Exploratory Analysis of Electric Demand Using Informational Tools

An original analysis proposal is presented to detect relevant characteristics and particular events of the electric power demand signal in the Argentine Interconnected System (in Spanish, Sistema Interconectado Argentino, SADI). In this sense, permutation entropy has been used to analyze the degree of uncertainty of a signal. As a result, it is possible to find certain patterns in the SADI hourly electricity demand records, which contribute to make the data exploration task faster, more efficient and easier to implement. The simple observation and interpretation of this metric allows an improvement of the prediction of future values by 7%. Together with the inclusion of temperature data, by means of a polynomial model of order two, it represents an overall improvement of 19% over a simple linear regression model on the original signal data. This result has great economic implications, since it allows a better management of generation, and environmental implications since it would optimize the dispatches of electric power demand.