Forecasting Solar Activity with Two Computational Intelligence Models (A Comparative Study)

TL;DR

This paper compares the effectiveness of BELFIS, a neural-inspired fuzzy inference system, with ANFIS in forecasting solar activity across multiple solar cycles, aiming to improve prediction accuracy for space weather events.

Contribution

It introduces a performance evaluation of BELFIS for solar cycle prediction and compares it with the established ANFIS model, highlighting its potential advantages.

Findings

BELFIS shows competitive forecasting accuracy.

BELFIS outperforms ANFIS in certain solar cycle predictions.

The study demonstrates BELFIS's suitability for chaotic system forecasting.

Abstract

Solar activity It is vital to accurately predict solar activity, in order to decrease the plausible damage of electronic equipment in the event of a large high-intensity solar eruption. Recently, we have proposed BELFIS (Brain Emotional Learning-based Fuzzy Inference System) as a tool for the forecasting of chaotic systems. The structure of BELFIS is designed based on the neural structure of fear conditioning. The function of BELFIS is implemented by assigning adaptive networks to the components of the BELFIS structure. This paper especially focuses on performance evaluation of BELFIS as a predictor by forecasting solar cycles 16 to 24. The performance of BELFIS is compared with other computational models used for this purpose, and in particular with adaptive neuro-fuzzy inference system (ANFIS).