Forecasting of PV Power Generation using weather input data-preprocessing techniques

Stochastic nature of weather conditions influences the photovoltaic power forecasts. The present work investigates the accuracy performance of data-driven methods for PV power ahead prediction when different data preprocessing techniques are applied to input datasets. The Wavelet Decomposition and the Principal Component Analysis were proposed to decompose meteorological data used as inputs for the forecasts. A time series forecasting method as the GLSSVM (Group Least Square Support Vector Machine) that combines the Least Square Support Vector Machines (LS-SVM) and Group Method of Data Handling (GMDH) was applied to the measured weather data and implemented for day-ahead PV generation forecast.