Improved Grid Reliability by Robust Distortion Detection and Classification Algorithm

Deviations from normal power grid operations, such as incipient faults, equipment damage, or weather related effects, have characteristic signatures in the current and voltage waveforms. Detecting and classifying such signal distortions as quick as possible can contribute to grid reliability since grid events can be responded to in time, i.e. before they lead to an outage. This paper proposes a new distortion detection algorithm, based on computationally very lightweight operations. The method does not require large datasets, has a small memory footprint, and therefore can be easily implemented on decentralized, embedded systems. This detection method constitutes the core of an overarching algorithm which accurately classifies the event even in case of a malfunctioning device and normal switching action. The paper investigates the performance of this new algorithm and evaluates it with four case studies for High Impedance Faults occurring on an IEEE 9 bus system.