Capacitor Voltage Balancing Control Based on Multiple Sequence Prediction

ABSTRACT Modular multilevel converter (MMC) has been widely adopted in flexible direct current transmission due to the high flexibility of its modular topology and has become one of the core components in modern power electronic conversion systems. But the unbalanced submodule (SM) capacitor voltage severely limits the MMC's performance. The complexity of capacitor voltage balancing control closely depends on the number of SMs. This paper analyzes the factors of SM capacitor voltage unbalance, identifies the shortcomings of traditional capacitor voltage balancing control, derives the relationship between capacitor voltage deviation and its capacitance value, and proposes capacitor voltage balancing control using multiple sequence predictions. The spectral characteristics of capacitor voltage and voltage deviation are analyzed using Fourier decomposition, and the main spectrum is selected to complete the real‐time prediction of capacitor voltage and voltage deviation. The SM trigger sequence is anticipated, and its control is performed using the neural network. By doing so, the speed of capacitor voltage balancing control is greatly improved while satisfying the accuracy, and the capacitor voltage offset is reduced. The MMC model is built in MATLAB/Simulink to verify the feasibility of the proposed scheme.