基于VMD-LSTM-ELMAN模型的国际原油价格人工智能预测研究

Research on artificial intelligence predictions of international crude oil prices based on variational modal decomposition,long short-term memory,and the Elman neural network

针对国际原油价格序列的高度非线性、非平稳性和时变性等复杂特征,本文提出一种基于变分模态分解(Variational modal decomposition, VMD)和组合预测模型LSTM-ELMAN的方法对国际原油价格进行预测。首先采用VMD方法将原始原油价格分解为不同频率的子序列;然后采用不同模型分别对高频和低频序列进行预测,利用ELMAN神经网络(Elman neural network, ELMAN)预测最后一个高频分量,长短期记忆网络(LSTM,Long short-term memory network)作为主要的预测模型来预测其他子序列;最后重构不同模型的子序列预测值,进而得到最终的预测结果。实证研究结果表明,本文所提出的VMD-LSTM-ELMAN混合模型相较于对比模型不仅能够明显提高国际原油价格的预测精度,而且在不同训练集长度和市场环境下仍能保持预测优势,具有较强的泛化性与可靠性。总体而言,基于国际原油价格的实验证明了VMD-LSTM-ELMAN是一种有效且稳定的预测模型,能够为政府和企业提供有效的智能技术支持。

Aiming at the complex characteristics of international crude oil prices,which are highly nonlinear,non-stationary,and time varying,this paper proposed a method based on variational modal decomposition(VMD),long short-term memory network(LSTM),and the Elman neural network(ELMAN)to predict international crude oil prices.First,the original crude oil prices were decomposed into subsequences of different frequencies with the VMD method.Then,different models were used to predict high frequency and low frequency sequences.ELMAN was used to predict the last high frequency component,and LSTM was used as the main prediction model to predict other subsequences.Finally,the subsequence prediction values of different models were reconstructed to obtain the final prediction results.The empirical results showed that the VMD-LSTM-ELMAN hybrid model proposed in this paper not only significantly improved the prediction accuracy of international crude oil prices compared with the comparison model,but also maintained strong prediction advantage under different training set lengths and market conditions.It was shown that the model had strong generalization ability and was reliable.Overall,experiments based on international crude oil prices demonstrated that the VMD-LSTM-ELMAN method was an effective and stable forecasting model that provided effective intelligent technical support for governments and businesses.