Parallel Sphere Decoding Algorithm for Long-Prediction-Horizon FCS-MPC

Researchinterest in finite-control-set model-predictive control (FCS-MPC) for power conversion devices has grown in recent years. Particularly, long-prediction-horizon FCS-MPC provides promising results in recent research works. However, its practical implementation is not generally straightforward...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on power electronics Jg. 37; H. 7; S. 7896 - 7906
Hauptverfasser: Zafra, Eduardo, Vazquez, Sergio, Regalo, Carlos, Lecuyer, Vicente Baena, Alcaide, Abraham Marquez, Leon, Jose I., Franquelo, Leopoldo G.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York IEEE 01.07.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
Algorithms
Cost function
DC–AC power converter
digital control
Energy conversion
Field programmable gate arrays
Iterative methods
Mathematical models
Power harmonic filters
Prediction algorithms
Predictive control
Predictive models
Switches
System on chip
Uninterruptible power systems
ISSN:0885-8993, 1941-0107
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Researchinterest in finite-control-set model-predictive control (FCS-MPC) for power conversion devices has grown in recent years. Particularly, long-prediction-horizon FCS-MPC provides promising results in recent research works. However, its practical implementation is not generally straightforward due to its inherently large computational burden. To overcome this obstacle, the problem can be formulated as a least-squares integer program. A sphere decoding algorithm (SDA) is a branch-and-bound algorithm proposed in previous works as an efficient approach to solve this problem. In these works, the SDA is formulated as an iterative process, where simultaneous search is not possible. A parallel and fully scalable SDA design is proposed in this article. The design is implemented in the field-programmable gate array of a modern field-programmable system-on-chip platform. Thanks to the proposed parallelization, the required execution time is greatly reduced. Experimental results prove the feasibility and performance improvements of the proposed implementation.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0885-8993
1941-0107
DOI:10.1109/TPEL.2022.3150433