| 引用本文: | 邢超,高敬业,毕贵红,陈仕龙.基于集成神经网络的特高压直流输电线路初始电压行波小波变换模极大值比单端测距方法[J].电力自动化设备,2022,42(11): |
| XING Chao,GAO Jingye,BI Guihong,CHEN Shilong.Single-end fault location method used ratio between WT modulus maximum values of initial voltage traveling wave for UHVDC transmission line based on integrated neural network[J].Electric Power Automation Equipment,2022,42(11): |
|
| 摘要: |
| 针对现有故障测距方法存在对高阻故障不灵敏、二次行波波头难以捕捉的问题，提出一种基于集成神经网络的特高压直流输电线路初始电压行波小波变换模极大值比的单端测距方法。首先，推导出故障距离与初始电压行波的线模量和地模量的小波变换模极大值比之间的近似公式，公式表明两者之间具有非线性关系，且此关系与过渡电阻无关。然后，利用AdaBoost-Elman集成神经网络拟合两者之间的非线性关系，提取不同小波尺度下初始电压行波各模量分量的小波变换模极大值比作为集成神经网络输入量，将故障距离作为输出量，构建集成神经网络故障测距模型。将各小波尺度下的初始电压行波各模量分量的小波变换模极大值比输入训练完成的集成神经网络模型即可达到故障测距的目的。仿真结果表明，所提方法测距精度高，且不受过渡电阻影响。 |
| 关键词: 特高压直流 输电线路 初始电压行波 模极大值比 AdaBoost-Elman集成神经网络 故障测距 |
| DOI：10.16081/j.epae.202205008 |
| 分类号:TM721.1;TM771 |
| 基金项目:国家自然科学基金资助项目(52067009) |
|
| Single-end fault location method used ratio between WT modulus maximum values of initial voltage traveling wave for UHVDC transmission line based on integrated neural network |
|
XING Chao1, GAO Jingye1,2, BI Guihong2, CHEN Shilong2
|
|
1.Electric Power Research Institute of Yunnan Power Grid Co.,Ltd.,Kunming 650217, China;2.College of Electrical Engineering, Kunming University of Science and Technology, Kunming 650500, China
|
| Abstract: |
| Since the existing fault location methods have the problems of insensitivity to high resistance faults and difficulty in identifying the secondary traveling wave head, a single-end fault location method used the ratio between WT(Wavelet Transform) modulus maximum values of initial voltage traveling wave for UHVDC(Ultra High Voltage Direct Current) transmission line based on integrated neural network is proposed. Firstly, the approximate formula for the fault distance and the ratio between WT modulus maximum values of initial voltage traveling wave of aerial-mode and zero-mode components is derived. The formula shows that there is a nonlinear relationship between the WT modulus maximum values of initial voltage traveling wave of aerial-mode and zero-mode components, and this relationship is independent of the transition resistance. Then, the AdaBoost-Elman integrated neural network is used to fit the non-linear relationship. Taking the ratio between WT modulus maximum values of the modal components of initial voltage traveling wave at different wavelet scales as the input of the integrated neural network and the fault distance as the output, a fault location model based on integrated neural network is built. The fault distance can be obtained by inputting the ratio between WT modulus maximum values of initial voltage traveling wave of modal components under each wavelet scale into the trained model. Simulative results verify that the proposed method has high fault location precision and is not affected by transition resistance. |
| Key words: UHVDC transmission line initial voltage traveling wave ratio between modulus maximum values AdaBoost-Elman integrated neural network electric fault location |
