摘要
针对传统选线方法在消弧线圈过补偿工况下存在的特征混淆、拓扑适应性差等问题,提出了一种数据驱动的智能选线方法,通过构建零序等效电路模型,阐明了补偿电流与容性电流的动态平衡机制,论证了过补偿策略对故障零序电流幅相特性的影响规律;提出了采用皮尔逊相关系数量化线路间零序电流波形相似度的故障选线新方法,引入线路层级编码规则表征电网拓扑关联性,通过改进的K-means聚类算法实现特征空间的自适应划分,有效解决了噪声干扰与补偿波动导致的特征混淆问题;开发了基于PSCAD/EMTDC平台的仿真验证体系,验证了方法的工程适用性。研究成果可通过新型矿用隔爆感知装置实现技术转化,为智能煤矿建设提供了理论支撑与装备保障。
A data-driven intelligent line selection method is proposed to address the problems of feature confusion and poor topology adaptability in traditional line selection methods under overcompensation conditions of arc suppression coils.By constructing a zero sequence equivalent circuit model,the dynamic balance mechanism between compensation current and capacitive current is elucidated,and the influence of overcompensation strategy on the amplitude and phase characteristics of fault zero sequence current is demonstrated.A new method for fault line selection using Pearson correlation coefficient to quantify the similarity of zero sequence current waveforms between lines is proposed.The line hierarchical coding rule is introduced to characterize the topological correlation of the power grid.The improved K-means clustering algorithm is used to achieve adaptive partitioning of the feature space,effectively solving the problem of feature confusion caused by noise interference and compensation fluctuations.A simulation verification system based on the PSCAD/EMTDC platform is developed to verify the engineering applicability of the methodology.The research results can be transformed into technology through a new type of mining explosion-proof sensing device,providing theoretical support and equipment support for the construction of intelligent coal mines.
作者
张晋铭
胡玉东
王文涛
梁睿
卢仕祺
唐家保
金智俊
ZHANG Jinming;HU Yudong;WANG Wentao;LIANG Rui;LU Shiqi;TANG Jiabao;JIN Zhijun(School of Electrical Engineering,China University of Mining and Technology,Xuzhou,Jiangsu 221116,China;Shandong Ameksi Electric Co.,Ltd.,Heze,Shandong 274000,China)
出处
《自动化应用》
2025年第15期107-111,114,共6页
Automation Application
基金
山东省科技型中小企业创新能力提升工程项目“全屏蔽、全绝缘、全接地智能矿用隔爆兼本质安全型组合配电装置研发与应用”(2023TSGC0072)
中国矿业大学研究生创新计划项目资助“多维特征深度学习的高比例新能源配电网安全状态评估”(2025WLJCRCZL359)。
关键词
煤矿配电网
单相接地故障
K-MEANS聚类
故障选线
coal mine distribution network
single-phase grounding fault
K-means clustering
fault line selection
