ANFIS-based intelligent control for capacitor voltage stabilization and harmonic mitigation in multilevel active power filters

The issue of harmonics in power systems has grown increasingly significant because of the extensive utilization on nonlinear loads, leading to voltage distortion and degradation of power quality. To address these problems, a single-phase shunt-active power filter utilizing a 5-level Cascaded H-Bridge (CHB) inverter is developed, as it provides notable benefits such as modularity, flexibility, and component reduction. However, the CHB inverter encounters voltage imbalance issues across its DC-link capacitors due to uneven switching operations and variations in charging and discharging rates, which negatively affect system efficiency. To mitigate these problems, an Adaptive Neuro-Fuzzy Inference System (ANFIS) control strategy is utilized to ensure the stability of voltages among H-bridge modules and enhance the system’s dynamic performance. Furthermore, a dual-frequency multicarrier pulse width modulation (DFMPWM) technique is integrated to distribute the switching pattern evenly across each H-bridge module to improve voltage symmetry. With the combined ANFIS–DFMPWM approach, the proposed system achieves significant harmonic reduction, maintaining total harmonic distortion well below the IEEE-519 standard limit, while improving the power factor ( P f ) close to unity and minimizing the angle difference between the supply voltages by the source and the resulting current, thereby enhancing overall power quality.