RBF神经网络电力负荷预测模型研究

Research on RBF Neural Network Models of Power Load Forecasting

运用Matlab神经网络工具箱建立了一个RBF神经网络,依据某地实际的历史电力负荷数据和天气数据作为训练样本和测试样本,进行了考虑历史天气状况因素的电力系统短期负荷的预测和仿真,预测结果平均相对误差较小,满足精度要求,并将此RBF负荷预测模型与BP神经网络建立的短期电力负荷预测模型的预测结果进行了比较,显示了在相同预测条件下,RBF神经网络相比于BP神经网络在电力系统短期负荷预测方面的优越性.

A RBF neural network model built by Matlab neural network toolbox is presented. The actual his- torical electric load and weather data is used as training and test samples in this model. And the forecasting result of the simulation shows that the average relative error is small which is satisfied with the topic design accuracy requirement. Comparing with the results obtained by the RBF model and BP neural network model, the RBF one is proved to have better effect in the field of the electrical power system short-term load forecast under the same forecast condition.