基于贝叶斯框架和回声状态网络的日最大负荷预测研究

Daily Peak Load Forecasting Based on Bayesian Framework and Echo State Network

为克服神经网络中的伪回归问题,对标准的回声状态网络进行改进,用贝叶斯理论提高网络的泛化能力。在实证算例分析中,采用某地区的实际负荷数据和相关气候数据,对该地区的日最大负荷进行预测,验证所提方法的有效性和适用性。对比试验的预测结果表明,改进的回声状态网络比标准回声状态网络和前馈神经网络预测效果更精确,网络泛化能力更强。

To overcome the pseudo-regression in neural network,the standard echo state network is improved and generalization ability of standard echo state network is enhanced by Bayesian framework.To verify the availability and adaptability of the proposed method,in the analysis on empirical example actual load data and related climatic data of a certain region are used as input variables to forecast the daily peak load of this region.Comparison of the forecasted daily load curve with actual daily load curve shows that the forecasted load curve by improved echo state network is more accurate than those forecasted by BP neural network and standard echo state network,and the generalization ability of the improved echo state network is more stronger.