Deep Learning Application in Power System with a Case Study on Solar Irradiation Forecasting

Abstract

Power systems are developing day by day due to the inclusion of latest digital technologies. Due to the increasing complexities in power systems and collection of high volume of data, Deep Learning (DL) techniques are becoming most suitable technologies for its future development and success. Due to high performance computing with decreased computational cost, availability of huge amount of data, and better algorithms, DL has entered into its new developmental stage. This article introduces state of the art of application of Deep Learning in power systems, and presents a novel case study on the solar irradiance forecasting required for PV generation. The case study is prediction of hourly, daily and total solar irradiation forecasting for a year ahead using Long-Short Term Memory (LSTM). Year ahead data is important from the point of view of installation planning and market.