基于改进Elman网络的燃气负荷预测

Gas load forecasting with improved Elman network

为提高燃气负荷预测的精度,分析了燃气小时负荷的变化规律和影响因素,建立了燃气小时负荷预测模型,采用具有输出-输入反馈机制的改进Elman(OIF Elman)网络对燃气小时负荷进行预测.与传统的Elman网络比较,OIF Elman网络不仅计入了隐层节点的反馈,而且考虑输出层节点的反馈,以便从有限的训练样本中获得更多的信息.预测结果表明,在样本点较少时,无论在训练速度上,还是在预测精度上,OIF Elman网络明显优于Elman网络.OIF Elman网络提高了网络的泛化能力,既降低了对训练样本个数的需求,又能提高预测精度,在燃气负荷预测中得到成功的应用.

In order to improve the forecasting accuracy,in accordance with the influence factors and characteristics of hour gas load,a model has been established to forecasting hour gas load with OIF(output-input feedback) Elman network. Compared with convertional Elman network,OIF Elman network takes into account not only the hidden nodes feedback but also the output feedback so as to obtain more information from limited sampling spots.The emulation illustrates that OIF Elman network is better than Elman network not...