神经网络预测中长期电力负荷对比研究

Research and comparison of neural networks for forecasting mid-long term power load

为避免传统方法预测中长期电力负荷建模的复杂性,根据电力负荷历史数据,研究了基于LM算法的BP网络、RBF网络在中长期电力负荷预测中的应用,通过神经网络对训练样本的学习,自动提取影响中长期电力负荷的诸多因素。从训练速度、预测误差等方面分析对比了两种神经网络预测能力,仿真和实例数据表明了两种神经网络在中长期电力负荷预测方面的可行性和良好效果。

In order to avoid the complex forecasting model of mid-long term power load by traditional methods, the research of mid-long term power load based on BP network and RBF network is done according to the past power load data. Two networks can obtain lots of factors from training samples automatically which influence power load. The ability of two networks for forecasting is compared with the training speed and forecasting errors. Two networks have a good performance in forecasting the power load verified by the simulation results and the measured values.