A Machine Learning Modeling Method for Switched Reluctance Motors Based on Few Preprocessed Flux Linkage Data

This article proposes a machine learning method to build the model of a switched reluctance motor (SRM) using few preprocessed flux linkage data. Firstly, the improved torque balance method is used to obtain the accurate flux linkage data without redundant experiments. Secondly, two special data pre...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Journal of electrical engineering & technology Jg. 20; H. 5; S. 3445 - 3456
Hauptverfasser: zhao, Yan, Zhu, Jingwei, Ren, Ping, Jing, Zhe
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Singapore Springer Nature Singapore 01.07.2025
Springer Nature B.V
대한전기학회
Schlagworte:
Electrical Engineering
Electrical Machines and Networks
Electronics and Microelectronics
Engineering
Instrumentation
Original Article
Power Electronics
전기공학
Nonlinear modeling
Switched reluctance motor
Data preprocessing
Improved tuna swarm algorithm
Support vector regression algorithm
ISSN:1975-0102, 2093-7423
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This article proposes a machine learning method to build the model of a switched reluctance motor (SRM) using few preprocessed flux linkage data. Firstly, the improved torque balance method is used to obtain the accurate flux linkage data without redundant experiments. Secondly, two special data preprocessing steps are proposed, which are nonlinear preprocessing and angle mapping, respectively. The first step provides beneficial nonlinearity for algorithms, and the second step improves the linearity of flux linkage at small angles by the proposed mapping function. Thirdly, support vector regression algorithm optimized by the improved tuna swarm algorithm (ITSO-SVR) is employed to establish the flux linkage model. Based on the flux linkage model, the current and torque models are easily built by ITSO-SVR to complete the nonlinear modelling of SRM. Finally, the effectiveness of the proposed method is verified. The preprocessing method is verified to reduce the modeling difficulty. Besides, ITSO-SVR facilitates the swift and efficient modeling without any pre-storge or complex calculations. The experiments under the CCC and APC algorithms indicate that the established model exhibits high accuracy, fast speed and strong generalization capability.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1975-0102
2093-7423
DOI:10.1007/s42835-025-02265-8