一种中长期负荷预测多模型筛选新方法  

A novel method for multi-model sifting for medium and long term load forecast

作　　者：李媛媛[1] 牛东晓[1] 刘达[1] 

机构地区：[1]华北电力大学工商管理学院,北京102206

出　　处：《华东电力》2007年第11期23-26,共4页East China Electric Power

基　　金：国家自然科学基金资助项目(70671039);高等学校博士点专项基金资助项目(20040079008)

摘　　要：提出了一种基于趋势拟合评判的中长期负荷预测多模型筛选新方法。通过改进的灰色关联分析方法,定量衡量预测曲线与实际曲线的趋势拟合程度,从多种负荷预测模型中自动筛选出若干优越模型,再利用方差—协方差方法对筛选得到的模型进行组合,得到最终的优化组合预测模型。研究算例表明,该方法预测效果优于以各时点误差分析为基础的中长期负荷组合预测,而且随着步长的增加,优势越发明显。A novel method for multi-model sifting for medium and long term load forecast based on evaluation of trend comparability is presented. By using the improved gray relation analysis, the trend comparability between the forecast curve and the actual curve was calculated, which helped sift several optimal models from multiple load forecast models. The sifted models were then integrated using the variance-covariance method, and the final optimized integrated forecast model was obtained. Sample calculation shows that the novel method is more effective than other models which are based on error analysis of each point. It will be more advantageous with the increase of forecast step-length.

关 键 词：国家自然科学基金资助项目(70671039) 高等学校博士点专项基金资助项目(20040079008) 

分 类 号：TM715[电气工程—电力系统及自动化]