Hydro-power is a widespread source of energy, which currently provides over 60% of total renewable electricity production. As such, it plays a key role in green power generation, and has a fundamental influence on power market prices, because it can be used as a buffer for more volatile renewable sources, and it is relatively cheap to ramp up and down. For these reasons, it is of paramount importance to accurately predict the monthly hydro-power production capacity of wide geographical zones of the electricity market. In fact, future hydro-power production capacity depends on meteorological and climatic processes, water storage as result of pumping activity in the plant, and, of course, actual production, and this makes it extremely difficult to obtain an accurate prediction using traditional techniques, such as auto-regressive models. In this paper we propose a methodology based on machine learning (ML) regression techniques, mainly artificial neural networks and support vector machines, and feature reduction mechanisms, such as principal component analysis and feature grouping techniques. We apply these techniques to model the relationship between the meteorological and climatic variables and the total water in the reservoir used for the hydro-power generation. We show how ML regression techniques are able to obtain an accurate prediction of the hydro-power capacity in a real life example in Northern Italy.
Condemi, C., Casillas-Perez, D., Mastroeni, L.C.L., Jiménez-Fernández, S., & Salcedo-Sanz, S. (2021). Hydro-power production capacity prediction based on machine learning regression techniques. KNOWLEDGE-BASED SYSTEMS, 222 (2021) 107012 [https://doi.org/10.1016/j.knosys.2021.107012].
|Titolo:||Hydro-power production capacity prediction based on machine learning regression techniques|
|Data di pubblicazione:||2021|
|Citazione:||Condemi, C., Casillas-Perez, D., Mastroeni, L.C.L., Jiménez-Fernández, S., & Salcedo-Sanz, S. (2021). Hydro-power production capacity prediction based on machine learning regression techniques. KNOWLEDGE-BASED SYSTEMS, 222 (2021) 107012 [https://doi.org/10.1016/j.knosys.2021.107012].|
|Appare nelle tipologie:||1.1 Articolo in rivista|