摘要
提出了类电磁机制算法优化的小波对角递归神经网络的短期负荷预测模型,在常规的对角递归神经网络的隐含层神经元之间增加了同层神经元之间的相互连接,使隐含层单元之间存在相互的信息交换,模型的动态性能得到增强;隐含层函数采用小波函数,通过伸缩因子和平移因子的引入,使模型具有较强的逼近能力和容错能力。采用类电磁机制算法对小波对角递归神经网络进行优化,具有全局优化能力强、编程实现简单、收敛性好等优点。经实际负荷系统预测仿真测试,结果表明所提出的预测模型能得到满意的预测精度。
Short-term load forecasting model of wavelet diagonal recursive neural network optimized by electromagnetism-like mechanism algorithm is constructed in this paper.The connection of neurons in the hidden layer can exchange information,which increases dynamic performance of model compared with conventional diagonal recursive neural networks.The function of hidden layer adopts wavelet function,and the scale factor and translation factor are used,which increases approximation capabilities and fault-tolerant performance of model.Wavelet diagonal recursive neural network is optimized by electromagnetism-like mechanism algorithm,which possesses global optimization ability,simple programming realization and good convergence.A kind of actual load system is used to simulate,and the testing results show that the proposed forecasting model can obtain satisfactory forecasting precision.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2010年第5期51-55,共5页
Proceedings of the CSU-EPSA
基金
山东电力集团公司科学技术项目(2010A-41)
关键词
类电磁机制
小波对角递归神经网络
短期负荷预测
电力系统
electromagnetism-like mechanism
wavelet diagonal recursive neural network
short-term load forecasting
power system