Classifying Reactive Power Control Laws of Behind-the-Meter Solar Photovoltaic Inverters

Various types of reactive power control laws are heterogeneously used for inverter-based resources (IBRs) in distribution grids. As a result, grid operators may not be aware of which type of control law is used by a particular IBR. Different control laws imply different voltage support behaviors, which need to be known for the development of accurate computational models for power system analyses. To help mitigate this challenge, this paper develops two classification algorithms to identify which type of control law governs the reactive power output of a behind-the-meter solar photovoltaic inverter when the specific control law selected by the IBR owner is unknown. In particular, the two algorithms require only aggregated smart meter measurements to identify candidate reactive power control laws (constant power factor or volt-var control) for distribution network inverters. We present a case study to assess their classification accuracy and evaluate the algorithms' performance in the context of noisy measurements. Our neural network-based classifier is shown to have a higher classification accuracy and performs better under varying levels of noise.