Al Amin, Verdana Marchian Ilham Putra (2017) Prediksi Kedalaman Lapisan Batuan Gamping Di Area Selat Sunda Utara Menggunakan Multilayer Perceptron Duelist Algortihm. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2413100031_Undergraduate_Theses.pdf - Published Version Download (3MB) | Preview |
Abstract
Eksplorasi minyak bumi selalu dioperasikan pada lapisan batuan pasir yang semakin hari kian menipis. Batuan gamping dapat mengandung minyak bumi hingga 50%. Sumur Indri A-1 Maxusu sebagai sumur latih dan sumur Widuri A-1 Maxusu sebagai sumur uji di Area Selat Sunda Utara. Data logging pada sumur latih diolah menggunakan beberapa metode. Metode uji multikoliniearitas digunakan untuk mengetahui variabel yang mengalami multikoliniearitas dan diatasi dengan metode PCA-PLS. Didapatkan 8 variabel yang tidak multikoliniearitas, yaitu ILD, ILM, SFLU, SP, CALI, RHOB, DEPT, dan GR. Pada sumur uji, terdapat 2 variabel yang tidak ada, yaitu SP dan SFLU. Sehingga, dilakukan normalisasi data dengan menggunakan Jaringan Syaraf Tiruan arsitektur Multilayer Perceptron Levenberg Marquadt. Data yang sudah lengkap ini dilatih untuk mendapatkan hasil kedalaman batuan gamping yang divalidasi dengan menggunakan Log image. Dari hasil prediksi, didapatkan koefisien korelasi sebesar 0,327 terhadap kedalaman lapisan batuan gamping dengan titik terdalam 1276 – 1288 ft. Kemudian, dibandingkan dengan menggunakan 2 hidden node antara metode update bobot LM dengan Duelist Algorithm. DA dapat menghasilkan bobot hingga RMSE sebesar 0,11, sedangkan dengan LM sebesar 0,257464.
===========================================================================================
Petroleum exploration has always operated on a layer of sandstone, but the more days the thinning. Limestone may contain petroleum by 50%. Indri A-1 Maxusu well as well trained and Widuri A-1 Maxusu well as test well in the North Sunda Strait Area. Data logging on well trained processed using several methods. Multikolinierity test method used to determine which variables are experiencing multikolinierity and solved with PCA-PLS. Obtained 8 multikolinierity variables, that is ILD, ILM, SFLU, SP, CALI, RHOB, DEPT, and GR. In the test well, there are 2 variables that do not exist, namely SP and SFLU. Thus, normalization of data by using artificial neural network Multilayer Perceptron architecture Levenberg Marquadt. The complete data are trained to get the depth of limestone were validated by using the Log image. From the predicted results, obtained correlation coefficient of 0.327 against the limestone rock layer depth with the deepest point of 1276 to 1288 ft. Then, as compared with using a hidden node 2 between weights update method with Duelist LM Algorithm. DA can produce a weight of up RMSE of 0.11, while the LM amounted to 0.257464.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Batuan gamping, Jaringan Syaraf Tiruan, Prediksi kedalaman, Duelist Algorithm |
| Subjects: | Q Science > Q Science (General) Q Science > QE Geology Q Science > QE Geology > QE571 Sedimentation and deposition. Sediment transport. Erosion. |
| Divisions: | Faculty of Industrial Technology > Physics Engineering > 30101-(S2) Master Thesis |
| Depositing User: | VERDANA PUTRA AL AMIN |
| Date Deposited: | 17 Apr 2017 08:27 |
| Last Modified: | 17 Apr 2017 08:51 |
| URI: | http://repository.its.ac.id/id/eprint/3890 |
Actions (login required)
 |
View Item |