Technology Category
Machine Learning: Learning on the Edge & Model CompressionSearch and Optimization: Mixed Discrete/Continuous SearchCognitive Modeling & Cognitive Systems: Neural Spike Coding
Physics-Embedded Neural ODEs for Sim2Real Edge Digital Twins of Hybrid Power Electronics Systems
🤖 AI Summary
Power electronic systems (PES) exhibit inherent continuous–discrete coupled dynamics that evolve with operating conditions, rendering existing modeling approaches inadequate for high-fidelity, generalizable simulation on resource-constrained edge devices.
Method: We propose a physics-embedded neural ordinary differential equation (Neural ODE) architecture: an event-driven automaton explicitly models discrete switching events, while prior physical differential equations are structurally embedded into the neural network—ensuring both interpretability and end-to-end trainability. Integrated with neural ODE solvers and an FPGA-based edge deployment toolchain, the framework enables cloud-edge collaborative real-time inference.
Results: Experiments across multiple operational scenarios demonstrate significant improvements in modeling accuracy, a 75% reduction in neuron count, and validation of the unified feasibility of high precision, low computational overhead, and enhanced real-time control capability.
Technology Category
Application Category
📝 Abstract
Edge Digital Twins (EDTs) are crucial for monitoring and control of Power Electronics Systems (PES). However, existing modeling approaches struggle to consistently capture continuously evolving hybrid dynamics that are inherent in PES, degrading Sim-to-Real generalization on resource-constrained edge devices. To address these challenges, this paper proposes a Physics-Embedded Neural ODEs (PENODE) that (i) embeds the hybrid operating mechanism as an event automaton to explicitly govern discrete switching and (ii) injects known governing ODE components directly into the neural parameterization of unmodeled dynamics. This unified design yields a differentiable end-to-end trainable architecture that preserves physical interpretability while reducing redundancy, and it supports a cloud-to-edge toolchain for efficient FPGA deployment. Experimental results demonstrate that PENODE achieves significantly higher accuracy in benchmarks in white-box, gray-box, and black-box scenarios, with a 75% reduction in neuron count, validating that the proposed PENODE maintains physical interpretability, efficient edge deployment, and real-time control enhancement.
Problem
Research questions and friction points this paper is trying to address.
Model hybrid dynamics in Power Electronics Systems accurately
Improve Sim-to-Real generalization on edge devices
Maintain physical interpretability with efficient deployment
Innovation
Methods, ideas, or system contributions that make the work stand out.
Physics-Embedded Neural ODEs for hybrid dynamics
Event automaton governs discrete switching explicitly
Cloud-to-edge toolchain enables efficient FPGA deployment
🔎 Similar Papers
No similar papers found.
