基于加权正则化极限学习机与改进梯度投影法的谐波源定位  

Study on a Harmonic Source Location Method Based on WRELM and Improved Gradient Projection Method

作　　者：沈清野 SHEN Qingye(State Grid Zhoushan Power Supply Company,Zhoushan,Zhejiang 316000,China)

机构地区：[1]国网舟山供电公司,浙江舟山316000

出　　处：《山西电力》2023年第4期13-18,共6页Shanxi Electric Power

摘　　要：针对配电网谐波源分布稀疏、监测装置少及定位难度高等特点,提出了采用基于加权正则化极限学习机与改进梯度投影法相结合的方法进行谐波源定位。具体的做法是,先构建系统量测节点的优化配置模型,采用模拟退火遗传算法进行求解,然后利用从监测节点获取的量测数据构建基于加权正则化极限学习机的谐波源定位模型,确定含有谐波源的嫌疑节点,实现谐波源的定性分析,最后再采用改进梯度投影法求取嫌疑节点的谐波电流,完成谐波源的定量分析。仿真结果表明:采用加权正则化极限学习机与改进梯度投影法相结合的谐波源定位方法具有定位精度高、速度快的优点,有一定的推广价值。A new method based on weighted regularization extreme learning machine(WRELM)and improved gradient projection method(IGPM)is proposed to solve the problem of harmonic source location which has the characteristics of sparse distribution,few monitoring devices and high location difficulty of power distribution network.The specific ways are as follows:Firstly,the optimal configuration model of the system measurement nodes is established and solved by simulated annealing genetic algorithm(SA+GA).Then,based on the measured data obtained from monitoring nodes,a model of harmonic source location based on WRELM is constructed to identify the suspected harmonic source nodes and complete the qualitative analysis of harmonic sources.Finally,IGPM is used to calculate the harmonic current of the suspected nodes and complete the quantitative analysis of harmonic sources.The simulation results show that the harmonic source location method of WRELM combined with IGPM has the advantages of high accuracy and high speed,which has certain promotional value.

关 键 词：谐波源定位 改进梯度投影法 加权正则化极限学习机 模拟退火遗传算法 

分 类 号：TM711[电气工程—电力系统及自动化]