基于模糊概率论的变压器局放信号模式识别法

电力变压器局部放电信号的模式识别是局放研究中的一个难点和重点问题.文中通过对高采样率采集到的4种局放模式下的单个UHF脉冲信号进行小波包分解,从熵值和能量角度提取出6个特征量,运用了模糊理论基本思想,通过简单的计算求出各种放电模式的信息分布表和放电标量作为参考量.当给出一个待确定模式的局放信号时,求出其具体的放电标量,按照相对误差最小原则就可以确定出其模式归属.经实测局放信号验证,该方法思路清晰,简单有效.

Fuzzy Probability Theory Based Pattern Recognition Method of UHF Partial Discharge Signals in Power Transformers

Pattern recognition of partial, discharge is a difficult and key issue in power transformer field. In the paper, a method of PD pattern recognition for power transformer based on fuzzy probability theory is introduced, using six features extracted from entropy and energy of results of wavelet packet transformation. The message distribution table and discharge scalar quantity are therefore calculated as reference values by adopting fuzzy theory. When an uncertain discharge UHF pulse signal is given, the discharge scalar quantity is worked out in accordance with the least error principle and then the PD pattern of this signal is confirmed. In addition, recognition result of signals collected from laboratory indicates that the method proposed is simple, feasible and has great potential for practical application.