Data-driven model for rooftop excess electricity generation. Kiguchi, Y., Heo, Y., & Choudhary, R. In 14th International Conference of IBPSA, Building Simulation 2015, pages 2631–2638, Hyderabad, 2015. IBPSA.
Data-driven model for rooftop excess electricity generation [pdf]Paper  abstract   bibtex   
Large scale rooftop PV deployment generates significant amounts of excess electricity (the difference between real-time PV generation and on-site consumption). Accurate forecasting of excess electricity generation becomes important for energy traders participating in a liberalised electricity market in order to avoid penalties imposed by the market in case of imbalance between actual and predicted values. This paper proposes a data driven Gaussian process model for forecasting excess generation of electricity. An illustrative study, with a-year long dataset from 287 households, is used to derive the forecasting model. Results show that a-year long data from 18 households is sufficient for accurate prediction of excess generation across a uniform geographic and socio-economic setting.

Downloads: 0