基于Bagging的双向GRU集成神经网络短期负荷预测

Short-term Load Forecasting Based on Bidirectional GRU Neural Network Integrated by Bagging Algorithm

为了提高电力系统短期负荷的预测精度,提出了一种基于Bagging算法的双向加权门控循环单元GRU(gated recurrent unit)集成神经网络短期负荷预测模型。对双向门控循环单元BiGRU(bidirectional gated recurrent unit)神经网络两个方向的隐含层状态进行加权求和处理,使得对负荷点的预测可以同时考虑过去和未来的信息。通过Bagging算法对双向加权GRU神经网络进行集成处理来提高模型的泛化能力。按照某地区真实负荷数据,并与反向传播BP(back propagation)神经网络、长短期记忆LSTM(long short-term memory)神经网络、单向GRU神经网络和双向GRU神经网络进行对比可以得出,所提模型有更好的预测效果。

To improve the forecasting accuracy of short-term load of power system,a short-term load forecasting modelis proposed,which is based onthe bidirectional weighted gated recurrent unit(GRU)neural network integrated by the Bagging algorithm.The hidden layer states of the bidirectional gated recurrent unit(BiGRU)neural network in two directions are weighted and summed,so that the past and future information can be considered simultaneously in the load forecasting.The Bagging algorithm is used to integrate the bidirectional weighted GRU neural network,thus improving the model’s generalization capability.According to the real load data in a certain area and compared with the back propagation(BP)neural network,long short-term memory(LSTM)neural network,GRU neural network,and BiGRU neural network,it can be concluded that the proposed model has better prediction effect.