Battery State of Health Estimation and Incremental Capacity Analysis under Dynamic Charging Profile Using Neural Networks

Incremental capacity analysis (ICA) and differential voltage analysis (DVA) are recognized methods for battery degradation monitoring, but their real-world implementation has been constrained by the need for constant-current (CC) charging profiles. This research addresses this limitation by proposing a framework that extends ICA/DVA-based monitoring to dynamic charging profiles. A novel concept termed virtual incremental capacity (VIC) and virtual differential voltage (VDV) is introduced. Two convolutional neural networks (CNNs), U-Net and Conv-Net, are devised to generate VIC/VDV curves and accurately estimate the state of health (SOH) based on dynamic charging across various state-of-charge (SOC) ranges. To enhance computational efficiency, adaptations named Mobile U-Net and Mobile-Net serve as streamlined replacements for the original networks without compromising performance. The efficacy of the proposed CNNs is validated using a comprehensive experimental dataset comprising battery modules, demonstrating their ability to produce precise VIC/VDV curves and facilitate ICA/DVA-based battery degradation monitoring across diverse fast-charging protocols and SOC conditions.