[Truncated] Power transformers are one of the key devices in the power industry. The reliability, quality, and economic cost of electrical power are affected by a transformer’s health conditions. Most catastrophic failures of power transformers are caused by winding problems, which may lead to huge economic loss and serious environmental impact. Therefore, the faults of power transformers, especially winding faults, are of much concern and have been investigated extensively.
This thesis focuses on a vibration-based condition monitoring method for power transformers. The advantages of this method lay in its on-line, non-invasive monitoring ability and real-time failure diagnosis. It uses vibration signals measured on the transformer tank to evaluate the health condition of the monitored transformer and to detect its potential failures. Successful evaluation and detection requires a clear understanding of the vibration characteristics of power transformers, in particular, the vibration features of different winding failure modes. The objective of this project is to provide some related experimental and theoretical information for improving existing vibration-based monitoring systems for power transformers.
An integrated monitoring device, TranstethoscopeTM, is developed for use in the power industry. It has been installed on in-service power transformers for monitoring purposes. By analyzing monitoring data recorded from two different types of power transformers, it is found that the on-line vibration of transformers is highly dependent on the operating signals (input voltage and loading current). Therefore, a monitoring model for power transformers is established based on the System Identification Method. This model uses the transformer operating signals as the inputs and the on-line vibration of the transformer as the output to estimate the difference between the measured vibration and the predicted model output. This error is an indicator for evaluating the health condition of the transformer. From the practical application of this monitoring model to the two in-service power transformers, it is found that this model is more suitable for a power transformer with a relatively stable loading current, but its function is very limited on a transformer that bears frequent heavy loading impacts. Another shortcoming is that this model does not have a clear physical meaning and requires a large database for failure diagnosis.
|Qualification||Doctor of Philosophy|
|Publication status||Unpublished - Jun 2015|