采用荧光光纤检测局部放电的自适应阈值方法

Adaptive Threshold Method for Partial Discharge Detection by Fluorescent Fiber

针对光测法实际检测中光电器件噪声及阈值设定困难的问题,提出了一种自适应的阈值方法。根据放电信号与噪声信号的相位分布差异,对采样数据进行分组量化与多周期累加。在充分分析累加序列功率成分的基础上,采用自回归(AR)谱估计理论,计算累加序列的功率谱并提取信号功率、噪声功率及谐波基波比三项功率谱指标。基于三项功率谱指标定义了一种放电评价参数,该参数能够表征信号功率占放电总功率的加权比值。通过噪声功率的波动幅度完成自适应阈值的设定。实验结果表明,所提方法可以根据采样数据动态调整阈值,实现放电的自动检测,无需人工干预,检测灵敏度高。该研究对于电力设备故障初期的预警及长期在线监测具有重要的意义。

Aiming at the noise of photoelectric device and difficulty of threshold setting,an adaptive threshold method is proposed.According to the distribution differences of partial discharge（PD）signals and noises,the sample data are divided into groups and accumulated by multi-cycles in time domain.Sufficiently investigating the power features,the AR spectrum estimation is applied to calculate the power spectrum of the sequence,and three power indexes,i.e.,total power,signal power and harmonic ratio,are extracted.Analyzing the indexes,a parameter is defined which can characterize the weighted ratio of signal power to total power,and the adaptive threshold is set in terms of the fluctuation of noise power.Experiments show that the proposed method is able to adapt the threshold according to sample data and realize autodetection of PD with high sensitivity and without human intervention,especially for early warning and online monitor of electrical equipment.