Tehnički vjesnik, Vol. 28 No. 5, 2021.
Izvorni znanstveni članak
https://doi.org/10.17559/TV-20200122112551
Segmentation of a Road Vehicle Vibration Signal Using Multiple Comparison Procedures between Paired Samples
László Róbert Hári
orcid.org/0000-0001-5280-7744
; Department of Logistics and Forwarding, University of Győr, Egyetemtér 1, 9026 Győr, Hungary
Sažetak
Segmentation of road vehicle vibration (RVV) signals can occur by the need to analyse or synthesise vibrations obtained in passenger cars or on the stowage of vans, trucks. A general and widely used measure to quantify RVV signals is its description via power spectral density (PSD). From a given PSD a Gaussian signal can be generated in a shaker testing laboratory. However, actual RVVs tend to have a non-Gaussian and nonstationary nature, which can be modelled as a composition of different segments, each with a different length and RMS content. For simulation purposes of nonstationary vibration signals, different approaches have been introduced yet each with its unique signal segmentation approaches. The current paper proposes a signal segmentation method implemented in the time-frequency domain in order to find segments within an RVV signal, where each segment has similarity in-between and is dissimilar to neighbouring segments. For this purpose, multiple comparison tests had been utilised between the Short-time Fourier transforms (STFT) applied on given fractions of an RVV. Different countermeasures had been applied against the Type I. error inflation.
Ključne riječi
multiple comparison test; paired t-test; performance analysis; segmentation; short-time Fourier transform; type I. error rate control; Wilcoxon signed rank test
Hrčak ID:
261336
URI
Datum izdavanja:
15.8.2021.
Posjeta: 1.238 *