Flare Identification by Forecasting Sunspot Numbers Using Fuzzy Time Series Markov Chain Model

Autor: Dian Candra Rini Novitasari, Nanang Widodo, Nurul Ardhiyah
Rok vydání: 2019
Předmět:
Zdroj: 2019 International Seminar on Intelligent Technology and Its Applications (ISITIA).
DOI: 10.1109/isitia.2019.8937242
Popis: The energy that move the global atmosphere, especially is coming from the sun. Sun has three main activities that can affect the state of the Earth, such as sunspots, the explosion of the sun (flare), and the coronal mass ejection (CME). One of the easiest solar activity observed on the Earth are sunspot and flare. The period of the solar activity cycle takes an average of 11 years. International sunspot number (R) is a key that can indicating the existence of solar activity. In this study, used the method of fuzzy time series Markov Chain models to predict the sunspot number. Fuzzy time series Markov chain model is a combination of fuzzy time series method with Markov chain which aims to have the greatest probability obtained by using Markov transition probability matrix. The forecasting of sunspot numbers using fuzzy time series Markov Chain model has an accuracy rate of 6,54% by using MAPE and 9,5% on test data. Then the results of sunspot number forecast in April 2019 was 6,4 and in May 2019 was 6,5 which indicated not potentially to flare because the sunspot numbers generated are less than 20.
Databáze: OpenAIRE