基于GA优化BP神经网络的有源配电网高阻接地故障选线方法

当配电网发生高阻接地故障时，逆变型分布式电源的接入会向零序网络中注入不平衡的谐波电流，改变原有故障特征的分布规律，导致传统高阻故障选线方法失效。考虑光伏电源接入对配电网的影响，提出了一种基于GA优化BP神经网络通过融合多种故障特征的有源配电网高阻接地故障选线方法。首先，利用Matlab/Simulink搭建谐振接地系统仿真得到选定周波的故障零序电流，根据小波包变换从中提取小波包能量熵和模极大值，并将其作为数据样本。然后，将数据输入优化后的网络中进行训练，得到能够实现智能选线的机器学习模型。最后，算例分析表明该方法较传统算法提高了迭代速度和训练精度，在多种复杂故障条件下具有良好的选线容错率，且具有一定的抗噪能力。

High Resistance Grounding Fault Line Selection Method for Active Distribution Network Based on GA Optimized BP Neural Network

When the high resistance grounding fault occurs in the distribution network， the access of inverter interfaced distributed generation will inject unbalanced harmonic current into the zero-sequence network， change the distribution law of original fault characteristics， and lead to the failure of traditional high resistance fault line selection method. Considering the influence of photovoltaic power access to distribution network，GA-optimized BP neural network is proposed based on high resistance grounding fault line selection method of active distribution network by integrating various fault characteristics. Firstly，the resonant grounding system is built by Matlab/Simulink to simulate the fault zero sequence current of selected frequency wave， and the wavelet packet energy entropy and mode maximum are extracted from the system according to the wavelet packet transform and taken as data samples. Then， the data is inputted into the optimized network for training，and the machine learning model that can realize intelligent route selection is obtained. Finally，an example analysis shows that the proposed method can improve the iteration speed and training accuracy compared with the traditional algorithm，has good fault tolerance rate of line selection and certain anti-noise ability under a variety of complex fault conditions.