基于WT-CNN-LSTM混合神经网络的电力系统负荷预测模型

Power System Load Forecasting Model Based on WT-CNN-LSTM Hybrid Neural Network

随着电力在我国能源占比中的持续提升,电力预测在现代能源管理中具有不可替代的作用。由于电力结构的多元化以及影响因素的复杂化,传统的预测模型在电力负荷预测中存在局限性。本文结合小波变换(WT)与神经网络CNN-LSTM,将WT-CNN-LSTM混合神经网络应用于电力系统的负荷预测,并与传统机器学习模型、时间序列预测模型进行对比,结果表明WT-CNN-LSTM神经网络在电力负荷预测上具有更高的准确性,能够为电力系统运行和规划提供参考依据。

As the proportion of electricity in China's energy continues to increase,electricity forecasting plays an irreplaceable role in modern energy management.Due to the diversification of the electric power structure and the complexity of influencing factors,the traditional prediction model has limitations in electric load forecasting.This paper combines the wavelet transform WT and neural network CNN-LSTM,and applies the WT-CNN-LSTM hybrid neural network to power system load forecasting,and conducts comparative experiments with the traditional machine learning model and time series forecasting model.The results show that the WT-CNN-LSTM neural network has a higher accuracy in electric load forecasting,and can provide reference basis for the operation and planning of the power system.