基于小波包和概率神经网络算法的短路故障识别方法

The Distinguish Method of Short Circuit Fault Based on Wavelet Packet and Probabilistic Neural Network

为快速准确识别电力系统短路故障类型,根据电力系统发生短路故障时零序电流的能量特征,提出一种基于小波包和概率神经网络算法相结合的短路故障识别新方法。首先利用Matlab/Simulink建立仿真模型,对系统进行短路故障仿真。然后利用小波包对零序电流进行采样和分解,提取小波包分解重构系数,对各频段内的能量进行归一化处理,得到能量特征向量。最后对特征向量进行概率神经网络的训练和测试,将预测样本代入训练结果进行验证,其结果证明该方法能够快速准确地识别短路故障。

In order to distinguish short circuit fault of power system quickly and accurately,a method of short circuit fault detection method based on wavelet packed and probabilistic neural network algorithm is presented according to the energy characteristic of zero sequence current.Firstly,simulation model is built using Matlab/Simulink to carry out short circuit fault simulation.Then,the zero sequence current is sampled decomposed by wavelet packet,the decomposition and reconstruction coefficient of wavelet packet is extracted,and the energy in each frequency band is normalized to obtion the energy eigenvector.Lastly the probabilistic neural network is trained and tested for characteristic vector and take the test sample into training result.The results prove that the method will distinguish short circuit fault fast and accurate.