摘要
提出了可应用于电力系统负荷预测的混合模型神经网络方法 ,该方法同时具有电力系统负荷预测的传统方法的优点及人工神经网络方法的优点 .该方法中 ,不同的负荷分量采用不同类型的预测方法 ,并采用基本频率的谐振分量作神经网络的输入 ,神经网络的训练采用快速的学习算法进行 .该方法具有很强的实时性和适应性 ,适用于没有气象资料的应用场合 .仿真计算的结果表明 ,预测精度较传统方法来得高 .
This paper presents a hybrid model neural network (HMNN) based short term electric load forecasting approach.This approach combines the traditional time series model with the neural network approach.Some load components are forecasted with traditional methods and others with neural network approaches.The base component,which is periodic for the 24 hour forecasting,is modeled with a neural network.The harmonic components of the intrinsic frequency are chosen as input variables of the neural network and the neural network is trained with a rapid convergent learning algorithm.Simulation results indicate that the hybrid model neural network based load forecasting approach produces more accurate load forecasts in comparison to the traditional method and can be applied to the case of no whether material.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2000年第1期69-72,共4页
Control Theory & Applications
基金
国家自然科学基金!( 697740 0 2 )
关键词
混合模型
神经网络
短期负荷预测
电力系统
hybrid model neural networks
short term electric load forecasting
