基于LSTM-NPGARCH的电力市场售电量预测模型  

Electricity Market Electricity Sales Forecast Model Based on LSTM-NPGARCH Algorithm

作　　者：王蕾[1] 李斌[1] 吴飞 王鹏 Wang Lei;Li Bin;Wu Fei;Wang Peng(Northeast Electric Power University,Jilin,China;Jilin Jiantong Electric Technology Co.,Ltd.,Jilin,China)

机构地区：[1]东北电力大学,吉林吉林 [2]吉林市简通电气科技有限公司,吉林吉林

出　　处：《科学技术创新》2023年第20期209-212,共4页Scientific and Technological Innovation

基　　金：吉林省科技发展计划项目《基于新一代人工智能的电力市场售电智慧服务云平台》(20200401097GX)。

摘　　要：售电量的准确预测对推动电力市场的发展和建设具有十分重要的意义,考虑售电量具有非平稳性、非线性和随时间变化的复杂特性,本文提出基于小波变换和LSTM算法的短期售电量预测模型。首先采取小波变换法将售电量数据分解为细节分量和近似分量,然后使用LSTM模型进行预测,得到初步预测结果,再使用NPGARCH模型进行预测结果修正,最后将预测的结果累加,得到最终售电量预测结果。在实验中采用某售电公司的真实数据集,基于历史统计售电量数据的预测结果分析表明,本文提出的预测模型具有良好的预测精度。The accurate prediction of electricity sales is of great significance to promote the development and construction of the electricity market.Considering the complex characteristics of non-stationary,nonlinear and time-varying electricity sales,this paper proposes a short-term electricity sales forecast based on wavelet transform and LSTM algorithm Model.First,the wavelet transform method is used to decompose the electricity sales data into detailed components and approximate components,and then the LSTM model is used for prediction to obtain preliminary prediction results,and then the NPGARCH model is used to correct the prediction results,and finally the prediction results are accumulated to obtain the final electricity sales forecast result.In the experiment,the real data set of an electricity sales company is used,and the analysis of the prediction results based on the historical statistical electricity sales data shows that the prediction model proposed in this paper has good prediction accuracy.

关 键 词：售电量预测 小波分解 长短期记忆神经网络 非参数广义自回归条件异方差模型 

分 类 号：F426.61[经济管理—产业经济]