Predictive algorithm for harmonic mitigation in non-linear loads based on active filters

This paper shows a new strategy to calculate the reference current in a power active filter used to mitigate harmonics injected by non-linear loads. The current is predicted with a Kalman filter a sampling period ahead to reduce and compensate the delay time effects caused by the acquisition times, the data processing and the performance of the filter. The validity of the algorithm is demonstrated by means of simulations in Matlab/Simulink. Several Kalman filters were studied and their capabilities compared to find out the best performance at the lowest computational cost.