MDCNet: Long-term time series forecasting with mode decomposition and 2D convolution

• Published in: Knowledge-based systems 2024-09, Vol.299, p.111986, Article 111986
• Main Authors: Su, Jing, Xie, Dirui, Duan, Yuanzhi, Zhou, Yue, Hu, Xiaofang, Duan, Shukai
• Format: Article
• Language: English
• Subjects: Convolutional neural networks; Long-term time series forecasting; Mode decomposition
• ISSN: 0950-7051; 1872-7409
• DOI: 10.1016/j.knosys.2024.111986

Long-term time series forecasting is widely used in various real-world applications, such as weather, traffic, energy, healthcare, etc. Recently, time series decomposition techniques have been adopted in many mainstream forecasting models, such as the prevalent Transformer-based models, to help capture sophisticated temporal patterns and achieve success in several benchmark tasks. However, conventional decomposition algorithms are often based on simple operations or limited to specific fields, and therefore are not effective and applicable enough, especially for complex time series. In this paper, we propose Mode Decomposition and 2D Convolutional Network (MDCNet), a structure-simple yet effective forecasting architecture based on a more effective decomposition method and a multi-frequency time series feature extraction network with multi-scale 2D convolution. Specifically, we first introduce a Variational Mode Decomposition Block to discover intricate time patterns, which decompose time series into trend components and stationary modal components at different main frequencies. Then, we design a Trend Prediction Block and an Intrinsic Mode Functions Prediction Block to capture global correlation and hidden dependencies within different main frequencies, respectively. Furthermore, a Frequency Enhancement Module is designed to further reduce the impact of noise in long-term time series. Experiments on eight benchmark datasets show that MDCNet significantly reduces the error of the previous state-of-the-art method by 15.1% and 11.5% for multivariate and univariate time series, respectively. •A simple, efficient and robust time series forecasting network, MDCNet, is proposed.•The mode decomposition block is designed to decompose complex time series smoothly.•MDCNet uses 2D convolution to extract inter-pattern and intra-pattern dependencies.•MDCNet outperforms SOTA methods and achieves 15.1% relative performance improvement.