A new approach for event detection based on energy functions

This paper proposes a novel framework for event detection based on energy functions. The key idea is that any event (or disturbance) in the system will leave a signature (like a fingerprint) in Wide Area Measurement Systems (WAMS) data-sets. We show that signatures for four broad classes of disturbance events are buried in the components that constitute the energy function for the system. This establishes a direct correspondence (or mapping) between an event and certain component(s) of the energy function. Dynamic states required to compute the energy function are estimated with a particle filter. The estimated states along with wide-area phasor information from PMUs is used to reconstruct the components of the energy function. The individual components of the energy function are tracked dynamically and the sensitivity of these components is used to detect a disturbance in the system. Results show promising distinction between signatures of different events.