LAUSR

This paper is motivated by the fact that behind-the-meter solar farms are being increasingly deployed in California and elsewhere in recent years. The objective is to use real-world micro-PMU measurements at a 4.3 MW behind-the-meter solar Photovoltaic (PV) farm to build a foundation for event-based situational awareness and its data-driven application. Two essential tasks are conducted. First, through developing an automated event region identification mechanism, we identify whether an event at a behind-the-meter solar farm is “locally-induced”, i.e., it is caused by the solar farm, thus potentially indicating internal issues in the solar farm, or “grid-induced”, i.e., it is caused by something else on the grid, thus revealing how the solar farm responded to external disturbances. We show that this is a highly challenging task in practice: the conventional impedance-based method is ineffective, the statistical method and the machine learning method each has its weaknesses. Accordingly, a novel mixed-integrated method is proposed and tested that can achieve very high performance metrics. The proposed mixed-integrated method also closes the gap between the accuracies in identifying grid-induced events versus locally-induced events. Second, the outcome of automated event region identification is used to unmask the constructive use of the proposed analysis. Practical use cases are proposed to take advantage of the situational awareness that we gain from analyzing both types of events to provide critical reporting, unmask trends and relationships, adjust control parameters, or take remedial actions when needed.