부산대학교 도서관

This paper presents a hybrid solar irradiance forecasting to assist management of an isolated microgrid powered by photovoltaic systems (PV) and equipped with an energy storage system (ESS), lithium-ion batteries, and electrical vehicle (EV). Accurate production forecasting is needed to reduce the uncertainty of photovoltaic array caused by solar irradiance intermittency and control optimally the microgrid: schedule production, storage and load to maximize renewable resources production, ESS lifetime and minimize the utilization costs. The energy management system (EMS) must take into account the uncertainty of the forecasting models (electricity production and demand) to elaborate control strategies. The short-term solar forecasting (from 30 minutes up to 6 hours) is based on satellite data (visible (VIS) and infrared (IR) channels) coupled with an auto-regressive model. First of all, we describe the experimental microgrid “Rivesaltes-Grid”. Then we present the clear sky model used and its assessment. Moreover, the forecasting method is detailed with the results for seven locations. Finally, we conclude and discuss the performance of this forecasting model and the future perspectives. The short-term prediction algorithm will be integrated into La Compagnie du Vent's computer system. It will be used for microgrid management but also in the future perspectives for the sale of electricity from photovoltaic power plants on the intraday market.