Learning to leverage flexible electric loads

Electric loads like air conditioners and electric vehicles can enhance the efficiency and reliability of power systems by allowing system operators to occasionally moderate their power consumption. This is called demand response, and is regarded as a key enabler of renewable energy. A major challenge to demand response is uncertainty – precise information about loads like buildings and appliances is often unavailable. Hence, power system operators often can only precisely measure electric loads by utilizing them for demand response. This research aims to design machine learning algorithms that deploy loads for demand response while learning their uncertain characteristics for future use