Extended Kalman Filtering for Floating Capacitor Voltage Estimation on Triple Star Bridge Cells

Abstract

Modular multilevel cascade converters (MMCCs) have emerged as one of the most attractive topologies for medium and high-voltage applications due to their modularity, scalability, redundancy, and high power quality. Voltage balancing in power submodule (SM) capacitors plays a critical role in the internal energy balancing of the MMCC, making monitoring SM capacitor voltages a crucial task. However, achieving higher operating voltages requires a substantial increase in the number of voltage sensors and communication lines. This escalation in hardware complexity renders the system more reliant on sensors, reducing its reliability. Several techniques for estimating capacitor voltages have been presented to address this control and design burden. This work proposes an extended Kalman filter (EKF)-based observer for capacitor voltage estimation of all SMs in a triple-star bridge converter (TSBC). The proposed approach operates effectively under both open-loop and closed-loop conditions during transients and steady-state operation, enabling a decoupled controller using just one voltage and one current sensor per cluster. Experiments conducted in a TSBC composed of 27 SMs demonstrate the effectiveness of the proposed approach during transients and steady-state operation.