Προβλεπτικός‘ελεγχος Αντιστροφέων Τριών Επιπέδων Για Σύνδεση Στο Δίκτυο Με Lcl Φιλτρο Model Predictive Control Of Grid Connected Neutral Point Clamped Converters Through Lcl Filters

Abstract

The objective of this thesis is the development of a grid connected control system of a three level Neutral Point Clamped converter interfacing the grid through LCL filters with switching losses minimization and active damping of the filter resonance. In order to achieve optimum efficiency, an online Model Predictive optimization strategy is considered minimizing switching losses, namely Model Predictive Direct Power Control(MPDPC), and is properly extended to meet active damping of LCL filter requirements. As an outcome of the online op- timization process, switching frequency of the converter is variable, thus it is compared in simulation environment with the traditional Direct Power Con- trol(DPC) technique. DPC with and without LCL output filter are thoroughly presented and an analytical way of DPC lookup table design is proposed based on virtual flux quantities. Moreover an effective three-level hysteresis controller design is proposed in order to smooth spikes present in reactive power with conventional DPC. While presenting the MPDPC solution algorithm utilized in this thesis, two new techniques are introduced, aiming at lowering the com- putational effort required for conventional MPDPC reach a solution. Finally, MPDPC with LCL filters and active damping is evaluated in a broad range of operating points, and proof of concept hardware is implemented. The main conclusions drawn from this study is that high performance converters can be designed utilizing MPC concepts making use of today computational power, while providing a framework for traditional techniques, i.e active damping, to be integrated with Online Optimization Controllers.