Implementasi Machine Learning pada Data Well Log

Wicaksono, Satrio Hanif (2023) Implementasi Machine Learning pada Data Well Log. Project Report. [s.n], [s.l.]. (Unpublished)

Text 05111940000103-Project_Report.pdf - Accepted Version Download (2MB)

Abstract

Kualitas data well logging dipengaruhi oleh beberapa masalah seperti data yang tidak tersedia, nilai yang salah, masalah penyelarasan, resolusi, dan kesalahan manusia. Ahli petrofisika menghabiskan waktu banyak untuk Quality control. Algoritma machine learning dapat digunakan untuk memprediksi data yang tidak tersedia, mendeteksi outlier, dan klasifikasi fasies reservoir. Machine learning juga dapat membantu memprediksi data log sonik jika terdapat hubungan kuantitatif antara kurva well log yang tersedia. Dalam Kerja Praktek ini, penulis menggunakan dataset dari Norwegian Petroleum Directorate untuk identifikasi Badhole dan prediksi Shear Sonic Log menggunakan algoritma DBSCAN, OPTICS, Boosted Random Forest, dan LSTM. Hasil analisis conventional well log menunjukkan bahwa algoritma DBSCAN lebih efektif dalam mengidentifikasi Badhole dibandingkan dengan algoritma OPTICS. Selain itu, algoritma Boosted Random Forest memberikan prediksi Shear Sonic Wave Log yang lebih akurat dibandingkan dengan algoritma LSTM.
===========================================================================================================================
The quality of well logging data is influenced by several issues such as unavailable data, incorrect values, alignment problems, resolution, and human errors. Petrophysicists spend a lot of time on quality control. Machine learning algorithms can be used to predict unavailable data, detect outliers, and classify reservoir facies. Machine learning can also help predict sonic log data if there is a quantitative relationship between the available well log curves. In this Internship, the author used a dataset from the Norwegian Petroleum Directorate for Badhole identification and Shear Sonic Log prediction using the DBSCAN, OPTICS, Boosted Random Forest, and LSTM algorithms. The results of conventional well log analysis showed that the DBSCAN algorithm is more effective in identifying Badhole compared to the OPTICS algorithm. Additionally, the Boosted Random Forest algorithm provides more accurate predictions of the Shear Sonic Wave Log compared to the LSTM algorithm.

Item Type:	Monograph (Project Report)
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Satrio Hanif Wicaksono
Date Deposited:	20 Jun 2023 08:15
Last Modified:	20 Jun 2023 08:15
URI:	http://repository.its.ac.id/id/eprint/98144

Actions (login required)

View Item