INTELLIGENT ACOUSTIC LOCALIZATION SYSTEM BASED ON HYBRID NEURAL NETWORKS WITH ADAPTIVE FEATURE WEIGHTING
Measuring Equipment and Metrology · 2026 · 87 (2) · 116-122
DOI: 10.23939/istcmtm2026.02.116 ↗Kurz gefasst
Fachartikel in der Zeitschrift Measuring Equipment and Metrology, 2026. Verfasst von Sviatoslav Sidletskyi und 1 weiteren Autor:innen. Die Arbeit nennt 14 Literaturangaben.
Abstract (Originalsprache)
The intelligent acoustic localization system based on a hybrid neural network with adaptive feature weighting is designed to improve the accuracy of sound source coordinate determination under conditions of noise and uncertainty. The proposed model combines a multilayer perceptron and an attention mechanism, which ensures adaptive extraction of the most informative features and reduction of noise components in the signal. The feature space is formed based on TDoA and environmental parameters, including temperature, humidity, and wind speed. A synthetically generated acoustic environment with added Gaussian noise was used for training. A comparison of the classical MLP model and the proposed hybrid architecture was conducted. Experimental results showed a 29.1% reduction in loss function, 30.2% reduction in MSE, 11.2% reduction in median error, and a 28.2% improvement in noise robustness. The results confirm the effectiveness of attention mechanisms in acoustic localization tasks and the promising applicability of the proposed model in real-time intelligent systems
Zitieren
APA
Sidletskyi, S., & Seheda, O. (2026). INTELLIGENT ACOUSTIC LOCALIZATION SYSTEM BASED ON HYBRID NEURAL NETWORKS WITH ADAPTIVE FEATURE WEIGHTING. Measuring Equipment and Metrology, 87(2), 116-122. https://doi.org/10.23939/istcmtm2026.02.116
Vancouver
Sidletskyi S, Seheda O. INTELLIGENT ACOUSTIC LOCALIZATION SYSTEM BASED ON HYBRID NEURAL NETWORKS WITH ADAPTIVE FEATURE WEIGHTING. Measuring Equipment and Metrology. 2026;87(2):116-122. doi:10.23939/istcmtm2026.02.116
BibTeX
@article{Sidletskyi2026,
author = {Sidletskyi, Sviatoslav and Seheda, Oleh},
title = {INTELLIGENT ACOUSTIC LOCALIZATION SYSTEM BASED ON HYBRID NEURAL NETWORKS WITH ADAPTIVE FEATURE WEIGHTING},
journal = {Measuring Equipment and Metrology},
year = {2026},
volume = {87},
number = {2},
pages = {116--122},
doi = {10.23939/istcmtm2026.02.116}
}
EUVIMED