摘要
利用支持向量机(SVM)和遗传算法(GA)建立24个不同的混合模型来对夏季24点负荷进行滚动预测。通过追加最新的负荷和天气信息来更新混合模型的输入,滚动预测下一小时负荷。利用SVM建立预测模型,利用GA自动选择SVM模型的参数。经过GA优化后的最终SVM模型用于滚动预测下一小时的负荷。研究实例表明,GA简化了SVM参数选择,优化了SVM模型;滚动预测效果要明显好于常规预测方法。
This paper presents hybrid models of support vector machines (SVM) and genetic algorithm (GA) to forecast summer 24 hourly loads. These models were applied to rollingly forecast the loads respectively with their inputs updated by newly obtained information of the hourly. SVM were applied to build a series rolling forecasting models. Parameters in the SVM models were automatically selected by GA to simplify the complex modeling. These optimized models were then used to forecast the rest unknown hourly loads of the day. A studied case shows that the forecasting errors of the dynamical models is significantly lower than that of the compared methods.
出处
《电工技术学报》
EI
CSCD
北大核心
2007年第6期148-153,共6页
Transactions of China Electrotechnical Society
基金
国家自然科学基金(50077007)
高等学校博士点专项基金(20040079008)资助项目。
关键词
支持向量机
小时负荷预测
遗传算法
滚动预测
Support vector machines, hourly load forecasting, genetic algorithm, rolling forecasting