Assessment of voltage harmonic compensation with H-bridge circuit using EMD derivatives

Abstract

The extraction of fundamental voltage and controlling the H-bridge circuit, called series active power filter (SeAPF), for voltage harmonic compensation have always been a research concern. This paper reports data-adaptive methods for accurately extracting fundamental voltage and harmonics. In this context, empirical mode decomposition (EMD) and its derivatives have recently become powerful harmonic detection methods. These data-adaptive versions decompose non-stationary polluted signals into frequency-dominated intrinsic mode functions ($imfs$). In this framework, harmonics for compensation are extracted using SeAPF and synthesized using EMD derivatives. These include EMD, ensemble EMD (EEMD), and complete EEMD with adaptive-noise (CEEMDAN) algorithms. CEEMDAN addresses the mode-mixing issue of EMD and the amplitude deficiency of EEMD techniques. Further, the optimal switching signals for the SeAPF circuit are accomplished by a model predictive controller (MPC). The EMD-variants with MPC prove to be a strong asset in improving the performance of SeAPF. The efficacy of the EMD variants is demonstrated through MATLAB/Simulink and real-time simulations conducted using an OPAL-RT OP4510 real-time simulator. Compared to EMD and EEMD, the results show that CEEMDAN has improved fundamental extraction with low total harmonic distortion (THD), meeting the IEEE 519-2022 standards. Further, the active filtering efficiency of SeAPF has significantly improved with the CEEMDAN approach.