基于GA优化BP神经网络小电流接地系统故障选线方法

Fault Line Detection in Low Current Grounding System Based on GA Optimized BP Neural Network

为了达到小电流接地系统故障选线多判据融合的目的,以及克服BP神经网络对初始权值阈值敏感的问题,提出遗传算法优化BP神经网络的故障选线方法。利用MATLAB/Simulink搭建小电流接地系统模型,获取零序电流的故障特征分量,输入经遗传算法优化的BP神经网络模型,经过训练即可输出选线结果。数字仿真试验测试结果表明,方法收敛速度快,判断精度高,满足故障选线的快速性与可靠性。

In order to achieve the purpose of multi-criteria fusion for fault line detection of low current grounding system and to overcome the problem that BP neural network was sensitive to the initial weight threshold,a technique for fault line detection using GA optimized BP neural network was proposed.This technique adopted MATLAB/Simulink to build a model for low current grounding system in order to obtain fault signature components of zero-sequence current.Upon input of GA optimized BP neural network and after training it was available to output line detection results.The results of digital simulation test show that the proposed technique has the advantages of high convergence rate and high accuracy for detection that can satisfy the rapidity and reliability in fault line detection.