煤矿配电网电压暂降源辨识模型

针对煤矿配电网电压暂降信号特征提取困难和辨识准确率低的问题,应用小波熵结合支持向量机SVM(support vector machine)建立故障辨识模型,以故障信号的小波熵测度来表征故障特征,对电压暂降信号进行小波多分辨分析,选取采样序列的高频系数分量,计算其小波系数熵和小波时间熵,作为特征向量输入SVM,使故障信号特征更加明显,对故障源进行自动分类辨识。结果表明,与小波结合BP神经网络方法比较,无论在训练时间上还是在辨识准确率方面均有明显优势。

Voltage Sag Source Identification Model for Distribution Network of Coal Mine

The features of voltage sag signals in the distribution network of coal mine are difficult to extract, and the identification accuracy is also lower. To solve these problems, a fault identification model was established by combining wavelet entropy with support vector machine(SVM). The wavelet entropy measure of fault signal was used to cha-racterize the fault characteristics, and wavelet multi-resolution analysis was performed on the voltage sag signals. The high-frequency coefficients of the sampling sequence were selected to calculate the corresponding wavelet coefficient entropy and wavelet time entropy, which were further input into SVM as characteristic vector. In this way, the feature of the fault signal was more obvious, and the fault source can be classified and identified automatically. The proposed model was compared with an algorithm that combined wavelet with BP neural network, showing obvious advantages in both training time and identification accuracy.