Multistep model predictive control for power electronics and electrical drives

Download files
Access & Terms of Use
open access
Copyright: Baidya, Roky
Altmetric
Abstract
Model predictive control (MPC) has emerged as a promising control technique for power electronic applications due to its own features, mainly, handling of multi-variables, constraints, fast dynamic response. Among the MPC families, finite control set MPC has become one of the most popular strategies since it takes the switching nature of the power converter. The switches or voltage levels can be directly incorporated in the control formulation as manipulable control-inputs, and thus, no modulator is required. An optimal control problem is formulated using a cost function that forecasts the future system behavior over a finite prediction horizon. The standard optimization process to obtain the optimal control-input is the exhaustive search algorithm (ESA). To date, most of the research is limited to horizon one only (single-step MPC). Recently, it has been shown that increasing horizon length greater than one (multistep MPC), has significant benefits in the steady-state performance when compared to single-step MPCs. However, it poses a high computational burden, and thus, its practical implementation becomes challenging. The aim of this thesis is to study the use of multistep MPC for power electronics systems and electrical drives. The focus is on technical and theoretical issues regarding the optimization process that considers a computationally efficient optimizer namely sphere decoding algorithm (SDA). Moreover, a suitable MPC formulation which includes the full steady-state information (e.g., currents and voltages) is adopted, which is, in fact, a formal MPC approach used in control theory. Special attention is given to a crucial and so far unsolved computational issue faced by the SDA during the transient operation of the system. If not properly addressed, it may have a detrimental effect on the system operation. To solve this issue, an efficient preconditioning approach to the SDA is proposed that exploits the quadratic formulation of the optimal control problem. The outcome of this research is evidence of the potential for multistep MPC in power electronics systems and electrical drives. The theoretical and practical proposals in this thesis allow the multistep MPC to govern power converters and electrical drives over the whole operating conditions of the system with a fast dynamic response, reduced harmonic distortion, and higher efficiency, which enables it as a viable and attractive control alternative for power electronics applications.
Persistent link to this record
Link to Publisher Version
Link to Open Access Version
Additional Link
Author(s)
Baidya, Roky
Supervisor(s)
Fletcher, John
Aguilera Echeverria, Ricardo
Acuna Rios, Pablo
Creator(s)
Editor(s)
Translator(s)
Curator(s)
Designer(s)
Arranger(s)
Composer(s)
Recordist(s)
Conference Proceedings Editor(s)
Other Contributor(s)
Corporate/Industry Contributor(s)
Publication Year
2018
Resource Type
Thesis
Degree Type
PhD Doctorate
UNSW Faculty
Files
download public version.pdf 12.66 MB Adobe Portable Document Format
Related dataset(s)