基于 ARIMA 模型及线性神经网络的用电量需求预测研究

摘要

ARIMA(0,2,2) model was established by using stationary data of historical electricity consumption after second order difference,and the average relative error was calculated. The relationship between the histor-ical power was approached piecewise based on linear neural network. Finally,an empirical case study on Beijing City was done, and the results show that the effect of linear neural network is better than ARIMA model.%利用二阶差分后平稳的历史用电量数据,建立 ARIMA(0,2,2)模型,计算平均相对误差,并进一步采用线性神经网络分段逼近历史用电量之间的关系。最后以北京市为例进行实证研究,结果显示：线性神经网络的预测效果优于 ARIMA 模型。