Artificial Neural Networks for Nonlinear Time Series: A Case Study with the Difference Settlement Price

• Grant number: 25/14744-9
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: October 01, 2025
• End date: September 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Mathematics - Applied Mathematics
• Principal Investigator: Thays Aparecida de Abreu Santos
• Grantee: Ana Júlia Berchol Matias
• Host Institution: Faculdade de Engenharia (FEIS). Universidade Estadual Paulista (UNESP). Campus de Ilha Solteira. Ilha Solteira , SP, Brazil

Abstract

The Spot Price for Settlement of Differences (PLD) is one of the main indicators in the Brazilian electricity market, used for accounting transactions in the short-term market. Due to its high volatility and nonlinear behavior, PLD poses a challenge for modeling and forecasting. This Scientific Initiation project aims to analyze and forecast the PLD time series using a machine learning approach. Initially, an exploratory analysis of the historical series will be conducted, focusing on identifying components such as trend, seasonality, structural breaks, and outliers. The forecasting stage will employ artificial neural networks, using two distinct models: the Multilayer Perceptron (MLP), trained through error backpropagation, and the Fuzzy ARTMAP network, known for its continuous and adaptive learning capabilities. The results are expected to contribute to a better understanding of the series' behavior and to provide predictive models with potential applications in the electricity sector.