SPLTRAK Abstract Submission
An Evaluation of Empirical Models for use in Normalizing PV Plant Performance Data
Daniel J Fregosi& Michael L Bolen
Electric Power Research Institute, Palo Alto, CA, United States

Monitoring PV plant health informs operations and maintenance activities. Models of the plant are used to calculate expected power, given the meteorological conditions. The expected power is used to normalize plant performance. This work evaluates the accuracy of various empirical, or data-driven, models and compares them to physics-based models by measuring how closely the calculated expected power matches actual power. It was found that empirical models offer greater accuracy and ease of setup than physics-based models. Methods to improve model performance are proposed and evaluated.