周、月负荷预测的GS-ANN综合模型

GS-ANN based integrated model for weekly and monthly load forecasting

灰色系统(GS)和人工神经网络(ANN)作为2种行之有效的负荷预测工具有着各自的优缺点和适用范围。将灰色系统预测和人工神经网络预测的结合方式总结为5种,即并联型、串联型、灰色系统辅助人工神经网络型、人工神经网络辅助灰色系统型和补偿型。针对周、月负荷预测的规律和需要,通过对比推荐采用串联型综合模型,提出在GS的基础上将主要影响因素也作为ANN的输入,对周、月负荷序列进行建模。通过算例证明了其在周、月负荷预测中的精度明显优于单一的灰色系统模型和人工神经网络模型。

The gray theory and the artificial neural network as two well-established tools for load forecasting have their respective advantages and disadvantages and the applicable scope. The union ways of the two load forecasting tools in predecessor＇s foundation were summarized and the union ways were classified into five kinds of definition： synthetic parallel connection model, synthetic series connection model, neural network system with auxiliary grey system model, grey system with auxiliary neural network system model and compensated model. Based on the actual need, the synthetic series connection model was chosen for modeling of weekly ＆ monthly load. The actual examples prove that the synthetic series connection has better accuracy in weekly ＆ monthly load forecasting than the gray forecast model or the artificial neural network forecast model.