http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
주파수 대역 최적화 기반 노이즈에 강건한 볼 베어링 고장 진단 알고리즘 개발
최종현(Jonghyun Choi),박종민(Jongmin Park),박정호(Jungho Park),김수지(SuJ. Kim),정화용(Hwayong Jung),윤병동(Byeng D.Youn) 대한기계학회 2019 대한기계학회 춘추학술대회 Vol.2019 No.11
Fault signal of Rolling Element bearing modulate to certain frequency region due to rotation of bearing elements. To extract the fault signal for bearing diagnostic, demodulation to proper frequency band is necessary. Thus, many studies have been conducted to set quantification indicator for finding most informative frequency band among a full band spectrum. However, previous studies, such as Kurtogram and Autogram, failed to extract the fault informative band in high noise condition as they used indirect indicator which can be highlighted by impulsive and periodic noise. Also binary tree based band selection method which was not adaptive to bearing signal. To addreess these problems, this paper suggest new informative band indicator which is Fault to mean ratio(FMR) and new band selection method which based on particle swarm optimization(PSO). The FMR is direct measure of fault information in certain frequency band as it represent fault power density compare to other components. The compatibility of proposed method was checked under varying fault location and speed condition by Seoul National University(SNU) testbed data and under high noise environment by wheel bearing test data on real road condition. It was shown that proposed method present robustness to high noise environment compare to previously introduced band selection method.
최종현(Jonghyun Choi),박종민(Jonmin Park),김근수(Keunsu Kim),김수지(S.J Kim),정화용(Hwayong Jung),윤병동(Byeng D. Youn) 대한기계학회 2018 대한기계학회 춘추학술대회 Vol.2018 No.12
Envelope analysis is most conventionally used for detection of rolling element bearing faults by bandpass filtering the raw signal and then extracting its envelope and then analyzing fault frequency energy by Fourier transform. However, traditional Envelope analysis based algorithm such as high frequency resonance technique and spectral kurtosis has limitation in extracting faulty signal and constructing optimal bandpass-filter. This paper employs empirical mode decomposition(EMD) to decompose signal and sort fault implemented signal and particle swarm optimization(PSO)to optimize bandpass filter. The performance of “EMD+PSO Envelope analysis” was checked under varying fault location and load condition by CWRU data, and varying operating speed and fault size by SNU data. It was shown that suggested algorithm successfully diagnose four commonly occur bearing fault location and also naturally formed distributed fault under different speed and loading condition. It was also shown that suggested algorithm presents much higher performance compare to three previously introduced Envelope analysis algorithm.