Abstract:A new variational mode decomposition (VMD) based deep learning approach is proposed in this paper for time series forecasting problem. Firstly, VMD is adopted to decompose the original time series into several sub-signals. Then, a convolutional neural network (CNN) is applied to learn the reconstruction patterns on the decomposed sub-signals to obtain several reconstructed sub-signals. Finally, a long short term memory (LSTM) network is employed to forecast the time series with the decomposed sub-signals and the reconstructed sub-signals as inputs. The proposed VMD-CNN-LSTM approach is originated from the decomposition-reconstruction-ensemble framework, and innovated by embedding the reconstruction, single forecasting, and ensemble steps in a unified deep learning approach. To verify the forecasting performance of the proposed approach, four typical time series datasets are introduced for empirical analysis. The empirical results demonstrate that the proposed approach outperforms consistently the benchmark approaches in terms of forecasting accuracy, and also indicate that the reconstructed sub-signals obtained by CNN is of importance for further improving the forecasting performance.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the complexity and non - linearity in time - series prediction. Specifically, traditional statistical and econometric models perform poorly when dealing with time - series having complex non - linear patterns. In order to capture the non - linear and dynamic characteristics of these complex time - series data, researchers have introduced various non - linear models and deep - learning models. This paper proposes a new deep - learning method based on variational mode decomposition (VMD), aiming to improve the accuracy of time - series prediction through a decomposition - reconstruction - integration framework. The main innovation of this method lies in embedding the reconstruction, individual prediction and integration steps into a unified deep - learning framework, so as to better capture the complex patterns in the time - series and improve the prediction performance. ### Main problems: 1. **Limitations of traditional models**: Traditional linear models (such as ARIMA, GARCH, etc.) are not effective in dealing with complex non - linear time - series. 2. **Deficiencies of existing deep - learning models**: Although existing deep - learning models (such as LSTM, CNN, etc.) perform well in some time - series prediction tasks, they are usually applied in separate steps and lack a unified framework to combine the advantages of different models. 3. **Application of the decomposition - reconstruction - integration framework**: Although the existing decomposition - reconstruction - integration framework is effective, it is usually carried out step by step and fails to fully utilize the end - to - end training ability of deep - learning models. ### Solutions: The VMD - CNN - LSTM method proposed in this paper solves the above problems in the following ways: - **Variational mode decomposition (VMD)**: Decompose the original time - series into multiple sub - signals to simplify the prediction difficulty of each sub - signal. - **Convolutional neural network (CNN)**: Used to learn and reconstruct the decomposed sub - signals, generating reconstructed sub - signals that are easier to model. - **Long - short - term memory network (LSTM)**: Used to predict the decomposed and reconstructed sub - signals, and finally integrate these prediction results to obtain the final prediction output. In this way, this method can not only better capture the complex patterns in the time - series, but also achieve end - to - end training through a unified deep - learning framework, thereby improving the accuracy and robustness of prediction.

A New Unified Deep Learning Approach with Decomposition-Reconstruction-Ensemble Framework for Time Series Forecasting

A Novel Sequence-to-Sequence-Based Deep Learning Model for Multistep Load Forecasting.

M-EDEM: A MNN-based Empirical Decomposition Ensemble Method for Improved Time Series Forecasting

Enhancing Time Series Forecasting: A Hierarchical Transformer with Probabilistic Decomposition Representation

Revitalizing Multivariate Time Series Forecasting: Learnable Decomposition with Inter-Series Dependencies and Intra-Series Variations Modeling

A novel decomposed-ensemble time series forecasting framework: capturing underlying volatility information

Time series forecasting model for non-stationary series pattern extraction using deep learning and GARCH modeling

A Novel Encoder-Decoder Model for Multivariate Time Series Forecasting

A decomposition-guided mechanism for nonstationary time series forecasting

TDG4MSF: A temporal decomposition enhanced graph neural network for multivariate time series forecasting

Multivariate Time Series Modeling and Forecasting with Parallelized Convolution and Decomposed Sparse-Transformer

Financial Time Series Forecasting with the Deep Learning Ensemble Model

DDTCN: Decomposed dimension time-domain convolutional neural network along spatial dimensions for multiple long-term series forecasting

Reinforced Decoder: Towards Training Recurrent Neural Networks for Time Series Forecasting

DeformTime: Capturing Variable Dependencies with Deformable Attention for Time Series Forecasting

TCLN: A Transformer-based Conv-LSTM Network for Multivariate Time Series Forecasting

A hybrid framework for multivariate long-sequence time series forecasting

A modular neural network with empirical mode decomposition and multi-view learning for time series prediction

A novel hybrid deep learning time series forecasting model based on long-short-term patterns

Improving the Transferability of Time Series Forecasting with Decomposition Adaptation

Hybrid deep learning and empirical mode decomposition model for time series applications.