Banca de QUALIFICAÇÃO: ELIAS MARTINS VIEIRA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : ELIAS MARTINS VIEIRA
DATE: 29/08/2025
TIME: 13:00
LOCAL: https://conferenciaweb.rnp.br/sala/felipe-leite-coelho-da-silva
TITLE:

ANALYSIS AND FORECASTING OF ELECTRICITY CONSUMPTION IN THE STATE OF RIO DE JANEIRO: A BOTTOM-UP APPROACH AND NEURAL NETWORKS

KEY WORDS:

Electricity consumption; Bottom-up; Neural networks; Univariate time series; Forecasting.

PAGES: 69
BIG AREA: Ciências Sociais Aplicadas
AREA: Economia
SUMMARY:

Electricity is an essential input for socioeconomic development, and its planning is indispensable for sustainability, especially in Rio de Janeiro, where the residential, commercial, industrial, and other sectors (public activities/services) are major consumers. Traditionally, consumption forecasting models in the state are based on univariate analyses, which consider only historical consumption data and, therefore, limit the ability to capture the influence of external factors. Although the literature presents various approaches, few studies apply the bottom-up methodology focusing specifically on the complex socioeconomic context of Rio de Janeiro, which represents a gap that needs to be filled. This study therefore seeks to answer which variables, in addition to historical consumption, can improve the accuracy of energy forecasts in the state. The overall objective is to implement and compare time series models and neural networks for the residential, commercial, industrial, and other sectors, incorporating relevant explanatory variables. To this end, the research adopts a hybrid methodology that combines a bottom-up approach, aggregating sectoral forecasts to obtain total demand, with the flexibility of Artificial Neural Networks (NNAR and MLP) to model the non-linear relationships of the data. The proposed MLP model will integrate exogenous variables such as the number of consumers, the energy tariff deflated by the IPCA, and the state's average temperature. The models' performance will be evaluated and compared using a set of complementary statistical metrics, including the Mean Absolute Percentage Error (MAPE), the Root Mean Square Error (RMSE), the Mean Absolute Error (MAE), and the Coefficient of Determination (R²). The results are expected to generate valuable information for managers and policymakers, contributing to the improvement of forecasting methodologies and encouraging more efficient and sustainable energy planning in the state of Rio de Janeiro.

COMMITTEE MEMBERS:
Interna - 2829205 - DEBORA MESQUITA PIMENTEL
Presidente - 2639882 - FELIPE LEITE COELHO DA SILVA
Interno - 1447519 - PAULO JOSE SARAIVA