A Two-Staged Optimization Approach to Modulated Model-Free Predictive Current Control for RL-Connected Three-Phase Two-Level Four-Leg Inverters

Abstract

This paper proposes a modulated model-free predictive current controller for three-phase two-level four-leg RL-connected inverters based on a two-staged optimization. The proposed scheme is model-free and does not require any knowledge of system parameters nor the load model to perform current predictions. To improve current prediction accuracy, two input voltage vectors are linearly modulated in every sampling period, doubling the current difference calculation and update frequency. Moreover, the input voltage vectors are independently chosen and optimized in each stage using the integrated approach of cost function minimization and duty ratio optimization, known as the two-staged process. The obtained vectors are adaptive and modulated with variable durations. The proposed method is implemented in a drive circuit using TMS320F28379D and put to tests under various loading conditions of steady-state, dynamic, and parameter mismatch. Results from simulations and experiments demonstrate the feasibility of the proposed method.