Créé le : 11/02/2019
This paper presents experimental results obtained for the detection and identification of valves defects on air compressors. To this end, the cyclostationary approach is used after angular resampling of vibration signals collected on the machine. Moreover, the angle-frequency representation of these signals using the Wigner-Ville transformation allows defining specific monitoring indicators closely related to machine cyclic phenomena. The evolution of these parameters when different valve defects occur shows the effectiveness of the cyclostationary approach to detect and identify the various failure modes of the equipment. Finally, the best vibration sensor locations on the equipment are identified and an optimal monitoring system is proposed allowing reducing the overall cost of the condition monitoring system.