The Research On The Denoising Method Of Interferences For On-Line Partial Discharge Monitoring Of Transformer

As the key equipment in electric power system ,transformer plays a very important role in economic construction. Statistics show that the main transformer faults are insulation, which is the state of insulation as an important index in the monitoring of transformer partial discharge detection. In the partial discharge site there are a lot of large interfering signals, which affect the monitoring system sensitivity and reliability of detection severely. Therefore, it is necessary and important to do researches on partial discharge technology of noise reduction accurately and reliably.The thesis researches on the concept of transformer partial discharge, the disturbance characteristics, and the current detection and denoising technology. On the basis of study on the wavelet analysis and mathematical morphological method, the thesis does the main work as followings:Firstly, In-depth study of the wavelet analysis theory, the wavelet time-frequency filtering is deeply researched. The paper focus on the wavelet packet algorithm for wavelet packet algorithm. As a result of its complexity, the effectiveness of low-defect, an improved model was designed based on the wavelet packet filtering in the transformer partial discharge denoising method. Mainly the algorithm's accuracy and efficiency has been improved by optimizing the parameters of wavelet packet model, by sampling decomposition, prediction algorithm, and updating operator to optimize the ordinary wavelet packet algorithm.Compared with the traditional wavelet algorithm using wavelet packet algorithm, the improved partial discharge signal denoising algorithm not only improve the efficiency and the effect of noise removal is obvious.Secondly, the paper studies the basic theory of mathematical morphology, which will be applied to the technology of transformer partial discharge denoising. It is proposed and a generalized filtering algorithm based on mathematical morphology, and generalizes reasonable structural elements of morphological filter. Then the generalized morphological filter is applied on the technology of transformer partial discharge signal denoising. Compared with the traditional wavelet algorithm, the significance of morphological algorithm is clear in geometric meaning, and its advantages of high operational algorithm. The model not only retains the information characteristics of partial discharge signals, but also be very effective suppression of interferenceThirdly, the improved wavelet packet algorithm and mathematical morphology algorithms are applied into the collected on-site partial discharge signals processes. The results show that the improved wavelet packet model and the generalized morphological filter model for denoising partial discharge in transformer technology, not only effectively improve the signal to noise ratio, and remove the interference filter works well, but also retains the basic characteristics of partial discharge signals. Wavelet packet filtering algorithm and mathematical morphology filtering algorithm improve the superiority, which have good prospects in the denoising noise in the transformer partial discharge technology.