Robust Predictive Control of High-Density Cascaded DC Voltage Link Power Converters

Abstract

Cascaded dc voltage link power converters typically employ a significant amount of energy storage at the intermediate link. The dc bus voltage is assumed to be stiff enough to provide decoupling between the source and load, allowing dynamic regulation functions to be implemented at the load and source terminals independently. In these cases, the size of the dc bus capacitance is dictated by requirements of steady-state ripple voltage, ripple current rating, ride-through considerations, and, more importantly, voltage overshoot during source and load transients. Design tradeoffs in realizing a stable system with acceptable transient performance often place a lower bound in the value of dc bus capacitance. In this paper, a robust predictive controller based on integrated and coupled control at the intermediate dc link, source, and load terminals is presented. The approach permits the use of a small amount of intermediate energy storage compared to conventional decoupled designs. The predictive control aspect of the proposed approach features cost function minimization and accurate duty ratio calculations based on converter model that accounts for small values of dc link capacitance. Although this allows stable operation of the system, it leads to persistent steady state error and transient overshoot voltages. The robust aspect of the proposed approach is based on the theory of variable structure systems. It introduces a capacitor charge restoration function that overcomes the drawbacks of the predictive control by squarely addressing uncertainty of model parameters and operating conditions. The resulting significant reduction in size of the capacitive reactive components opens the prospect of replacing electrolytic capacitors that have a limited life span with film capacitors with much longer life in emerging applications. The proposed approach is presented in a step-by-step manner using a dc-to-dc converter example and further extended to a dc-to-ac inverter case. Extensive computer simulations and experimental results from a laboratory-scale prototype of an example system are used to demonstrate the viability of the proposed controller.