Reliable power disturbance detection using wavelet decomposition or harmonic model based Kalman filtering

This paper proposes a method for detecting Power Quality (PQ) disturbances measured on distribution networks. The detection efficiencies of two types of detection vectors are compared: the estimation error returned by a Kalman Filter (KF) based on a harmonic model and the detail coefficients given by a Multi-Resolution Analysis (MRA). The detection capabilities of different state-models and wavelet families are tested on waveform recordings of PQ disturbances including faults, transformer energizing, capacitor bank switching or ripple control.