Adaptive Model Predictive Control for <i>LCL</i>-Filter Grid-Tied Inverters
Hasan Zamani, Karim Abbaszadeh, Mohammad Hadi Karimi, Johan Gyselinck
Abstract
This article proposes adaptive continuous-control-set model predictive control (CCS-MPC) for three-phase <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCL</i> -filter grid-tied inverters in the stationary frame. The proposed control provides stability against the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCL</i> -filter resonance and removes the need for active damping or passive damping. Furthermore, it can track sinusoidal references in the stationary frame close to zero offsets. Inductance variation of a grid is a challenging issue, which can shift the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCL</i> -filter resonance. Even though CCS-MPC generally has good stability for parameter mismatches, the control is optimized for the nominal value of the parameters. A relatively large variation may deteriorate the performance of the control and cause instability in the closed-loop system. To solve this problem, the concept of CCS-MPC is developed to work with a variety of the grid-inductance changes. The controller coefficients are chosen with a sensitivity analysis to provide enough gain margins and phase margins for the whole interval of the presumption grid-inductance variation. Also, this selection guarantees the control robustness against mismatches of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCL</i> -filter capacitance and inductance. The real-time computation of advanced controllers is a challenging issue. The computation burden of the proposed CCS-MPC is mostly performed offline and just some simple numerical expression is executed in real time. Thus, it does not require a powerful but expensive processor for implementation. The simulation and experimental results prove the performance of the proposed controller for handling the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LCL</i> -filter resonance, grid-inductance variations, and tracking sinusoidal references.