Gradient-Based Predictive Pulse Pattern Control with Improved Steady-State and Dynamic Behavior

Abstract

This paper presents an optimal control strategy for grid-connected converters with LCL filters. Specifically, a gradient-based model predictive control (MPC) scheme is developed to manipulate optimized pulse patterns (OPPs) in real time such that very low grid current harmonic distortions are achieved even at switching frequencies close to the filter resonance frequency. To fully exploit the potential of OPPs—in terms of their ability to produce very low harmonic distortions—optimal steady-state reference trajectories are derived based on the OPP in use. Thanks to this, the proposed MPC scheme achieves excellent steady-state performance by producing the optimal output ripple. However, the rigid nature of OPPs may lead to the underutilization of the converter voltage margin during transients. To avoid this, a three-phase pulse insertion strategy is proposed wherein the OPP is locally overwritten such that the switch positions that can result in a very fast dynamic performance become available to the controller. The effectiveness of the proposed control strategy is verified with a system consisting of a two-level converter connected to the grid via an LCL filter.