环网柜电缆头局部放电检测与模式识别

Partial Discharge Detection and Pattern Recognition for the Cable Head of Ring Main Uint

电缆接头的局部放电现象是户外环网柜绝缘故障的主要表现形式,本文利用电磁耦合法对电缆头局部放电进行检测,建立环网柜电缆局部放电检测实验平台,应用局放仪测得的局放量作为局放发生判据,运用小波变换理论滤除信号中的噪声干扰,提取有效信号。通过大量的实验分析,选取8MHz-12MHz信号频谱的功率谱面积作为模式识别的特征向量,利用BP神经网络算法实现电缆头的局部放电识别。结果表明,该方法能有效区分局部放电发生与否,具有较高的识别率,为局部放电检测提供了一种新的算法。

Cable head partial discharge is the main fault in the Ring Main Uint, which is detected with the method of electromagnetic coupling in this article.The experimental platform is established for the partial discharge detection and the charges showed in the discharge device is the reference to judge the partial discharge.Wavelet transform is used to filter noises in the signal.Through lots of experiments, the power spectrum area in 8MHz-12MHz is the main feature of pattern recognition as the BP neural network the main algorithm of it.It turns out that BP neural network is useful to discriminate partial discharge of the cable head and with a high identification rate, which provides a new algorithm for partial discharge.