Validation of Inverter Labeling with Plant Transfer Functions

The large quantity of data sources found within a utility scale photovoltaic plant presents data quality control challenges. One potential issue is mislabeling of the plant's component outputs (e.g. production measurements made at the combiner or inverter level). If a component's output is incorrectly labeled, it presents an obstacle to plant monitoring and maintenance, as operators will not know where fixes are needed. This study aims to demonstrate the possibility of utilizing the Cloud Advection Model to perform quality checks on the labeling of production outputs at a plant component level based on information about the plant's spatial layout. Results utilizing simulated data showed that the plant transfer function predicted by the CAM could provide discrimination between plant segments that are separated in the cloud motion direction. The discrimination occurred primarily through the phase of the transfer function, but in cases where the spatial dispersion of the plant varied significantly in the cloud motion direction, changes to the transfer function bandwidth were also observable. This methodology shows promise using the simulated plant data in this study, which warrants further study and practical validation of this method utilizing real plant data.