Community Research and Development Information Service - CORDIS

Abstract

The application of several time-frequency signal processing algorithms, ie, spectrogram, smoothed Wigner-Ville distribution and scalogram, in processing vibration signatures to detect the link-joint backlash of a robot is examined. A systematic feature extraction approach is proposed to process the time-frequency representations of signals, so that the backlash of a robot link-joint can be detected, interpreted and also quantified. Practical aspects of the calculation of these algorithms are discussed briefly, and additionally the differences between these time-frequency signal representations are pointed out to address the characteristics of the three algorithms. Based on the physical phenomena, several features, ie local energy Omega, first-order local moment with respect to frequency, local maximum, the duration of certain effects in the time-frequency plane and the Itakura spectral distance d(t) extended for time-frequency representation and calculated for every oscillating cycle. As a result, the features Omega and especially Omega(f) and d(t) are quite promising with respect to quantifying backlash.

Additional information

Authors: PAN M C, Katholieke Universiteit Leuven, Department of Mechanical Engineering (BE);VERBEURE B, Katholieke Universiteit Leuven, Department of Mechanical Engineering (BE);VAN BRUSSEL H, Katholieke Universiteit Leuven, Department of Mechanical Engineering (BE);SAS P, Katholieke Universiteit Leuven, Department of Mechanical Engineering (BE)
Bibliographic Reference: Article: ISMA19 - Tools for Noise and Vibration Analysis, pp. 407-421
Record Number: 199610257 / Last updated on: 1996-03-18
Category: PUBLICATION
Original language: en
Available languages: en
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top