摘要
针对配电网中线路短路故障、大型感应电动机启动以及变压器投切等单一电压暂降源和由单一电压暂降源组合而成的复合电压暂降源导致的电压暂降现象,在MATLAB/Simulink中搭建改进的IEEE33节点配电网系统模型进行仿真分析和验证,提出一种先对电压暂降信号进行经验模态分解(empirical mode decomposition,EMD),得到一系列固有模态函数(intrinsic mode function,IMF)分量,然后分别计算三相电压的各相电压前3阶IMF的能量熵和样本熵,可得到各相电压的特征向量,最后把它们组合起来作为一组特征向量的方法。针对支持向量机(support vector machine,SVM)的惩罚因子和核函数参数在寻优过程中容易陷入局部最优解的问题,提出一种改进的人工蜂群(improved artificial bee colony,IABC)算法对SVM的惩罚因子和核函数参数进行寻优,构建IABC-SVM分类器,再把提取到的特征向量进行归一化处理之后输入到构造好的IABC-SVM分类器中对样本进行训练与识别,并与粒子群算法优化支持向量机、极限学习机、BP神经网络和人工蜂群算法优化支持向量机这4种分类器进行对比,仿真结果表明所提出方法具有准确性和快速性,能够准确实现对9种不同的电压暂降源信号的辨识,有利于解决实际的工程问题。
Aiming at the voltage sag caused by single voltage sag source such as line short-circuit fault,large induction motor starting and transformer switching in distribution network and composite voltage sag source composed of single voltage sag source,an improved IEEE 33-bus distribution network system model is built in MATLAB/Simulink for simulation analysis and verification.An empirical mode decomposition(EMD)method is proposed to decompose the voltage sag signal,and aseries of intrinsic mode function(IMF)components are obtained.Then the energy entropy and sample entropy of the first three order IMFs of each phase voltage of three-phase voltage are calculated respectively.The feature vectors of each phase voltage can be obtained,and finally they are combined as a set of feature vectors.Aiming at the problem that the penalty factor and kernel function parameters of support vector machine(SVM)are easy to fall into local optimal solution in the process of optimization,an improved artificial bee colony(IABC)algorithm is proposed to optimize the penalty factor and kernel function parameters of SVM,and the IABC-SVM classifier is constructed.Then the extracted feature vectors are normalized and input into the constructed IABC-SVM classifier to train and identify the samples,and compared with the four classifiers optimized by particle swarm optimization,extreme learning machine,BP neural network and artificial bee colony algorithm.The simulation results show that the proposed method is accurate and fast,which can identify nine different voltage sag source signals and is beneficial to solve the practical engineering problems.
作者
陈晓华
吴杰康
陈盛语
王志平
蔡锦健
杨国荣
许海文
彭宇文
CHEN Xiaohua;WU Jiekang;CHEN Shengyu;WANG Zhiping;CAI Jinjian;YANG Guorong;XU Haiwen;PENG Yuwen(School of Automation,Guangdong University of Technology,Guangzhou,Guangdong510006,China;School of Electrical&Intelligentization,Dongguan University of Technology,Dongguan,Guangdong523808,China)
出处
《广东电力》
2022年第2期11-18,共8页
Guangdong Electric Power
基金
广东省基础与应用基础研究基金区域联合基金项目——粤港澳研究团队项目(2020B1515130001)。
关键词
经验模态分解
改进人工蜂群算法
支持向量机
电压暂降源辨识
改进的IEEE33节点配电网系统
empirical mode decomposition
improved artificial bee colony algorithm
support vector machine
voltage sag source identification
improved IEEE 33-bus distribution network system
