短期负荷预测相关因素的自适应训练  

ADAPTIVE TRAINING APPROACHES CONSIDERING IMPACT FACTORS IN SHORT-TERM LOAD FORECASTING

作　　者：高峰[1] 康重庆[1] 程旭[1] 沈瑜[1] 夏清[1] 彭涛[1] 周安石[1] 

机构地区：[1]清华大学电机系,北京市100084

出　　处：《电力系统自动化》2002年第18期6-10,共5页Automation of Electric Power Systems

基　　金：国家重点基础研究专项经费资助项目 (G19980 2 0 311)

摘　　要：提高预测精度是短期负荷预测的基本目标。目前已提出了处理相关因素的规范策略和短期负荷预测的综合模型。在此基础上 ,将自适应训练的思想引入到短期负荷预测相关因素处理中 ,提出了相关因素自适应训练的若干概念 ,并分析了自适应训练中的基本问题 ,给出了短期负荷预测过程的抽象化模型 ,提出了两种训练负荷相关因素的算法 :摄动算法和遗传算法 ,最后比较了这两种算法的优缺点。算例分析表明 ,通过自适应训练相关因素 。To improve prediction precision is the most radical objective in short term load forecasting (STLF). On the basis of the presented unified approach considering load relative factors and integrated model of STLF, the idea of adaptive training is used to deal with load relative factors in STLF. The concept of adaptive training of load relative factors is proposed and studied in the paper. Some basic questions about adaptive training are analyzed and an abstract model describing the progress of STLF is presented. Moreover, two kinds of training methods, perturbation algorithm and genetic algorithm, are both proposed to optimize the load relative factors. The advantages and disadvantages of both methods are illustrated. Case study shows that the precision of STLF has been improved apparently after load relative factors have been trained.

关 键 词：短期负荷预测 虚拟预测 综合模型 摄动算法 遗传算法 电力系统 

分 类 号：TM761[电气工程—电力系统及自动化]