摘要
电价预测有助于电力市场的优化调度。长短期记忆(long short-term memory,LSTM)网络作为一种特殊的循环神经网络(recurrent neural network,RNN),在解决此类问题时具有良好的性能。为了提高电价预测的准确性,本文将变分模态分解(variational mode decomposition,VMD)、LSTM和Adam优化算法相结合,构建了一种VMD-Adam-LSTM混合模型。经VMD分解,将原始复杂的电价序列分解为波动简单且数量有限的固有模态函数(intrinsic mode function,IMF)。VMD分解能够有效克服经验模态分解(empirical mode decomposition,EMD)中存在的模态混叠问题。先将高效的随机梯度优化器Adam与LSTM结合,再对分解后的每个IMF进行预测,可以精确捕捉到电价的波动行为。本文将VMD-Adam-LSTM混合模型应用于实际电价数据。并与其他模型相比较,验证了该模型在电价预测上具有良好的性能。
Price forecasting is helpful to optimize the power market.Long short-term memory(LSTM)network,as a special recurrent neural network(RNN),has a good performance in solving electricity price forcasting problems.In order to improve the accuracy of electricity price prediction,a VMD-Adam-LSTM hybrid model is constructed by combining VMD,LSTM and Adam optimization algorithm.The intrinsic mode function(IMF)is used to decompose the original complex electricity price sequence into the simple and limited intrinsic mode function(IMF)after VMD decomposition.VMD decomposition could effectively overcome the modal aliasing problem in empirical mode decomposition(EMD).Firstly,the efficient stochastic gradient optimizer Adam is combined with LSTM,and then the decomposed IMF is predicted,which can accurately capture the fluctuation behavior of electricity price.In this paper,the VMD-Adam-LSTM hybrid model is applied to the actual electricity price data.Compared with other models,it is verified that this model has good performance in electricity price prediction.
作者
马丽莹
魏云冰
MA Liying;WEI Yunbing(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《智能计算机与应用》
2022年第12期142-146,152,共6页
Intelligent Computer and Applications
关键词
电价预测
长短期记忆网络
变分模态分解
Adam优化算法
electricity price forecast
long and short term memory network
variational modal decomposition
Adam optimization algorithm