Study Of On-line Monitoring System And Noise Elimination Based On Wavelet For Partial Discharge Within Power Transformers

The electrical insulation is important to the reliability of power transformers. As statistics shows, the failure of insulation is the prime reason for power transformers. Many researchers find that Partial Discharge (PD) is one of symptom of the insulation aging, and also the PD is the critical case of the insulation damage. The on-line PD measurements have been used as effective method for a long time to measure insulation deterioration of power transformers. After getting the on-line PD pulse, typical PD can be identified. The trending results of these statistic characteristics also be used in following the track of insulation deterioration, it allows maintenance personnel to become aware of developing insulation problems in power transformers, and take corrective action before failure occurs. However, PD measurements on operating rotating machines can be influenced by electrical noise, even more, the noises in the time will larger than signals in magnetite. It is difficult to distinguish the signal processing algorithm. How to detect the PD on-line from power noise and disturbance has been the critical aspect for the PD on-line monitoring of power transformers.The wavelet analysis be developed quickly recent years has good localization property both in time and frequency domain. It is very suitable for the processing of singular signal, just like PD signal. The paper presents a wavelet de-noising method, of which the selection of threshold is studied. Furthermore, an automatic shareholding rule and synthesis shareholding processing is presented to select thresholds in order to obtain the optimal de-noising result. With the experiments of noisy PD signals can be de-noised very effectively and steadily and provides the precondition for pattern recognition. A set of on-line monitoring system for partial discharge within power transformer is developed. The presented system can detect the current signals of multiplex pulses and the ultrasonic signals caused by partial discharge within the power transformer. The on-site operation experience shows that the presented system possesses higher detection accuracy and can offer creditable basis for condition-based maintenance of power transformers.The accomplishments above will play a certain role in improving the performance of PD on-line monitoring system and practicability of thetechnique.