Long-Prediction-Horizon FCS-MPC of Six-Phase PMSM Motor Based on Merge Sorting Sphere Decoding

The control performance of model predictive control based motor drives can be enhanced as the prediction horizon increases, but at the expense of large computational burden. To solve this problem, this paper proposes a sphere decoding algorithm (SDA) based on merge sorting for the MPC-controlled six...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on energy conversion Ročník 40; číslo 4; s. 2884 - 2893
Hlavní autoři: Wang, Mengdi, Yang, Kai, Luo, Yixiao
Médium: Journal Article
Jazyk:angličtina
Vydáno: New York IEEE 01.12.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Témata:
Decoding
Merge sorting
model predictive control
Motors
Prediction algorithms
Predictive control
Search methods
Search process
six-phase permanent magnet synchronous motor
Sorting
Sorting algorithms
sphere decoding algorithm
Switches
Synchronous motors
Thermal stability
Torque
Vectors
Voltage control
ISSN:0885-8969, 1558-0059
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The control performance of model predictive control based motor drives can be enhanced as the prediction horizon increases, but at the expense of large computational burden. To solve this problem, this paper proposes a sphere decoding algorithm (SDA) based on merge sorting for the MPC-controlled six-phase PMSM drive. First, the form of the current predictive control optimization problem is reconstructed. Then, instead of depth-first strategy in conventional SDA, the search strategy of the search tree is changed to breadth-first. Meanwhile, the merge sorting algorithm is proposed in the horizontal comparison, which significantly simplifies the search process. In this way, the problem of heavy computational burden of MPC is solved. Meanwhile, the control performance is enhanced through long prediction horizon MPC. The experimental results verify the excellent performance of the proposed method.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0885-8969
1558-0059
DOI:10.1109/TEC.2025.3556144