摘要
提出了一种将混沌和神经网络相结合的方法用于短期负荷预测。利用混沌方法重构相空间系统吸引子,用前向神经网络拟和吸引子上的全局整体映射,构成了混合预测模型。在实际应用中,将基于奇异值分解的噪声消减滤波算法应用到数据的预处理中,并用混沌学习算法来训练神经网络参数,预测结果表明了该方法的有效性。
A hybrid method based on chaos and neural network was used in the study of the electric power system short-term load forecasting. This article presents the use of chaos method to reconstruct attractors in phase spaces and a multi-layer feed forward neural network to fit the attractor抯 global map, to construct a hybrid prediction model. Moreover, this article shows the efficiency of a noise-suppressing method based on single value decomposition (SVD), and a new neural network learning algorithm, chaotic learning algorithm, is proposed. The result indicates that the method is effective.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2002年第9期15-18,共4页
Proceedings of the CSEE
基金
国家重点基础研究973专项经费项目(G1998-030405)。~~
关键词
电力系统
混沌
神经网络
短期负荷预测模型
load forecasting
chaos
neural network
single value decomposition
chaotic learning algorithm