基于粗糙集降维和相关向量机的长期用电需求预测方法  

Long-term electricity demand forecasting method based on rough set reduction and relevance vector machine

作　　者：郭晓鹏[1] 杨淑霞[1] 杨里 

机构地区：[1]华北电力大学经济与管理学院,北京102206 [2]福建省电力科学研究院,福建福州350007

出　　处：《中南大学学报（自然科学版）》2013年第12期5133-5138,共6页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(71071054);中央高校基本科研业务费专项资金资助项目(11QR34)

摘　　要：提出基于粗糙集降维的相关向量机用电量预测模型。选取1996—2010年北京市的GDP作为输入值,对应的全社会用电量作为输出值进行分析验证。研究结果表明:相关向量机是一种新的监督学习方法,与支持向量机相比,它更加稀疏,泛化能力更强且不需要设置惩罚因子,而粗糙集降维被用于从多个相关因素中筛选出适用于RVM回归模型的输入向量集,进而提高算法效率;基于相关向量机的用电量预测模型比经过优化参数后的支持向量机预测模型更优。The electricity demand forecasting model based on the rough set and relevance vector machine was studied. To verify the validity of the model, the GDP and the electricity consumption data of Beijing from 1996 to 2010 were selected and analyzed with the GDP data was selected as input data, and the electricity consumption data used as output data. The results show that the relevance vector machine is a new supervising learning method. Compared with the support vector machine, it is sparser, with more generalization abilities and does not need to set the penalty factor. Rough set reduction is used to filter out the input vector for RVM regression model from a number of related factors, thus improving the efficiency of the algorithm. The electricity demand forecasting model based on the relevance vector machine is better than the support vector machine prediction model based on particle swarm optimization parameters.

关 键 词：粗糙集 相关向量机 RVM回归模型 预测 用电需求 

分 类 号：TU457[建筑科学—岩土工程] TU413.6[建筑科学—土工工程]