High-speed nonlinear model predictive control of an interleaved switching DC/DC-converter

Abstract

This paper presents a high speed nonlinear model predictive control (MPC) strategy for (interleaved switching) buck-type DC/DC-converters. Compared to known MPC strategies the proposed approach allows for longer prediction horizons while keeping the calculation times in the range of a few µs. This is achieved by tailoring the MPC strategy to the efficient implementation on an FPGA, exploiting the inherent parallel computation capabilities of FPGAs. Moreover, the utilization of a numerically efficient envelope model in the MPC is an important building block to reach the desired fast calculation times. The MPC strategy systematically takes into account input and state constraints, features a high dynamic and accurate control performance, exhibits constant switching frequencies and enables an interleaved switching operation. The proposed MPC strategy is tested in detailed simulations and in hardware-in-the-loop (HIL) experiments. An accurate and fast closed-loop dynamics can be achieved, also with respect to unknown fast load variations and parameter uncertainties. It is finally shown that the proposed MPC strategy has the potential to outperform the control performance of existing state-of-the-art controllers for DC/DC-converters.