电力负荷预测作为电力行业的重要参数,它为电力相关部门的规划设计提供了决策支持.但是面对电力负荷数据的非线性和强波动的特点,传统的预测结果精准度很低.为此,提出一种基于LSTM(Long Short Term Memory Networks)循环神经网络中长期电力负荷智能预测的方法.首先介绍基于LSTM神经网络的工作原理和模型搭建;然后结合算法流程图,对某地区电力负荷历史数据和该地区的经济、气象等历史数据,进行算法试验和效果验证;通过和目前本领域主流的算法比较数据训练过程中相关参数的变化判断所提算法的性能和鲁棒性.结果表明,该方法能够很好地提高电力负荷预测的精准度.
Based on LSTM Network For Medium-Term and Long-Term Power Prediction Based on Pytouch
Power Prediction can provide scientific decision support for the forward-looking construction of relevant guidance de-partments of power system.However,in the face of the nonlinear and strong fluctuation characteristics of power load data,the accura-cy of traditional prediction results is very low.Therefore,this paper proposes an intelligent power load forecasting method based on CNN optimized LSTM cyclic neural network based on pytouch deep learning platform.The results show that this method can improve the accuracy of power load forecasting.
万方数据
维普
