저널베어링 이상상태 진단시스템의 데이텀 정의

정준하, 전병철, 윤병동 et al. · 대한기계학회 춘추학술대회 · 2014

Data-driven diagnosis systems for journal bearings frequently use vibration data acquired from sensors. However, a small disruption in the system leads to perturbation in the rotating speed, which results in unstable vibration signal. In this research, a datum unit for feature extraction is proposed to embrace a sudden change in vibration data. It presents the angular resampling method, common for ball bearing systems but not for journal bearing systems, and definition of datum for extracting features. The proposed approach enables the raw vibration data to be processed consistently despite the variations in signals, and defines effective cycles for extr acting features. The data is acquired from analytic signal from a calculation program as well as experiment data from journal bearing test-bed. Overall, the increased separability measure values shows that proposed idea is effective and the improved classification result adds reliability in diagnosis system for journal bearings.

Cited by 0 works.

Authors