Model-Free Predictive Control of DC–DC Boost Converters: Sensor Noise Suppression With Hybrid Extended State Observers
Oluleke Babayomi, Zhenbin Zhang, Zhen Li
Abstract
DC–DC boost converters find versatile application as power conversion interfaces to renewable energy sources including wind, solar photovoltaic, fuel cell, and energy storage systems. The model predictive control (MPC) of these power converters is characterized by multivariable, constrained, optimal control with higher performance than linear control methods. However, MPC is limited by sensor measurement noise and model uncertainties. Although model-free predictive control based on the ultralocal model and conventional extended state observer (MFPC-ESO) can mitigate model uncertainties, MFPC-ESO is limited in noise suppression. Noise suppression is an important control objective in order to keep the steady state ripples bounded and guarantee the stability and robust performance of power converters. Therefore, this study proposes the novel hybrid parallel-cascade ESO, which has better measurement noise suppression than the conventional linear ESO. New higher order hybrid ESO structures are also discussed along with their detailed theoretical analyses. The generalized selection and design guidelines for robust control with measurement noise suppression using hybrid ESOs (including cascade-parallel ESO) are presented. The proposed control schemes are experimentally demonstrated for the robust MFPC of a bidirectional dc–dc boost power converter, under conditions including load, input voltage, measurement noise, and model uncertainties.