Skip to main navigation Skip to search Skip to main content

Enhanced Performance of a Grid-Connected 15-Level Switched-Capacitor Multilevel Inverter Using Inverse Model Predictive Control

Research output: Contribution to journalArticlepeer-review

Abstract

This paper presents an enhanced single-phase 15-level switched-capacitor multilevel inverter (15L-SCMLI) that addresses the critical limitations of SCMLI designs, including passive capacitor voltage balancing, which requires oversized capacitors; inrush charging current spikes, which necessitate high-rated power switches; and high total harmonic distortion (THD). The proposed inverter generates 15 voltage levels from a dual DC source by combining an H-bridge with a switched-capacitor topology that can boost the input voltage. The topology achieves high voltage gain with reduced component count, lowering cost and complexity while minimizing stress on power switches and the capacitor. To enhance dynamic performance and reliability, an inverse model predictive control (IMPC) strategy is integrated with a capacitor voltage-balancing algorithm and used to regulate the inverter current. This approach ensures a fast transient response, minimal overshoot, reduced computation time, and THD. Comprehensive analysis and design procedures are presented to validate the proposed approach. The experimental findings of a laboratory-scale prototype connected to the grid confirm the effectiveness of the topology and control strategy. Benchmarking against state-of-the-art solutions demonstrates the superiority of the proposed 15L-SCMLI in efficiency, power quality, and overall component cost-effectiveness.

Original languageEnglish
JournalIEEE Transactions on Power Electronics
DOIs
Publication statusAccepted/In press - 2026

Keywords

  • capacitor balancing
  • inverse model predictive control
  • Multilevel inverters
  • switched capacitor

Fingerprint

Dive into the research topics of 'Enhanced Performance of a Grid-Connected 15-Level Switched-Capacitor Multilevel Inverter Using Inverse Model Predictive Control'. Together they form a unique fingerprint.

Cite this