Circulating current mitigation for renewable-based modular seven-level converter using deep learning-optimized fractional-order proportional resonant controller

Modular multi-level converters (MMCs) are often used for high and medium voltage applications. However, to reduce losses and costs, many researchers prefer a half-bridge converter. In addition, the half-bridge-based MMC is vulnerable in the event of an error, so the full-bridge MMC is used here to work with faulty network states. The losses and harmonics in the system could be reduced by using an appropriate arm voltage and circulating current control model. In order to operate the MMC in a grid-tied renewable system, both outer and inner loop control were performed. In order to realize outer-loop control, a fractional-order proportional–integral–derivative controller using a deep learning technique is proposed. An active power filter-based fractional-order proportional resonant controller with improved pulse width modulation achieves arm balancing with harmonic mitigated circulating current regulation. The simulation shows that the proposed method reduced the current and voltage harmonics to 71.56% and 10.42% through an improved control strategy based on pulse width modulation.