Input-Weighted Newton Fading Extended Kalman Filter for Experimental Output Estimation in Control Systems
Amesia · 30.06.2026 · 7 (1) · 25-37
DOI: 10.54559/amesia.1899823 ↗Kurz gefasst
Fachartikel in der Zeitschrift Amesia, online erschienen am 30.06.2026. Verfasst von Levent Özbek. Die Arbeit nennt 23 Literaturangaben.
Abstract (Originalsprache)
This paper proposes an Input-Weighted Newton Fading Extended Kalman Filter (IW-NF-EKF) for real-time output estimation in control systems subject to modeling uncertainty and excitation-dependent disturbances. Unlike conventional adaptive fading Kalman filters that rely exclusively on innovation statistics, the proposed method explicitly incorporates normalized input magnitude into a chi-square-based innovation-consistency condition. The resulting nonlinear fading equation is solved online via Newton iterations, yielding an excitation-aware covariance inflation mechanism. The method is experimentally validated using real measurement data obtained from a computer-aided temperature control system. Comparative results against a fixed Kalman filter and a Newton fading EKF without input weighting demonstrate that the proposed approach achieves lower estimation error and improved goodness-of-fit while preserving stable residual behavior. The results indicate that input-aware fading provides a practically effective and statistically coherent extension of innovation-driven adaptive filtering.
Zitieren
APA
Özbek, L. (2026). Input-Weighted Newton Fading Extended Kalman Filter for Experimental Output Estimation in Control Systems. Amesia, 7(1), 25-37. https://doi.org/10.54559/amesia.1899823
Vancouver
Özbek L. Input-Weighted Newton Fading Extended Kalman Filter for Experimental Output Estimation in Control Systems. Amesia. 2026;7(1):25-37. doi:10.54559/amesia.1899823
BibTeX
@article{zbek2026,
author = {Özbek, Levent},
title = {Input-Weighted Newton Fading Extended Kalman Filter for Experimental Output Estimation in Control Systems},
journal = {Amesia},
year = {2026},
volume = {7},
number = {1},
pages = {25--37},
doi = {10.54559/amesia.1899823}
}
EUVIMED