Capacitor voltage estimation for predictive control algorithm of flying capacitor converters

Multilevel converters have emerged as a promising alternative to traditional two level converters, especially flying capacitor converters because of the fact that this topology requires only a main DC-link voltage and presents a easy way to increase the output voltage levels by increasing the number of cells. Unfortunately, a balancing of capacitor voltage is required. Recently, predictive control algorithms have been presented in order to control not only the output current but also to achieve good performance in the balancing of the capacitor voltages. For this purpose, it is necessary to know the state of these voltages generally taking a measurement of them, therefore the number of sensors required will be increased regarding the output voltage levels desired. This paper presents an estimation of the capacitor voltages using a discrete Kalman filter. This algorithm is employed to determinate correctly the system state and thus provides this information to the predictive controller in order to determinate the best switching combination to be applied in the next sample period.