Solar Photovoltaic output prediction using Jackknife Regression

摘要

A statistical model for predicting the output power and energy of Solar Photovoltaics (PV) has been developed. The multiple input single output (MISO) system is based on Jackknife regression and generates PV power in kilo-watts in response to inputs that include irradiance, precipitation, ozone, ambient temperature and atmospheric aerosol components. The model is trained and tested on data from National Renewable Energy Laboratory and residual statistical tests are applied to validate the estimation results. An absolute error of less than 1 kW is observed for 90.6 % of the predicted values that corresponds to a percentage error of less than 8.33 % for the 12 kW system under study.