EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Real-Time Identification and Nonlinear Control of a Permanent-Magnet Synchronous Motor Based on a Physics-Informed Neural Network and Exact Feedback Linearization

Sergio Velarde-Gomez, Eduardo Giraldo

Information · 19.09.2024 · 15 (9) · 577

DOI: 10.3390/info15090577 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Information, online erschienen am 19.09.2024. Verfasst von Sergio Velarde-Gomez und 1 weiteren Autor:innen. Die Arbeit wurde bisher 20-mal zitiert (Crossref) und nennt 20 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This work proposes a novel method for the real-time identification and nonlinear control of a permanent-magnet synchronous motor (PMSM) based on a Physics-Informed Neural Network (PINN) and the exact feedback linearization approach. The proposed approach is presented in a direct-quadrature framework, where the quadrature current and the rotational speed are selected as outputs and the direct and quadrature voltages are selected as inputs. A nonlinear difference equation is selected to describe the physical dynamics of the PMSM, and a PINN is designed based on the aforementioned structure. A simplified training scheme is designed for the PINN based on a least-squares structure to facilitate online training in real time. A nonlinear controller based on exact feedback linearization is designed by considering the nonlinear model of the system identified based on the PINN. Therefore, the proposed approach involves identification and control in real time, where the PINN is trained online. In order to track the reference for the rotational speed, a nonlinear controller with integral action based on exact feedback linearization is designed based on a linear quadratic regulator. As a result, the proposed approach can be used to identify the system to be controlled in real time, and it is able to track any small change in the real model; in addition, it is robust to both external and internal disturbances, such as variations in torque load and resistance. The proposed approach is evaluated through simulation and using a real PMSM, and the results of reference tracking are evaluated under disturbances. The identification performance is evaluated by using a Taylor diagram under closed-loop and open-loop structures, where ARX and NARX structures are used for comparison. It is thereby verified that this novel proposed control approach involving a PINN-based model can adequately track the dynamics of a PMSM system, where the performance of the proposed nonlinear control is maintained even when using the identified model based on the PINN.

Zitieren

APA

Velarde-Gomez, S., & Giraldo, E. (2024). Real-Time Identification and Nonlinear Control of a Permanent-Magnet Synchronous Motor Based on a Physics-Informed Neural Network and Exact Feedback Linearization. Information, 15(9), 577. https://doi.org/10.3390/info15090577

Vancouver

Velarde-Gomez S, Giraldo E. Real-Time Identification and Nonlinear Control of a Permanent-Magnet Synchronous Motor Based on a Physics-Informed Neural Network and Exact Feedback Linearization. Information. 2024;15(9):577. doi:10.3390/info15090577

BibTeX
@article{VelardeGomez2024,
  author = {Velarde-Gomez, Sergio and Giraldo, Eduardo},
  title = {Real-Time Identification and Nonlinear Control of a Permanent-Magnet Synchronous Motor Based on a Physics-Informed Neural Network and Exact Feedback Linearization},
  journal = {Information},
  year = {2024},
  volume = {15},
  number = {9},
  pages = {577},
  doi = {10.3390/info15090577}
}
Kostenlos starten →