Long-term solar generation forecasting

摘要

The rapid growth of solar Photovoltaic (PV) technology has been very visible over the past decade. Such increase in the integration of solar generation has brought attention to the forecasting issues. This paper presents a new approach to tackle the long-term forecasting challenge and accordingly reduce the uncertainty of the PV forecast, which would accordingly help facilitate its integration into the electric power grid. The new method includes a set of pre- and post-processes that will be undertaken before the data is fed to the forecasting model and after the forecast is obtained. Using the proposed method, the historical solar PV radiation data, which is non-stationary, is converted to a set of stationary data which will accordingly allow utilization of a larger set of data for forecasting. Numerical simulations exhibit the performance of the proposed method.