基于混合LSTM深度学习的特高压直流线路模量幅值比故障测距研究  

Fault Location of Modulus Amplitude Ratio of UHVDC Transmission Line Based on Hybrid LSTM Depth Learning

作　　者：魏荣智 陈仕龙[1] 毕贵红[1] 邓小伟 牛元有 姚红涛 WEI Rongzhi;CHEN Shilong;BI Guihong;DENG Xiaowei;NIU Yuanyou;YAO Hongtao(Faculty of Electric Power Engineering,Kunming University of Science and Technology,Kunming 650500,China;Honghe Power Supply Bureau of Yunnan Power Grid Co.,Ltd.,Honghe 661100,China)

机构地区：[1]昆明理工大学电力工程学院,云南昆明650500 [2]云南电网有限责任公司红河供电局,云南红河661100

出　　处：《电力科学与工程》2023年第12期1-10,共10页Electric Power Science and Engineering

基　　金：国家自然科学基金资助项目(52067009)。

摘　　要：对特高压直流(Ultra high voltage direct current,UHVDC)输电系统而言,准确可靠的测距方案可确保故障线路快速恢复、提高供电可靠性。提出一种基于线模分量和零模分量幅值比的特高压直流线路单端故障测距方法。首先推导出故障距离与初始电压行波模量幅值比之间的近似公式。公式表明,两者之间存在非线性关系,且与过渡电阻无关。然后选用混合长短期记忆(Long short-time memory,LSTM)深度学习网络进行训练和学习,提取各模量初始电压行波首波头的幅值比作为深度学习网络的输入量,以故障距离作为输出量,构建深度学习故障测距模型。搭建云广±800 kV特高压直流输电系统仿真模型。仿真实验表明,所提出的测距方法不受故障类型和过渡电阻的影响,且测距结果具有较高的准确性,相对误差不超过0.34%。For ultra high voltage direct current(UHVDC)transmission system,accurate and reliable location scheme can ensure the fast recovery of fault lines and improve the reliability of power supply.A method for single-terminal fault location of UHVDC transmission lines based on the amplitude ratio of linear module component and zero module component is proposed.Firstly,the approximate formula between fault distance and initial voltage traveling-wave modulus amplitude ratio is derived.The formula shows that there is a nonlinear relationship between the two and it is independent of transition resistance.Then,the hybrid long short-term memory(LSTM)deep learning network is used to train and learn,and the amplitude ratio of the initial voltage traveling wave head of each modulus is extracted as the input of the deep learning network,and the fault distance is taken as the output to build the deep learning fault location model.A simulation model of Yunguang±800 kV UHVDC transmission system is built.The simulation results show that the proposed method is not affected by fault types and transition resistance,and the ranging results have high accuracy,with a relative error less than 0.34%.

关 键 词：特高压输电 直流输电 长短期记忆 模量幅值比 深度学习 故障测距 

分 类 号：TM723[电气工程—电力系统及自动化] TM773