Vibration analysis metrics of a ball bearing during different operational states

This labeled dataset is provided in the form of a CSV file and contains vibration analysis metrics (v-RMS, a-RMS, a-Peak, Temperature, Crest Factor) of a 6204 2RS ball bearing by the use of an IFM VVB001 vibration sensor and measurements were captured every two minutes by the sensor device.
An experimental assembly was set up comprised by a 0.75kW - 1450rpm Bonfiglioli BN80B4 FD motor with an i=80 reduction gear, a coupler and an axle with the ball bearing mounted on the latter for the purpose of conducting experiments based on three operational states of the bearing in regard to the level of grease applied.

Attribute description as per the sensor’s manual and operation instructions:

• The v-RMS (effective value of the vibration velocity) measures the total load of a rotating machine. The most frequent types of overload (unbalance, alignment errors, etc.) are reflected in the v-RMS. An increased load can damage the machine in the long term (fatigue, fatigue strength) or, in extreme cases, destroy it within a short time. It is expressed in m/s.
• The a-RMS (effective value of the acceleration) detects mechanical contact of machine components. This contact typically occurs in case of wear (faulty bearing, worn out toothed wheels, etc.) or problems with lubricants (contaminated grease, water in oil, etc.). It is expressed in m/s².
• The a-Peak monitors the maximum value of the acceleration. Shocks in the acceleration can occur once or periodically, as in a crash, for example in the event of bearing damage. a-Peak is a measure for the forces occurring on the machine. It is expressed in m/s².
• The Crest (or crest factor) is a described characteristic value of the signal analysis. It is defined as the ratio of the maximum value to the effective value (peak/RMS). In condition monitoring the characteristic value is used for the evaluation of the bearing condition. The high-frequency signals with a short pulse duration of a bearing damage generate higher peak values in relation to the effective value. This relation can be read from the crest factor.
• The Temperature attribute is self explanatory and is expressed in degrees of Celsius.
• Labels of the Bearing State regarding the recorded measurements:
  1 - sealed ball bearing with recommended amount of industrial grease
  2 - unsealed ball bearing with no amount of grease
  3 - unsealed ball bearing with excess amount of grease

This dataset was produced for the purpose of training an Artificial Neural Network model for the task of multi-class classification in the frames of a Predictive Maintenance strategy.