摘要
为了提升智能电网负荷预测准确率,提出了一种基于深度学习的短期电力负荷预测模型.在长短时记忆网络和卷积神经网络基础上,构建混合CNN-LSTM预测模型结构.利用基于叠加卷积降噪自动编码器对电力数据进行特征提取,提出包含2个堆叠的LSTM层和1个线性输出层的负荷预测模型.24 h短期负荷预测结果表明,所提模型MAE、RMSE、MAPE和R2 指标分别为 232.08、292.19、0.0322、0.909,与 XGBoost 模型相比,性能分别提升 74.8％、73.8％、70.8％和 10.9％.
Abstract
In order to improve the accuracy of smart grid load forecasting,a short-term power load forecasting model based on deep learning is proposed.On the basis of the long-short term memory network and convolutional neural network,a hybrid CNN-LSTM prediction model structure is constructed.The automatic encoder based on superposition convolution noise reduc-tion is used to extract the features of power data,and a load forecasting model with two stacked LSTM layers and a linear out-put layer is proposed.The 24 h short-term load forecasting results show that the MAE,RMSE,MAPE and R2 indicators of the proposed model are 232.08,292.19,0.0322 and 0.909,respectively,and the performance is improved by 74.8％,73.8％,70.8％and 10.9％,respectively,compared with XG Boost model.
维普
万方数据