Model predictive control based selective harmonic control-PWM strategy for active power filters

Active power filters are widely used to improve power quality and mitigate harmonic distortion caused by non-linear loads and grid-connected generation units. In high-power and medium-voltage applications, conventional high-frequency carrier-based pulse width modulation techniques results in excessive switching losses, reduced efficiency, and converter current derating. This work proposes a novel optimal harmonic control strategy for high-power/medium-voltage active power filters operating at low-switching frequencies. The proposed approach reduces switching losses, increases power density, and lowers system costs. Unlike existing low-switching-frequency methods that primarily regulate the fundamental voltage, the proposed strategy enables precise control of both the magnitude and phase of multiple low-order harmonics, ensuring full active power filter functionality. A Kalman filter provides accurate real-time estimation of grid voltage and current harmonic distortions, which are processed by an optimal model predictive controller. This controller is integrated with an advanced selective harmonic control pulse width modulation technique to regulate current harmonics efficiently. To reduce the computational burden of real-time selective harmonic control pulse width modulation, an artificial neural network is employed for fast and efficient execution. The proposed strategy is compared with a conventional proportional-integral control approach and validated experimentally using a three-level neutral point clamped converter-based active power filter.