基于CNN-LSTM混合神经网络模型的短期负荷预测方法

为了更好地挖掘海量数据中蕴含的有效信息,提高短期负荷预测精度,针对负荷数据时序性和非线性的特点,提出了一种基于卷积神经网络(CNN)和长短期记忆(LSTM)网络的混合模型短期负荷预测方法,将海量的历史负荷数据、气象数据、日期信息以及峰谷电价数据按时间滑动窗口构造连续特征图作为输入,先采用CNN提取特征向量,将特征向量以时序序列方式构造并作为LSTM网络输入数据,再采用LSTM网络进行短期负荷预测。使用所提方法对江苏省某地区电力负荷数据进行预测实验,实验结果表明,文中所提出的预测方法比传统负荷预测方法、随机森林模型负荷预测模型方法和标准LSTM网络负荷预测方法具有更高的预测精度。

Short-term Load Forecasting Method Based on CNN-LSTM Hybrid Neural Network Model

In order to screen out the effective information contained in massive data and improve the accuracy of short-term load forecasting, a hybrid model of short-term load forecasting method based on convolutional neural network(CNN)and long short-term memory(LSTM)network is proposed due to the timing and nonlinear characteristics of load data, which takes massive historical load data, meteorological data, date information and peak-valley electricity price data as input by constructing a continuous feature map of time sliding window. Firstly, CNN is used to extract feature vectors. The feature vectors are constructed as time series and used as input data for LSTM network, which is utilized to forecast the short-term load. The proposed method is used to predict the power load data of an area in Jiangsu province. The experimental results show that the proposed prediction method has higher prediction accuracy than the traditional load forecasting method, the random forest forecasting method and the standard LSTM network forecasting method.