Semiparametric Estimation of Solar Generation

Because of the high variability of solar generation, integrating solar energy into the grid requires estimating the total energy demand and total solar generation behind a node using such available data as load measurements at the node and irradiance data. In this work, we propose a flexible, robust semiparametric solar generation estimation methodology with uncertainty bounds, which nonparametrically estimates the total energy demand and parametrically estimates the solar generation based on a photovoltaic performance model. We demonstrate the effectiveness of our methodology on experiments that combine publicly available hourly load and per minute irradiance data with simulated data. We also demonstrate the high accuracy achieved by our methodology if per minute load measurements became available. November 7, 2017