摘要
随着新能源并网发电的渗透率提高,非计划孤岛的危害也受到了人们的关注。当非计划性孤岛发生时,公共耦合点的电压电流会产生巨大波动。为降低对大电网运行稳定性的影响,减轻对维修人员、电力设备的危害,准确辨识出孤岛状态具有很高的研究价值。由于多机并网系统特征参数的稀释效应,大部分孤岛检测方法的准确率会降低。故提出基于随机森林算法,对并网状态和孤岛状态下运行参数进行清洗、重构、处理,选择特征向量集,生成决策树,构建孤岛检测模型。并根据特征的重要性排序,构建最优的决策树,强化其表决能力,优化随机森林模型,提高孤岛检测的准确度。
With the increasing permeability of new energy connected to the grid,the harm of unplanned island has been concerned.When the unplanned island occurs,the voltage and current of PCC would fluctuate greatly.In order to reduce the influence on the operation stability of large power grid,and lower the harm to maintenance personnel and power equipment,it is of great research value to accurately detect the island state.Due to the dilution effect of the characteristic parameters of multi-inverter grid-connected system,the accuracy of most islanding detection methods would be reduced.Therefore,based on the random forest algorithm,it is proposed to clean,reconstruct and process the operating parameters in the grid-connected state and the islanding state,select the feature vector set,generate the decision tree,and build the islanding detection model.Based on the importance of features,the optimal decision tree is constructed to strengthen the voting ability,optimize the random forest model,and improve the accuracy of islanding detection.
作者
孙雯
余思翰
Sun Wen;Yu Sihan(School of Electrical and Electronic Engineering,Anhui Science and Technology University,Bengbu,Anhui 233000,China)
出处
《黑龙江工业学院学报(综合版)》
2024年第12期126-132,共7页
Journal of Heilongjiang University of Technology(Comprehensive Edition)
基金
安徽省教育厅自然科学研究项目“基于随机森林模型的孤岛检测方法研究”(项目编号:2023AH051864)。
关键词
随机森林
孤岛检测
多机并网系统
random forest
island detection
multi-machine grid-connected system
