Publication detail
Time-Frequency Analysis of Accelerometery Data for Detection and Identification of Faults on a Pneumatic Production Machine
DOBOSSY, B. ADÁMEK, R. RADIL, F. BRABLC, M.
English title
Time-Frequency Analysis of Accelerometery Data for Detection and Identification of Faults on a Pneumatic Production Machine
Type
conference paper
Language
en
Original abstract
As pneumatic cylinders are nowadays an integral part of manufacturing machines in a variety of industries, there is a growing demand for development of cost-effective and reliable techniques for detection and identification of faults on these machines. In this paper, we propose a novel algorithm for condition indicator extraction from the time-frequency domain. The algorithm uses signal-based condition indicators extracted from a frequency spectrum obtained by a modified version of the Short-Time Fourier Transform (STFT). The aim of this article is to demonstrate that the processing of readings from an accelerometer by the proposed method provides higher accuracy in detecting and identifying faults. We achieve all this by processing only a fraction of the measured signal. As result of that, the computational requirements are decreasing, leading to lower hardware cost which is advantageous as the proposed method is developed for industrial deployment.
English abstract
As pneumatic cylinders are nowadays an integral part of manufacturing machines in a variety of industries, there is a growing demand for development of cost-effective and reliable techniques for detection and identification of faults on these machines. In this paper, we propose a novel algorithm for condition indicator extraction from the time-frequency domain. The algorithm uses signal-based condition indicators extracted from a frequency spectrum obtained by a modified version of the Short-Time Fourier Transform (STFT). The aim of this article is to demonstrate that the processing of readings from an accelerometer by the proposed method provides higher accuracy in detecting and identifying faults. We achieve all this by processing only a fraction of the measured signal. As result of that, the computational requirements are decreasing, leading to lower hardware cost which is advantageous as the proposed method is developed for industrial deployment.
Keywords in English
fault diagnosis; pneumatic system; Short-Time Fourier Transform; signal-based features
Released
07.12.2022
Publisher
Institute of Electrical and Electronics Engineers
Location
Plzeň
ISBN
978-1-6654-1040-3
Book
Proceedings of the 2022 20th International Conference on Mechatronics - Mechatronika (ME)
Pages count
6
BIBTEX
@inproceedings{BUT182779,
author="Barnabás {Dobossy} and Roman {Adámek} and Filip {Radil} and Martin {Brablc},
title="Time-Frequency Analysis of Accelerometery Data for Detection and Identification of Faults on a Pneumatic Production Machine",
booktitle="Proceedings of the 2022 20th International Conference on Mechatronics - Mechatronika (ME)",
year="2022",
month="December",
publisher="Institute of Electrical and Electronics Engineers",
address="Plzeň",
isbn="978-1-6654-1040-3"
}