基于广义回归神经网络的有源配电网网供负荷预测方法

Load Forecasting Method of Active Distribution Network Based on Generalized Regression Neural Network

随着大量分布式电源(DG)接入配电网和用户终端再电气化程度的提高,提出了一种基于广义回归神经网络(GRNN)的有源配电网网供负荷预测方法。首先,对含DG与电采暖负荷的配电网网供负荷特性进行了分析;其次,基于广义回归神经网络,建立了有源配电网网供负荷预测的模型;再次,以该地区近几年的国民生产总值、人口、电力消费以及网供负荷峰值等经济数据作为模型的输入,得到了该地区在目标年配电网网供负荷的预测结果;最后,分别与采用前馈反向传播神经网络(FFBNN)以及级联正反向传播神经网络(CFBNN)得到的负荷预测结果进行了对比,验证了所提方法具有更高的预测准确性。

With a large number of distributed generation(DG)access to the distribution network and the increasing re-electrification degree of user terminal re electrification,a generalized regression neural network(GRNN)based load forecasting method for active distribution network substation network is proposed.Firstly,the load characteristics of substation distribution network including DG and electric heating load are analyzed.Secondly,based on the research of generalized regression neural network,the model of load forecasting for substation network of active distribution network is established;then,the economic data of GDP,population,power consumption and peak load of power supply in recent years in this area,are taken as the input of the model,and the target year of the region is obtained.Finally,the load forecasting results obtained by using the feedforward back propagation neural network(ffbnn)and cascaded forward and backward propagation neural network(cfbnn)are compared respectively,and the results are verified The method proposed in this paper has higher prediction accuracy.