摘要
将支持向量回归(SVR)算法引入短期负荷预测,为提高预测速度,根据负荷预测的特点,提出了一种SVR的在线训练算法,该算法通过不断输入新的负荷数据来更新回归函数,以获得更快的计算速度和较好的预测结果。和传统的SVR算法比较,它能在保证精度的同时大大减少支持向量的数目,具有更快的收敛性。仿真结果表明了算法的有效性。
An online training algorithm for short-term load fast forecasting based on SVR(Support Vector Regression) method is presented. In order to provide the accurate forecasted load, the regression function is updated by inputting new load data in the proposed algorithm. The online training algorithm not only results in a smaller number of support vectors with the same accuracy preserved but also has a much faster convergence and a better generalization performance compared with the conventional SVR algorithms. The results obtained from experiments show that the algorithm can achieve great forecasting accuracy and high computing speed. This project is supported by XJ Bounty of China Electricity Fund.
出处
《继电器》
CSCD
北大核心
2005年第9期17-20,49,共5页
Relay
基金
中华电力基金会许继奖教金资助项目~~
关键词
短期负荷预测
支持向量机
支持向量回归
short-term load forecasting
support vector machine
support vector regression