합성곱 신경망을 사용한 회전체 오일-휩 및 오일-휩 초기 상태 상태 진단 연구

초록

Failure of a rotating machine can lead to not only loss of system performance but massive losses. Therefore, Condition monitoring technologies of rotating machine for detecting a failure has been actively applied in many industries. Recently, as many techniques were developed to collect and analysis of data, condition monitoring which machine learning technology is applied has been studied. In this paper, the machine learning method is used to detect oil whip phenomenon and classify the normal state, oil whip initial state and oil whip. Oil whip which cand lead to large amplitude of vibration is the most common fault cause of sub-synchronous instability in hydrodynamic journal bearings. Convolution neural network which is widely used in image dimensionality reduction is used to detect oil whip phenomenon. As the input data of the neural network, an orbital image which can represent the feature of the normal state, the oil-whip initial state, and the oil-whip is used.

키워드