Szczegóły publikacji
Opis bibliograficzny
Multi-step neural state-space identification of an overhead crane from experimental data / Adam Szulecki, Patryk BAŁAZY, Bartłomiej Pajdziński // W: 27th International Carpathian Control Conference (ICCC) [Dokument elektroniczny] : 1-3 June 2026, Szilvásvárad, Hungary : proceedings. — Wersja do Windows. — Dane tekstowe. — Piscataway : IEEE, cop. 2026. — Print on Demand(PoD) ISBN: 979-8-3195-3321-0. — e-ISBN: 979-8-3195-3320-3. — S. 501–505. — Wymagania systemowe: Adobe Reader. — Bibliogr. s. 505, Abstr. — Publikacja dostępna online od: 2026-07-07
Autorzy (3)
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 169794 |
|---|---|
| Data dodania do BaDAP | 2026-10-02 |
| Tekst źródłowy | URL |
| DOI | 10.1109/ICCC71363.2026.11593257 |
| Rok publikacji | 2026 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Wydawca | Institute of Electrical and Electronics Engineers (IEEE) |
Abstract
This paper presents a data-driven identification of a nonlinear overhead crane using a neural state-space model (NSSM) trained directly on experimental input-output data. Measurements of trolley position and load swing angle were collected from a one-dimensional crane under diverse excitation trajectories. To ensure reliable long-horizon simulation, the model is trained using a multi-step prediction strategy rather than conventional one-step loss. The identified model is validated on previously unseen trajectories using open-loop rollout, demonstrating good agreement between measured and simulated responses for both position and swing dynamics. Quantitative evaluation confirms that the proposed physics-informed NSSM captures the essential nonlinear behavior, including actuator limitations and pendulum effects. Specifically, the physics-informed multi-step training reduced the mean rollout RMSE by approximately 58.5% and the maximum error by 46.6% compared to the baseline model. The results indicate that the model can serve as an accurate digital twin suitable for simulation and further control-oriented studies.
