند: مقاله ژورنالی
زبان: English
View: 72
This Paper With 12 Page And PDF Format Ready To Download
- Certificate
- من نویسنده این مقاله هستم
استخراج به نرم افزارهای پژوهشی:
شناسه ملی سند علمی:
JR_JPSTR-9-4_006
تاریخ نمایه سازی: 29 آذر 1402
Abstract:
It is essential to obtain valuable information during drilling from the formation that is being drilled for rate optimization. In the drilling operation, the process of lithology and formation determination is extremely obscurant and it seems machine learning, as a novel prediction method that can model complicated situations having a high degree of uncertainty, could be beneficial. In this work, the real-time drilling data was applied to predict the formation type and lithology while drilling that formation using a genetic algorithm and Taguchi design of experiment optimized artificial neural network. Drilling data of twelve wells in one of Iranian gas fields were applied for this work. ۴۷۵۰۰ sets of data were selected, and after data control, ۳۱۲۰۰ data sets were selected as valid data and imported to artificial neural networks. For performing this research, by changing the network features and optimizing the structure of the network using the Taguchi method and optimizing the weight and biases of the network using the genetic algorithm, a unique artificial neural network was designed. The results show that the developed hybrid machine learning method can predict formation and lithology with a high degree of accuracy.
Keywords:
Authors
Alireza Moazzeni
Department of petroleum engineering, Amirkabir University, Tehran, Iran
Ehsan Khamehchi
Department of Petroleum Engineering, Amirkabir University of Technology
مراجع و منابع این Paper:
لیست زیر مراجع و منابع استفاده شده در این Paper را نمایش می دهد. این مراجع به صورت کاملا ماشینی و بر اساس هوش مصنوعی استخراج شده اند و لذا ممکن است دارای اشکالاتی باشند که به مرور زمان دقت استخراج این محتوا افزایش می یابد. مراجعی که مقالات مربوط به آنها در سیویلیکا نمایه شده و پیدا شده اند، به خود Paper لینک شده اند :