摘要
针对电力现货价格存在的高波动性、非线性特征的问题,采用变分模态分解(VMD)和WOA-ATT-BiLSTM相结合的方法实现了短期电价预测。首先使用VMD将原始电价序列分解成多个相对平稳的子序列,然后采用结合注意力机制的ATT-BiLSTM来提取电价子序列中的特征信息并进行预测,同时引入鲸鱼优化算法(WOA)优化ATT-BiLSTM的超参数来提高预测精度,最后为验证方法的有效性,使用了法国电力市场的数据进行实验比较。结果表明,基于VMD和WOA-ATT-BiLSTM模型的平均绝对百分比误差(MAPE)为2.91%,均方根误差(RMSE)为1.65欧元/MWh,平均绝对误差(MAE)为1.29欧元/MWh,相较于其他对比模型具有更准确的预测效果。
In order to solve the problem of high volatility and nonlinear characteristics of spot price of electricity,a short term electricity price prediction method is implemented by combining variational mode decomposition(VMD)and WOA-ATT-BiLSTM.First,VMD is used to decompose the original electricity price sequence into several relatively stable subsequences.Then,the ATT-BiLSTM combined with attention mechanism is used to extract the characteristic information of the electrovalence subsequence and predict it.At the same time,whale optimization algorithm(WOA)is introduced to optimize the hyper-parameters of ATT-BiLSTM to improve the prediction accuracy.Finally,to verify the effectiveness of the method,the data of French electricity market are used for experimental comparison.The experimental results show that the mean absolute percentage error(MAPE),root mean square error(RMSE)and mean absolute error(MAE)of VMD and WOA-ATT BiLSTM models are 2.91%,1.65 Euro/MWh and 1.29 Euro/MWh respectively,which has more accurate prediction effect compared with other comparison models.
作者
王超
陈奇
谷新梅
姜湖
郭芳
邓尚云
严海贤
WANG Chao;CHEN Qi;GU Xinmei;JIANG Hu;GUO Fang;DENG Shangyun;YAN Haixian(Guangzhou Southern Investment Group Co.,Ltd.,Guangzhou 510663,China;China Energy Engineering Group Guangdong Electric Power Design Institute Co.,Ltd.,Guangzhou 510663,China;Guangdong Kenuo Surveying Engineering Co.,Ltd.,Guangzhou 510663,China;School of Mechatronic Engineering and Automation,Foshan University,Foshan 528000,China)
出处
《内蒙古电力技术》
2023年第4期73-80,共8页
Inner Mongolia Electric Power
关键词
短期电价预测
变分模态分解
注意力机制
双向长短期记忆神经网络
short-term electricity price prediction
variational modal decomposition
attention mechanism
bidirectional long and short term memory neural network
