Model-Free Predictive Control of Three-Level DC/DC Converter Based on Adaptive Super-Twisting Sliding Mode Observer

Aiming at the problem that the control performance of three-level bidirectional DC/DC converter for energy storage is susceptible to parameter disturbance, a model-free predictive control strategy for three-level DC/DC converter based on adaptive super-twisting sliding mode observer is proposed. Firstly, the ultra-local model is used to replace the conventional prediction model with inductance and capacitance parameters to reduce the parameter dependence. Then, a super-twisting sliding mode observer is designed to estimate the unknown part of the ultra-local model, and an adaptive gain is introduced to dynamically adjust the observer parameters, which solves the challenge of gain selection. Finally, the cost function of multi-objective optimization is designed to realize model-free predictive control and input-side capacitor voltage balance. The simulation results show that the proposed strategy has good dynamic and steady-state performance and parameter robustness.