Mechanical Fault Diagnosis of On-load Tap Changer of Power Transformer by Clustering Algorithm

On-load tap changer (OLTC) is an important component in power transformer for voltage regulation, and mechanical fault is the main problem during its switch-over process. Vibration signals resulted from the operation of OLTC are closely related to its mechanical condition. To better recognize the incipient mechanical faults of OLTC with high accuracy, two kinds of typical mechanical faults of storage spring and static contact are simulated in an CM type OLTC with the measured vibration signals, which are always chaotic. Then the one-dimension vibration signal time series is reconstructed into high dimension phase space with the calculated delay time and embedding dimension. The improved K-means clustering algorithm is applied to classify the large amount of phase points into different clustering. The curve of weighted distribution information entropy of each clustering is calculated. Calculated results have shown that the defined information entropy curve is capable of identify the different mechanical fault of OLTC.