基于Hilbert-Huang变换和神经网络的带串补高压输电线路故障测距

Fault Location for High Voltage Transmission Lines With the Series Compensation Capacitor Based on Hilbert-Huang Transform and Neural Network

提出了一种针对带串补高压输电线路的分层结构神经网络模型故障测距新算法。第1层为粗略判断故障位置的神经网络模型,利用一种新的信号处理方法Hilbert-Huang变换获取能量故障特征作为第1层神经网络的输入,判断故障发生在电容前或后;第2层为精确确定故障位置的神经网络模型,通过对神经网络的离线训练和对单端故障测距结果的在线补偿,最终得到精确的故障距离。该算法考虑了过渡电阻和分布电容的影响,克服了传统故障测距算法由于忽略分布电容导致在高阻接地故障时故障定位不准确的缺点。

A new fault location algorithm based on artificial neural network （ANN） model with layered structure for high voltage transmission lines with series compensation capacitors is proposed. The first layer is a neural network model that roughly judge fault position where the Hilbert-Huang transform is used for signal processing to obtain energy fault feature that is taken as the input of the first layer of ANN to recognize whether the fault occurred ahead or behind the compensator; the accurate fault location is carried out by the second layer of ANN neural network, by means of the off-line training of ANN and on-line compensation for the single-end fault location result the accurate fault position is implemented. In the proposed algorithm the influences of transition resistance and distributed capacity are taken into account, therefore the defects of traditional fault location algorithms, i.e., the unallowable location error of high-resistance earth-fault and the inaccurate fault location due to the neglect of distributed capacitance.