UC San Diego

Electric grid transients stress power systems and can have cascading effects leading to widespread

damage. Transient effects and oscillations are made worse by an increased portion of power supplied by sources without any rotational inertia, such as DC to AC inverters connected to photovoltaics or energy storage media. Yet, the solid state architecture of modern inverters allows

unprecedented response time, enabling novel control schemes.

It is difficult to leverage the speed of modern inverters because of the comparably slower

response rate of microprocessors. Electric grid transients are characterized by a mix of gridfrequency and lower frequency components, and a high levels of harmonic distortion that accompanies the most common inverter architectures, forming challenges in signal processing. Furthermore, the complexity of electric grids makes model development and validation difficult.

This dissertation makes contributions toward overcoming these difficulties and demonstrates

feasibility of controlling inverter output to mitigate transient effects on an electric grid. A gridxix

connected reference circuit allows recreation of power oscillations when an inductive load is

switched in. Parallel processing via a field programmable gate array allows signal processing that

demodulates and filters inverter-supplied power, removing grid frequency components and higher

harmonics. The result is clean real-time active and reactive power signals sent to a microprocessor

for control. This data is used for system identification, greatly reducing the burden of model development. Finally the improved signal serves as input to a robust control scheme that adjusts inverter

active and reactive power output for disturbance rejection of transients created by the reference

circuit.