一种基于问题结构信息的大企业日用电量预测方法  

Large Enterprises′ Daily Electricity Consumption Forecasting Method Based on Structure Information of Problems

作　　者：钱峰[1] 周佃民[1] 李凯[1] 李关定[1] 李宇涛[1] 

机构地区：[1]宝钢股份宝钢分公司能源环保部,上海201900

出　　处：《冶金动力》2008年第5期1-4,共4页Metallurgical Power

摘　　要：对大型重工业企业用电特点的进行了分析,发现了"检修时间越长,日用电量越少;生产量越大,日用电量越多"的特点,以此作为问题的结构信息,提出了一种基于非负最小二乘的预测方法。该方法利用企业的检修计划和产量计划作为输入信息,构造回归函数,并把问题的结构信息引入到模型的构造中去,从而极大的提高了模型的预测精度和泛化能力。Electricity utilization characteristics of large heavy industry enterpnses were analyzed. It found out that the longer the troubleshooting time was, the less the daily electric- ity consumption was; the larger the output was, the more the daily electricity oonsumption was. Taking the above as the structure information of problems, a forecasting method based on non-negative least square was set forward. Taking troubleshooting scheme and output scheme of the enterprises as input information, this method structured regression function and intro- duced the structure information of problems into construction of a model. It greatly increased precision of prediction and generalization capacity of the model.

关 键 词：重工业企业 电量预测 检修计划 生产计划 结构信息 

分 类 号：TM7[电气工程—电力系统及自动化]