Experimental Data-Driven LSTM Modeling for IPMSM Drive Systems
IEEE Access · 2026 · 14 · 57463-57474
DOI: 10.1109/access.2026.3683374 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Yuki Shimizu und 1 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Interior permanent magnet synchronous motors are key components in electric vehicle powertrains, wind turbine generators, and industrial drives, where accurate current prediction models are needed for high-performance control and condition monitoring. However, conventional $dq$ -axis voltage equations and finite-element-based lookup models struggle to capture nonlinear phenomena such as magnetic saturation, cross-coupling, and spatial harmonics under realistic manufactured variations. This paper proposes a purely data-driven approach in which a long short-term memory network is trained directly on experimentally measured three-phase voltages, currents, rotor position, and speed. On test conditions not seen during training, the proposed model achieves root-mean-square errors as low as 0.069 A at 10 Arms and 0.032 A at 1 Arms, corresponding to about 96% error reduction compared with a $dq$ -axis voltage equation model and about 50% improvement over gated recurrent unit and Transformer baselines. Fast Fourier transform analysis shows that it reproduces phase-current total harmonic distortion within 0.2 percentage points and the main harmonic amplitudes within 7%. These results indicate that experimental-data-driven long short-term memory models can accurately represent the nonlinear behavior of interior permanent magnet synchronous motors and provide a practical basis for advanced control and monitoring functions in motor drive systems.
Zitieren
APA
Shimizu, Y., & Akatsu, K. (2026). Experimental Data-Driven LSTM Modeling for IPMSM Drive Systems. IEEE Access, 14, 57463-57474. https://doi.org/10.1109/access.2026.3683374
Vancouver
Shimizu Y, Akatsu K. Experimental Data-Driven LSTM Modeling for IPMSM Drive Systems. IEEE Access. 2026;14:57463-57474. doi:10.1109/access.2026.3683374
BibTeX
@article{Shimizu2026,
author = {Shimizu, Yuki and Akatsu, Kan},
title = {Experimental Data-Driven LSTM Modeling for IPMSM Drive Systems},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {57463--57474},
doi = {10.1109/access.2026.3683374}
}
EUVIMED