Validated real-time energy models for small-scale grid-connected PV-systems

This paper presents validated real-time energy models for small-scale grid-connected PV-systems suitable for domestic application. The models were used to predict real-time AC power output from a PV-system in Dublin, Ireland using 30-min intervals of measured performance data between April 2009 and March 2010. Statistical analysis of the predicted results and measured data highlight possible sources of errors and the limitations and/or adequacy of existing models, to describe the temperature and efficiency of PV-cells and consequently, the accuracy of power prediction models. PV-system AC output power predictions using empirical models for PV-cell temperature and efficiency prediction showed lower percentage mean absolute errors (PMAEs) of 7.9-11.7% while non-empirical models had errors of 10.0-12.4%. Cumulative errors for PV-system AC output power predictions were 1.3% for empirical models and 3.3% for non-empirical models. The proposed models are suitable for predicting PV-system AC output power at time intervals suitable for smart metering.