Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Electrical and Computer Engineering


Dr. Jin Jiang


Islanded microgrids powered primarily by photovoltaic (PV) arrays present a challenging control problem due to the intermittent production and the relatively close scale between the sources and the loads. Energy storage in such microgrids plays an important role in balancing supply with demand, and in extending operation during periods when the PV supply is not available or insufficient. The efficient operation of such microgrids requires effective management of all resources. A predictive energy management strategy can potentially avoid or effectively mitigate upcoming outages. This thesis presents an energy management system (EMS) for such microgrids. The EMS uses a predictive approach to set operational schedules in order to (a) prolong the supply to critical system loads and (2) minimize the chances and duration of system-wide outages, specifically through pre-emptive load shedding. Online weather forecast data has been combined with the PV system model to assess potential energy production over a 48 hour period. These predictions, along with load forecasts and a model of the energy storage system, are used to predict the state-of-charge of the storage devices and characterize potential power shortages. Pre-emptive load shedding is subsequently planned and executed to avert outages or minimize the duration of unavoidable outages. A bounding technique has also been proposed to account for uncertainties in estimates of the stored energy. The EMS has been implemented using an event-driven framework with network communication. The approach has been validated through simulations and experiments using recorded real-world solar irradiance data. The results show that the outage durations have been reduced by a factor of 87% to 100% for an example operating scenario, selected to demonstrate the features of the scheme. The impact of uncertainties in the prediction models has also been investigated, specifically for the PV system rating and the battery capacity. A technique has been developed to compensate for such uncertainties by analyzing the data streams from the source and storage units. The technique is applied to the developed EMS strategy, where it is able to shorten the total outage duration by a factor of 12% over a 42-day scenario exhibiting a variety of irradiance conditions.