The life expectancy of the rotating unit used in various places in the vehicle is difficult to observe and predict the state of life due to the change of the vibration pattern according to the cause of various defects. The pattern of vibration depends on the degree of degradation of the bearing, and even if you try to record and judge it, the capacity of the measured data increases rapidly with time, making it difficult to store and utilize it. Therefore, various methods for predicting the lifespan of a rotating unit, especially a bearing, have been studied, and it is necessary to accurately grasp the state information of the rotating unit. In this paper, a signal processing and clustering system for vibration information proposed by using an experimental data set with an attached radial sensor attached to the bearing test rig to understand the state information of the rotating unit. The clustered state information was proposed to enable faster and more efficient analysis of signal processed vibration information. This method shows that it is possible to classify the same fault information that occurred in different test sets, so it is expandable that can be classified even if new vibration information of a rotating unit is input. Also, it is an economical classification method because it does not need to deal with a large amount of accumulated data when building a monitoring system with a relatively simple clustering method.