基于改进BP神经网络的电力负荷预测

在现有文献研究的基础上,对BP神经网络进行了深入研究,提出了一种新的LAFBP模型,给出了模型的标准BP算法、改进BP算法、权值和阈值的初始化方法.在此基础上,用新的LAFBP模型与传统的标准BP模型对黑龙江省巴彦县的电力负荷进行了预测.预测结果表明,新的LAFBP模型不仅克服了传统的BP模型外推效果不好的缺点,而且在模型的拟合精度、学习时间和学习次数方面明显优于传统的BP模型.

Electrical Load Forecasting Based on Improved BP Neural Network

The paper makes further research on BP neural network based on present literature,and presents a new LAFBP model,the model is given the standard BP algorithm,improved BP algorithm,weights and threshold initialization method.On this basis,this new LAFBP model and the traditional standard BP model are used for power load forecasting of Bayan town Heilongjiang Province.Forecasting results show that the new LAFBP model not only overcome the disadvantages of traditional BP model extrapolation ineffective,but also it is obviously superior to the traditional BP model in the model fitting accuracy,learning time and learning frequency.