Hybrid data‐model method to improve generation estimation and performance assessment of grid‐tied PV: a case study

Increased installed capacity of distributed photovoltaic (PV) systems has necessitated accurate measurement and tracking of PV performance under locality‐specific conditions of irradiance, temperature, and derate factors. Existing PV generation estimation methods are strictly model based and not responsive to changes in weather and system losses. Metrics computed using these methods, therefore, do not capture the real PV behaviour well. This study proposes a hybrid data‐model method (HDMM) that uses historical PV data in addition to model information to improve the accuracy of generation estimation. The generation estimated by HDMM is used to compute performance metrics – performance ratio, yield, capacity factor, energy performance index, and power performance index – for two real‐world PV systems at Miami (M, 1.4 MW) and Daytona (D, 1.28 MW) for 2017. The significance of these metrics is then evaluated, and a preliminary analysis of inverter efficiencies is provided. Results from this study show that when compared with the existing estimation method, HDMM performs better on an average by 75% for D and 10% for M. Further, at a given point in time, system M is likely to perform better than D. The study gives system installers and other stakeholders better PV system visibility, enabling aggregation and transactive energy.