摘要
现有研究工作表明电力系统负荷数据具有弱混沌性。在负荷预测混沌建模方法中,Lvapunov指数预报模式具有理论基础强、模型简单、预测精度高等优点,但预测时限受负荷吸引子最大Lyapunov指数限制。针对Lyapunov指数预报模式的不足,提出了k-△t间隔采样混沌模型,首先将原始负荷序列分解为多个不相交的了序列,然后对各个子序列分别建立Lyapunov指数预报模型。改进了求解最大 Lyapunov指数的方法,探讨了原始负荷序列最大可分解子序列数目的确定依据。数值实验结果表明文中提出的模型能有效地提高负荷预测精度、增加预测时限。
The weak chaotic property of electric load has been proven by the current research work. Among the forecasting models based on chaos theory, the Lyapunov exponent forecasting model has the strong foundation of theory with simple model and precise forecasting performances, but the forecasting length is limited by the largest Lyapunov exponent of the load attractor. The k-At interval sampling chaotic model is presented in this paper to solve the limitation of the Lyapunov exponent forecasting model. The original load data is first divided into several sub-series and the largest Lyapunov exponent forecasting method is then applied to the sub-series. The method of calculating the largest Lyapunov exponent is also improved. In addition, the first local minimum of the mutual information function of the load data is suggested as the criterion for selecting the largest number of the sub-series The numerical experiments demonstrate that the precision is improved and the forecasting length is increased with the model proposed in this paper.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2006年第10期28-32,共5页
Proceedings of the CSEE
基金
国家自然科学基金项目(60225006)
高等学校博士点基金项目(20030335003)
关键词
电力负荷预测
最大Lyapunov指数预报模式
预测时限
互信息
Electric load forecasting: Largest Lyapunov exponent forecasting model
Time limit of the forecasting length
Mutual information