Physics-Based Battery Models: An Embedded System Perspective

Physics-based battery models (PBMs) can offer significant advantages over Equivalent Circuit Models or Machine Learning-based approaches by providing deeper insight into battery-internal states. While high-quality modeling tools exist in desktop environments, deploying these models on resource-constrained embedded BMS hardware remains a major challenge. This presentation lists some of the real-world constraints of embedded systems, including measurement limitations, system scalability to thousands of cells and further restrictions. Strategies to bridge this gap are discussed, including model order reduction techniques and efficient code generation workflows. A runtime analysis of a representative foxBMS configuration illustrates the available computational budget, and a new open-source SoX-API for integrating models into foxBMS is introduced. The key takeaway is that physics-based models are ready to leave the lab if they can run on embedded hardware.