Classification of unbalance and misalignment faults in rotor using multi-axis time domain features

Early and accurate detection of rotor faults is crucial for optimal performance of rotating machinery. Unbalance and misalignment are the most common faults occurring in the machinery. Using vibration-based conventional frequency analysis methods, it is often difficult to identify these faults because they exhibit similar frequency patterns. The balancing procedure of an unbalanced rotor is based on attachment or removal of certain amount of weight to or from a particular location of the rotor. The rotor may causes additional problems in machinery, if such treatment is applied to correct misalignment faults. Therefore, accurate diagnosis of these faults is extremely important prior to corrective action. This paper utilizes radial and axial vibrations for the purpose. Sensitivity of statistical time domain features, extracted from these multi-axis vibration signals, is investigated. Every pair of alike features is then further processed to maintain the length of feature vector for efficient data processing. Support vector machine (SVM) is used to determine the effectiveness of proposed method, and 100% accuracy is obtained for the problem at hand.