Data-Driven Approach to Predict Spot Market Price in Indian Electricity

The restructuring of vertically integrated utilities has made the energy market more competitive. The power trading in a deregulated power system is reliant on an auction-based contract between market participants including generators, retailers, and wholesale traders. The electricity market is subj...

Full description

Saved in:
Bibliographic Details
Published in2023 International Conference on Electrical, Electronics, Communication and Computers (ELEXCOM) pp. 1 - 5
Main Authors M. P., Subeekrishna, R. R., Lekshmi
Format Conference Proceeding
LanguageEnglish
Published IEEE 26.08.2023
Subjects
Adaptation models
Backpropagation algorithm
Data models
deregulated power system
electricity market
electricity price forecasting
Electricity supply industry
Machine learning
Machine learning algorithms
market clearing price
neural network
Prediction algorithms
Predictive models
Online AccessGet full text

Cover

More Information
Summary:The restructuring of vertically integrated utilities has made the energy market more competitive. The power trading in a deregulated power system is reliant on an auction-based contract between market participants including generators, retailers, and wholesale traders. The electricity market is subject to several constraints imposed by the inherent nature of energy, which necessitates perpetual parity between consumption and production on a continental scale. To address this issue, markets employ pricing algorithms. This paper aims to develop a forecasting model for short-term electricity prices employing a data-driven approach. The model develops a predictive mechanism based on statistical and neural network frameworks. A multi-layer neural network architecture incorporating the backpropagation algorithm is employed. The prediction model is developed using the historical energy price data of a 15-minute day ahead Indian electricity market, available in the IEX webpage. The network is trained using the data from the 1 st of March 2023 to the 30 th of April 2023. The accuracy of the model is analyzed through testing with the data from 1 st of May 2023 to 31 st of May 2023. Finally, the exactness of the model is verified using mean absolute percentage error and R2 score. This model provides statistically significant performance for electricity price forecasting.
DOI:10.1109/ELEXCOM58812.2023.10370466