Combining Mathematical Morphology and Unscented Kalman Filter for Digital Protective Relaying

Abstract

The paper presents a digital filter algorithm, combining mathematical morphology (MM) and unscented Kaman filter (UKF), for microprocessor-based protective relaying. Firstly, the state and observation equations of UKF are designed based on the state variables of DC-offset component, fundamental angular frequency, and fundamental component of the fault signal. Approximation signals and detail signals are derived from the multi-scale morphological decomposition of the sampling signal. The approximation signals are employed to update the observations of UKF to improve the convergence speed of the filter algorithm, and the detail signals are applied to calculate the measurement noise variance to improve the convergence precision of the filter. The proposed MM-UKF algorithm is evaluated on the data collected from simulation model built in Matlab/Simulink.