基于SVM的火电企业能效监督评价体系  

Energy efficiency supervision and evaluation system for thermal power companies based on SVM

作　　者：周永芳[1] 时章明[1,2] 沈浩 

机构地区：[1]中南大学能源科学与工程学院 [2]湖南节能评价技术研究中心 [3]山西省电力勘探设计院

出　　处：《冶金能源》2015年第4期3-6,45,共5页Energy For Metallurgical Industry

基　　金：湖南省发改委湘西地区研究开发课题(湘发改赈[2013]1618号)

摘　　要：针对现有评价方法在能效评价中存在的不足,提出了基于支持向量机的能效评价方法。收集了某电厂60组运行数据,提取涵盖7个工序的28个具体的监督指标训练、测试支持向量机。结果表明,提出的方法能够有效地对小样本能效数据分类,并且具有良好的泛化性。不同的模型参数对分类性能影响较大,采用网格寻参与遗传算法寻参(GA寻参)相结合的方式,对比选择最优参数。在此基础上完成了预测模块的建立,分别预测了电厂较为重要的两个指标:机组负荷与发电煤耗,完善了所建立的能效监督评价体系。In view of the existence about energy efficiency supervision and evaluation deficiencies,investigated the application of support vector machines（ SVMs） to energy efficiency supervision and evaluation system. 60 groups operating data of a power plant were collected to train and test models of SVMs,which include 28 specific monitoring indicators that belong to 7 processes. The results indicate that the proposed approach can classify small sample size energy efficiency data effectively with good generalization performance on which the different model parameters has a significant impact. A combination of grid search method and GA method was taken by comparing to select the best parameters.And on this basis prediction module of unit load and power consumption was established,which matter very much to a power plant. The prediction module has improved the energy efficiency supervision and evaluation system. These results demonstrate considerable potential in applying SVMs to energy efficiency supervision and evaluation.

关 键 词：能效监督评价 指标体系 支持向量机 火力发电 

分 类 号：TM621[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]