Purpose: Boosting the energy marketing strategy.
Objectives
Adoption of current Machine Learning techniques for application in the forecasting of flows and load used by the DESSEM model (commonly implemented by Eletrobras).
Key Results
- Technical: greater security in energy trading operations
- Strategic: Improving the process of forecasting the hourly price of energy, given that the affluent flow and energy demand data are two variables of great importance in the formation of the PLD (Settlement Price of Differences, in English)
- Economic: less exposure to the market, which could save ~R$ 300k/month
Main information
Scope
Development of a semi-hourly demand and flow forecasting system
Partners / Entities of support
UFPE/FADE
Investment (contract)
R$ 2,4MM
Scalable?
Yes
Stages
Design
October/2021
Kickoff
April/2022
Prototype
April/2024
Tests
July/2024
Validation
September/2024
Implementation
October/2024
Deadline for implementation (in months)
30 months
P.S.:
Deadline: October/2024
