贝叶斯框架下最小二乘支持向量机的中长期电力负荷组合预测  

Combination method of mid-long term load forecasting based on support vector machine within the Bayesian evidence framework

作　　者：牛东晓[1] 吕海涛[1] 张云云[1] 

机构地区：[1]华北电力大学工商管理学院,北京102206

出　　处：《华北电力大学学报（自然科学版）》2008年第6期62-66,共5页Journal of North China Electric Power University：Natural Science Edition

基　　金：教育部新世纪优秀人才支持计划(NCET-07-0281)

摘　　要：影响中长期负荷变化的因素较多,单一预测模型很难满足预测需要,组合预测能够较好地解决单一模型的缺点,借鉴单一预测模型的优点。提出贝叶斯框架下最小二乘支持向量机(LS-SVM)中长期负荷组合预测模型,利用结构化风险原则代替经验风险最小化,挖掘各单一预测模型的信息,以单一模型的预测数作为组合预测输入样本,通过贝叶斯后验理论确定最小二乘支持向量机参数,建立组合预测模型进行预测。通过算例表明,提出的模型具有较高的预测精度,能够较好地解决小样本下的预测问题,具有良好的泛化能力和预测精度。Mid-long term load is affected by many factors, it is difficult to forecast by a single method. Combination forecasting is a good method to ,solve this problem and make use of the superiority of single method. In this paper, combination method of mid-tong term load forecasting based on support vector machine within the Bayesian evidence framework is proposed. Traditional empirical risk minimization principle is replaced by the principles of structure risk minimization. The forecasting results gained form single method are used as input samples. The parameters of IS- SVM are obtained by Bayesian posteriori estimation to construct the combination forecasting method. The simulation results show this model has high precision and is adept to ,solve the forecasting of least samples, and it performs good in generalization ability and forecasting precision.

关 键 词：中长期负荷 组合预测 贝叶斯框架 最小二乘支持向量机 

分 类 号：TM715[电气工程—电力系统及自动化]