PREDIKSI SOLAR FLARES MENGGUNAKAN PRODUK DATA VECTOR MAGNETIC SDO/HMI DAN RANDOM FERNS

Rooseno Rahman Dewanto, - (2020) PREDIKSI SOLAR FLARES MENGGUNAKAN PRODUK DATA VECTOR MAGNETIC SDO/HMI DAN RANDOM FERNS. S1 thesis, Universitas Pendidikan Indonesia.

[img] Text
S_KOM_1304654_Title.pdf

Download (698kB)
[img] Text
S_KOM_1304654_Chapter1.pdf

Download (273kB)
[img] Text
S_KOM_1304654_Chapter2.pdf
Restricted to Staf Perpustakaan

Download (1MB)
[img] Text
S_KOM_1304654_Chapter3.pdf

Download (381kB)
[img] Text
S_KOM_1304654_Chapter4.pdf
Restricted to Staf Perpustakaan

Download (1MB)
[img] Text
S_KOM_1304654_Chapter5.pdf

Download (207kB)
[img] Text
S_KOM_1304654_Appendix.pdf
Restricted to Staf Perpustakaan

Download (16MB)
Official URL: http://repository.upi.edu

Abstract

Solar Flares (SFs) merupakan letusan energi tiba-tiba yang disebabkan oleh kekusutan, persilangan, atau penataan ulang garis medan magnet di dekat bintik matahari. Fenomena ini juga diketahui sebagai letusan paling kuat di tata surya yang sering kali memberi pengaruh buruk bagi cuaca ruang angkasa. Oleh karena itu, banyak peneliti dengan beragam pendekatan berusaha memprediksi akan kemunculannya. Salah satu yang berperan penting pada upaya prediksi ini adalah Instrumen Helioseismic and Magnetic Imager (HMI) pada Solar Dynamic Observatory (SDO) yang secara terus menerus mengamati full-disk photospheric vector magnetic field dari luar angkasa di saat kebanyakan penelitian berbasis pada observasi dari dalam bumi. Berbekal data flux SFs yang direkam oleh Instrumen X-ray Sensors (XRS) pada Geostationary Operational Environmental Satellite (GOES) dan dipetakan dengan data vector magnetic berdasarkan Active Region (AR) Numbers, pada rentang 01 Mei 2010 sampai 10 Mei 2020, penelitian ini mengajukan model prediksi multiclass dan binary SFs menggunakan Algoritma Random Ferns (RFe) dan Teknik Oversampling. Implementasi Naive Bayesian Classification pada RFe sendiri diketahui optimal menangani banyak features yang merupakan kunci untuk meningkatkan tingkat klasifikasi. Sementara Teknik Oversampling digunakan untuk menyeimbangkan kelas minor pada populasi. Hasil dari penelitian ini menunjukkan bahwa model prediksi SFs menggunakan RFe dapat mengungguli beberapa aspek penelitian terdahulu. Adapun nilai rata-rata tertinggi sensitivity/recall, precision, dan TSS multiclass SFs yang diraih penelitian ini secara berturut-turut adalah 74,4%, 50,3%, dan 58,7%. Sementara nilai rata-rata tertinggi sensitivity/recall, precision, dan TSS binary SFs yang diraih penelitian ini secara berturut-turut adalah 87,7%, 77,7%, dan 72,8%. Solar Flares (SFs) are sudden bursts of energy caused by tangling, crossing, or reorganizing of magnetic field lines near sunspots. This phenomenon is also known to be the most powerful eruption in the solar system which often yield adverse impact on space weather. Therefore, many researchers with various approaches try to predict its occurrence. One that plays an important role in this prediction effort is the Helioseismic and Magnetic Imager (HMI) Instrument on the Solar Dynamic Observatory (SDO) which continuously observes the full-disk photospheric vector magnetic field from space while most research is ground-based observations. By using SFs flux data recorded by the X-ray Sensors (XRS) Instrument on the Geostationary Operational Environmental Satellite (GOES) and mapped with vector magnetic data based on the Active Region (AR) Numbers, in the range 01 May 2010 to 10 May 2020, this study proposes a multiclass and binary SFs prediction model using Random Ferns (RFe) Algorithm and Oversampling Technique. The implementation of the Naive Bayesian Classification on RFe itself is known to be optimal in handling many features which are the key to increase the classification rate. While the Oversampling Technique is used to balance the minor classes in the population. The results of this study indicate that the SFs prediction model using RFe can outperform several aspects of previous research. The highest average scores for sensitivity/recall, precision, and TSS multiclass SFs achieved in this study were 74.4%, 50.3%, and 58.7%, respectively. Meanwhile, the highest average values for sensitivity/recall, precision, and TSS binary SFs achieved in this study were 87.7%, 77.7%, and 72.8%, respectively.

Item Type: Thesis (S1)
Additional Information: No Panggil : S KOM ROO p-2020; NIM :1304654
Uncontrolled Keywords: Solar Flares, Prediksi, Geostationary Operational Environmental Satellite, X-ray Sensors, Solar Dynamic Observatory, Helioseismic and Magnetic Imager, Active Region Numbers, Vector Magnetic, Oversampling, Random Ferns
Subjects: L Education > L Education (General)
Q Science > QA Mathematics > QA76 Computer software
Divisions: Fakultas Pendidikan Matematika dan Ilmu Pengetahuan Alam > Program Studi Ilmu Komputer
Depositing User: Rooseno Rahman Dewanto
Date Deposited: 01 Sep 2020 07:50
Last Modified: 01 Sep 2020 07:50
URI: http://repository.upi.edu/id/eprint/51602

Actions (login required)

View Item View Item