Performa Algotima Dual Classification Learning Rao (DL-Rao) dalam Inversi Metode Self-Potensial

Setiawan, Boby (2025) Performa Algotima Dual Classification Learning Rao (DL-Rao) dalam Inversi Metode Self-Potensial. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5001201150-Undergraduate_Thesis.pdf Restricted to Repository staff only until 1 April 2027. Download (4MB) | Request a copy

Abstract

Metode self-potensial (SP) adalah sebuah metode dalam eksplorasi geofisika yang sering digunakan untuk mengidentifikasi struktur pada bawah permukaan bumi. Dalam melakukukan pengukuran metode SP, pemodelan kebelakang (inversi) dijalankan untuk menentukan nilai-nilai parameter model. Proses inversi melibatkan perhitungan matematis dengan algortima tertentu dengan tujuan meminimalisasi fungsi objektif yang merepresentasikan kecocokan antara data observasi dengan data perhitungan. Banyak algoritma yang dapat digunakan untuk melakukan inversi, salah satunya ialah algoritma Dual Classification Learning Rao (DL-Rao). Algoritma DL-Rao adalah pengembangan dari algoritma Rao yang baru diterapkan pada data gravasi dan belum ada penelitian tentang performa algortima ini dalam menginversi data SP. Pada penelitian ini algoritma DL-Rao dilakukan pengujian untuk menyelesaikan permasalahan inversi data SP, kemudian algoritma ini dibandingkan dengan salah satu algoritma Rao yang cukup baik yaitu Hybrid Rao and generalized normal distribution optimization (HGNDO-Rao). Hasil pengujian menunjukan bahwa algoritma DL-Rao memiliki performa yang baik dalam menginversi data SP. DL-Rao memiliki konvergensi yang cepat, yaitu konvergen pada iterasi ke-20. Stabilitas fungsi objektif yang baik, dengan bukti interkuartil yang sangat kecil bahkan mencapai nol. Ketahanan (robust) terhadap noise, yang ditunjukkan oleh boxplot fungsi objektifnya yang tetap stabil saat diberi noise. Serta akurasi estimasi model yang tinggi, di mana semua nilai misfit (RMSE) hasil inversinya cenderung kecil, bahkan RMSE DL-Rao dapat mecapai sebesar 2,57×10^−7. Bahkan performa DL-Rao lebih baik dari algortima HGNDO-Rao yang baru dapat konvergen pada iterasi ke-80, interkuartil tidak pernah berada di bawah 10^−5, boxplot fungsi objektif yang mengalami peningkatan rentang, dan RMSE yang hanya sampai bernilai 1,87×10^−3.
=====================================================================================================================================
The self-potential (SP) method is a geophysical exploration method that is often used to identify subsurface structures. In conducting SP measurements, inverse modeling is performed to determine the values of model parameters. The inversion process involves mathematical calculations using certain algorithms to minimize an objective function that represents the fit between observational data and calculated data. Various algorithms can be employed for inversion, one of which is the Dual Classification Learning Rao (DL-Rao) algorithm. The DL-Rao algorithm is development of the Rao algorithm that has only been applied to gravity data and there has been no research on the performance of this algorithm in inverting SP data. In this study, the DL-Rao algorithm was tested to solve the SP data inversion problem, and then this algorithm was compared with one of the well-performing Rao algorithms, namely Hybrid Rao and Generalized Normal Distribution Optimization (HGNDO-Rao). The results indicate that DL-Rao demonstrates excellent performance in SP data inversion, characterized by fast convergence, reaching convergence at the 20th iteration. It exhibits high stability in the objective function, as evidenced by a very small interquartile range, even reaching zero. Additionally, it shows robustness against noise, as demonstrated by the stability of the objective function's boxplot under noisy conditions. Furthermore, DL-Rao achieves high model estimation accuracy, with all misfit (RMSE) values remaining low, reaching as small as 2,57×10^−7. In contrast, HGNDO-Rao exhibits lower performance, converging only at the 80th iteration, with its interquartile range never falling below 10^−5, an increasing range in the objective function's boxplot, and an RMSE value reaching only 1,87×10^−3.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	DL-Rao, Inversi, Self-Potensial.
Subjects:	Q Science > QA Mathematics > QA9.58 Algorithms
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Physics > 45201-(S1) Undergraduate Thesis
Depositing User:	Boby Setiawan
Date Deposited:	10 Feb 2025 07:22
Last Modified:	10 Feb 2025 07:22
URI:	http://repository.its.ac.id/id/eprint/118593

Actions (login required)

View Item