基于量子神经网络和证据融合的小电流接地选线方法  

A Fault Line Detection Method for Indirectly Grounding Power System Based on Quantum Neural Network and Evidence Fusion

作　　者：张海平[1] 何正友[1] 张钧[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031

出　　处：《电工技术学报》2009年第12期171-178,共8页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金(50407009);四川省杰出青年基金(06ZQ026-012);教育部优秀新世纪人才支持计划(NCET-06-0799)资助项目

摘　　要：在研究基于集成量子神经网络和Dempster-Shafer(DS)证据理论故障诊断模型的基础上,提出将该模型应用到小电流接地选线中。利用快速傅里叶变换和小波包变换从零序电流信号中提取故障特征量来训练多个量子神经网络,再用DS证据理论对各个神经网络的输出结果进行全局融合,得到综合选线结果。仿真结果表明该模型对小电流接地选线具有很强的适应性,且不受系统接地方式、合闸角、过渡电阻等因素的影响,解决了单一判据选线准确率低和高维输入神经网络训练收敛速度慢、诊断时间长等问题。A new fault diagnosis model based on intergrated quantum neural networks （QNNs） and dempster-shafer （DS） evidence theory to detect the fault line for indirectly grounding power system is proposed. According to fast fourier transform （FFT） and wavelet packet transform （WPT） algorithms, the fault features extracted from zero sequence current are used to train the quantum neural networks, then DS evidence theory is used for global diagnosis to gain a unified line selection result from the outputs of the networks. The simulation results indicate that the model has strong adaptability to the fault line detection for indirectly grounding system, and the process is not sensitive to earth mode, inception angles and transition resistance. The issues are solved, which are low accuracy of the detecting process with single criterion, slow convergence speed and long diagnosis time of the high dimension inputs neural netwok.

关 键 词：量子神经网络 DS证据理论 信息融合 小电流接地选线 

分 类 号：TM77[电气工程—电力系统及自动化]