摘要
短期负荷预测对指导电网日常调度具有重要意义。提出了一种基于自适应K-means和深度神经网络(DNN)的短期负荷预测数据挖掘方法。首先,利用弹性网(Elastic Net)分析法辨识影响负荷数据的主导因素;其次,采用基于Davies-Bouldin指数的自适应K-means算法聚类处理;接着,为了克服传统神经网络的过拟合和不稳定性的问题,引入深度网络DNN进行预测聚类后的负荷数据;最后,以实测的负荷及气象数据进行了仿真实验,预测结果与DNN、BP方法的预测结果对比,验证了所提方法具有更好的预测精度。
Short-term load forecasting is of great significance for guiding the daily dispatch of the power grid.A data mining method of short-term load forecasting based on adaptive K-means and deep neural networks(DNN)is proposed.Firstly,the dominant factors affecting load data are identified by Elastic Net analysis method.Secondly,the adaptive K-means algorithm based on Davies-Bouldin index is used for clustering.Then,in order to overcome the over fitting and instability of the traditional neural network,the depth network DNN is introduced to predict the load data after clustering.Finally,the simulation experiment is carried out with the measured load and meteorological data,and the prediction results are compared with those of BP and DNN methods,which proves that the proposed method has better prediction accuracy.
作者
张健
Zhang Jian(State Grid Shanxi Electric Power Company,Taiyuan 030021,Chiina)
出处
《电子测量技术》
2020年第17期58-61,共4页
Electronic Measurement Technology