転がり軸受の故障促進制御

抄録

Diagnosis of the condition of rolling bearings is important for maintaining the machine operating rate. In recent years, research on bearing condition diagnosis methods using machine learning has been actively conducted. In these diagnoses, the failure state data obtained from bearings that have been given initial defects such as indentations in advance are often used. However, in actual condition diagnosis of bearing, it is important to detect abnormalities at the initial stage, so it is necessary to use bearing data from the normal condition to the occurrence of damage in machine learning. In this study, an experimental system that can apply an arbitral motion with any frequency to the rotating shaft was developed. A vibration experiment of deep groove ball bearings was conducted. The 3 axes (x,y,z) acceleration, 2 axes (x,y) displacement, and acoustic emission signal were measured and analyzed. Measures of envelop spectrum and kurtosis were calculated and observed.