递归神经网络及其在电力短期负荷预测中的应用

Recursive Neural Network and Its Application to the Short─term Prediction or Power Load

本文用递归神经网络逼近非线性ＡＲＭＡ模型预测电力短期负荷。与传统方法以及前馈神经网格方法比较，递归神经网络由于其能自学习逼近非线性ＡＲＭＡ模型而具有较高的预测精度，预测方法也比较简单。这在我国电力供应紧张的情况下，对提高我国的电力负荷预测水平，合理安排电力生产计划具有一定的现实意义。

The recursive neural network based nonlinear approaching ARMA model is adopted for short-term power load prediction in this paper.Since the recursive neural network has the capability of self-learning the approaching nonlinear ARMA model,so it can achieve a better prediction accuracy in a simpler way than the conventional method or the feed forward neural network method. And at this time,it is very important and practical for improving the power load prediction and the rational distribution level of China.