A Study on Evaluation of Rolling Element Bearing Condition using Pattern Recognition Method

vibration signals was developed. The pattern recognition method was applied in this technique. The feature of the technique is that various kinds of obtained measurement values were spectrum conversion for every measurement value on multidimensional vector space that is expressed as a Mahalanobis Distance corresponding to a state. This technique was applied to diagnosis of rolling element bearing condition, and was verified by test fixture in the course of progressing degradation of rolling element bearing condition, from the normal state to abnormal state as running short of oil or seizure.