Adaptive Kalman filter for harmonic detection in active power filter application

This paper deals with the harmonic detection which is decoupled from the operation of active power filter. Kalman filter for harmonic detection based on a stochastic state-space model is proposed. However, it is a challenging task in large time varying system to know the process and noise covariance matrices Q and R. In this active power filter application, the current sensor TLC277CD and ADC LTC1403A which introduce load current measurement inaccuracies are analyzed to decide a rough R. Based on that R is exactly known, two adaptive Kalman filter algorithms to scale Q are proposed. One of the adaptive Kalman methods switches two basic Q matrices depending on the system in transient- or steady-state. The other Kalman algorithm tunes an optimal Q at each step by using the information of innovations sequence. The simulation results show that both adaptive Kalman filters have better dynamic performance than the regular Kalman filter.