Multiscale Dual-path Feature Aggregation Network for Remaining Useful Life Prediction of Lithium-Ion Batteries

TL;DR

This work tackles lithium-ion battery RUL forecasting by addressing the need to model both global degradation trends and local pattern fluctuations. It introduces MDFA-Net, a dual-path architecture with MF-Net for multiscale, information-preserving features and EC-Net, a CNN–Transformer hybrid that captures long-range dependencies and local cues, fused via a position-enhanced attention head. Experiments on NASA and CALCE datasets show MDFA-Net consistently outperforms strong baselines, with notable improvements in $R^2$, $E_{MAE}$, and $E_{RMSE}$, and acceptable real-time inference latency. The findings suggest that combining shallow, multiscale representations with deep, attention-guided global modeling yields robust RUL predictions under variable operating conditions, with transfer learning proposed for future work to mitigate data scarcity.

Abstract

Targeted maintenance strategies, ensuring the dependability and safety of industrial machinery. However, current modeling techniques for assessing both local and global correlation of battery degradation sequences are inefficient and difficult to meet the needs in real-life applications. For this reason, we propose a novel deep learning architecture, multiscale dual-path feature aggregation network (MDFA-Net), for RUL prediction. MDFA-Net consists of dual-path networks, the first path network, multiscale feature network (MF-Net) that maintains the shallow information and avoids missing information, and the second path network is an encoder network (EC-Net) that captures the continuous trend of the sequences and retains deep details. Integrating both deep and shallow attributes effectively grasps both local and global patterns. Testing conducted with two publicly available Lithium-ion battery datasets reveals our approach surpasses existing top-tier methods in RUL forecasting, accurately mapping the capacity degradation trajectory.

Figures (6)

• Figure 1: Architecture of MDFA-Net.
• Figure 2: The diagram of the proposed first path multiscale feature network (MF-Net).
• Figure 3: The diagram of the proposed second path encoding network (EC-Net).
• Figure 4: RUL prediction results on four test LIB.
• Figure 5: RUL prediction results on CALCE.