Real-Time Power and Intelligent Systems (RTPIS) Laboratory

Real-Time Predictive Modeling and State Estimation For Power Systems


To be able to push the power grid closer to its operating limit and still have enough security margin is a challenging task and one that requires foresight/prediction in order to implement proactive control rather than reactive control. An alternative approach to real-time predictive modeling and state estimation is investigated. It is aimed at providing insight as to how the system states will change from current time instance to the next. At first, unobservable states including the mechanical state variables of generators are made observable using data from phasor measurement unit (PMU). This data is then used by nonlinear function approximators and Kalman Filter based algorithms to predict system states for the next time instance. The algorithm is run in a distributed computing framework that allows for linear scale-up and hence of practical use for application in large scale power systems.