Robust Explicit Model Predictive Control based on State Feedback Linearization for Buck Converter

摘要

Explicit model predictive control (EMPC) regards the linear time-invariant model as the prerequisite to deduce the offline control laws for online control. Thus, EMPC cannot be directly applied to Buck converters with variable load. A tradeoff method is to linearize Buck model under a specific load to obtain the time-invariant model to perform EMPC. However, since the obtained offline control laws is derived under the specific load, its performance will be degraded when the load varies. To fix this problem, this paper proposes to utilize State Feedback Linearization (SFL) technology to transfer the variable load term from model parameter matrix to state variables and control parameter. Based on such an equivalent model, EMPC can be directly performed and load-independent offline control laws are obtained. With such a methodology, only the load-dependent state variables and control parameter need to be updated online to adapt load variation, and the obtained control laws need not to be changed and can be discretized into a look-up table to reduce online computational burden, which suits the high switching frequency scenarios well. Details of the proposed methodology are given in this paper. Simulation and experimental results are also provided to demonstrate the effectiveness of the proposed control methodology.