摘要
针对高压直流输电线路短路故障识别精度不高的问题,提出一种基于改进的北方苍鹰算法优化核极限学习机(INGO-KELM)的高压直流输电线路短路故障识别方法。为了解决虚假模态问题,采用改进的自适应噪声完备集合经验模态分解(ICEEMDAN)方法,对预处理后的故障电流信号进行了分解,从而获取了不同频率的固有模态函数分量(IMF);计算相关系数最高的2组IMF分量的样本熵值作为故障特征量,并组建特征数据集;采用Chebyshev混沌映射、混合正弦余弦算法以及Levy飞行策略,对NGO算法进行多方面的策略优化,将数据集输入到INGO优化后的KELM短路故障识别模型中实现故障识别。仿真结果表明,所提方法具备良好的故障识别准确度和鲁棒性,在不同过渡电阻和故障距离的情形下,均可以准确识别各种短路故障类型,识别精度达到98%以上。
Aiming at the problem of low accuracy of short-circuit fault identification in HVDC transmission lines,a short-circuit fault identification method for HVDC transmission lines based on improved northern goshawk optimization-kernel extreme learning machine(INGO-KELM)is proposed.In order to solve the spurious mode problem,the preprocessed fault current signal undergoes decomposition through the ICEEMDAN,aiming to derive the intrinsic modal function components(IMF)of each frequency band.The sample entropy values of the two groups IMF components with the highest correlation coefficients are used as the fault feature quantities and the feature dataset is formed.Chebyshev chaotic mapping,hybrid sinusoidal cosine algorithm,and Levy flight strategy are employed to optimize the NGO algorithm with multiple strategies.The dataset is inputted into the INGO-KELM short-circuit fault identification model to achieve fault identification.Simulation results show that the proposed method possesses good fault identification accuracy and robustness.The method can still accurately recognize various short-circuit fault types under different transition resistances and fault distances,and the recognition accuracy reaches 98%.
作者
赵岩
王梓毅
徐天
ZHAO Yan;WANG Ziyi;XU Tian(School of Electrical and Control Engineering,Heilongjiang University of Science and Technology,Harbin 150022,China)
出处
《黑龙江电力》
2025年第1期34-43,共10页
Heilongjiang Electric Power
基金
黑龙江省省属高等学校基本科研业务费项目(项目编号:2021-KYYWF-1476)。
