Prosumers' renewable small-size generation forecasting analyses with ARIMA models

Abstract

Renewable small-size generation is becoming more and more attractive for smart-cities. It replaces to some extend conventional power generation based on coal, oil, gas, etc. and can range between 30 W to 400 kW. This leads to less polluted cities and lower investment in onerous transmission and distribution grids since generation sources is closer to final electricity and thermal energy consumers. In this paper, we used input data from small-size generation renewable sources based on solar irradiance and wind speed. The input data have been undergone through the Extract-Transform-Load (ETL) process. After the ETL, we performed first and second order of autoregressive, moving average, combined autoregressive and moving average, and autoregressive integrated moving average that is a generalization of combined autoregressive and moving average. We considered two study cases for performing forecasting analyses: the wind power generators located in Tulcea city and photovoltaic panel that is located in Giurgiu city. These forecasts could be mainly used by prosumers to perform load profiles, improve their consumption optimization, electricity market related activities, etc. We applied several different forecasting methods (SVM, ANN), but the best results we obtained with ARIMA models.