逆止弁診断へのAI適用の検討
概要
As an attempt to apply AI in the field of maintenance, we have been developing a diagnosis method of check valves to detect the internal leakage and the signal before the leakage based on the signal analysis of their vibration measurements by a portable system with AE (Acoustic Emission) sensors. CNN (Convolutional Neural Network) was applied to the measured raw waveform and its frequency distribution corresponding to the leakage rates in order to classify the leakage rates. The leakage rate was gradually increased from the zero-leakage state using an experimental loop set-up. We also examined the effect of ambient noise on the diagnosis.
