基于改进BP网络的电力系统超短期负荷预测

Ultra-short Term Load Forecasting Based on Modified BP Network

超短期电力负荷预测是电力系统运行管理的重要工作之一。本文提出了一种基于改进BP网络的超短期负荷预测方法。考虑了训练样本中坏数据的剔除,建立了变结构的BP网络模型,然后采用变学习步长的方法对BP网络模型进行训练。对某地区实际负荷的预测结果表明该方法有较高的预测精度,能取得令人满意的结果。

Ultra - short term load forecasting is one of important operation managements of power system. An ultra - short term load forecasting method based on modified BP network is proposed. Bad data in exercise sample are excluded. Varied structure BP network is put forward, which is exercised via varied study step. The validity of the above method is illuminated by a practical testing example.