摘要
针对径向基函数(radial basis function,RBF)网络在电力系统短期负荷预测中的应用,提出了一种基于动态自适应RBF网络的概率性短期负荷预测方法。采用动态自适应最近邻聚类学习算法训练网络实现负荷预测。在此基础上,通过对历史负荷预测误差特性的统计分析,对各负荷分区内预测误差的概率密度函数建模,并结合确定性预测结果获得概率性负荷预测结果。通过分析实际电网数据,验证了该方法的实用性与有效性。
According to the application of radial basis function (RBF) network in power system short-term load forecasting, a probabilistic short-term load forecasting method based on dynamic self-adaptive RBF network is proposed. The dynamic self-adaptive nearest neighbor-clustering learning algorithm is adopted to training the network for load forecasting. On this basis, by means of the statistics of error characteristics of historical load forecasting, a probability density function model for forecasting errors in load areas is established, and combining with the results of deterministic load forecasting the results of probabilistic load forecasting are obtained. The practicality and effectiveness of the proposed short-term load forecasting approach are verified by data analysis of actual power network.
出处
《电网技术》
EI
CSCD
北大核心
2010年第3期37-41,共5页
Power System Technology
基金
"十一五"国家科技支撑计划重大项目(2008BAB29B08)
国家自然科学基金雅砻江联合研究基金重点项目(50539140)
科技部水利部公益性行业科研专项(200701008)~~
关键词
短期负荷预测
概率密度函数
区间预测
径向基函数
最近邻聚类
short-term load forecasting
probabilitydensity function
interval forecasting
radial basis function(RBF)
nearest neighbour-clustering
作者简介
周建中(1959-),男,教授,博士生导师,IEEE高级会员,主要研究方向为水电能源及其复杂系统分析的先进理论与方法以及发电生产过程控制、诊断与仿真,E-mail:jz.zhou@hust.edu.cn;
张亚超(1985-),男,硕士研究生,研究方向为电力负荷预测。