Neuromorphic Parameter Estimation for Power Converter Health Monitoring Using Spiking Neural Networks

2026-04-17

AI Summary

This work addresses the challenge of deploying conventional physics-informed neural networks on sub-milliwatt edge devices by proposing a low-power health monitoring approach that decouples spiking temporal processing from physical constraints. Specifically, a three-layer leaky integrate-and-fire (LIF) spiking neural network is employed to estimate passive component parameters in power converters, while a differentiable ordinary differential equation (ODE) solver introduces physics-consistency loss. The method achieves, for the first time, separate training of spiking neural networks and ODE-based physical losses, leveraging membrane potential memory to enable degradation tracking and event-driven fault detection. Evaluated on a synchronous buck converter under electromagnetic interference, the approach reduces estimation error for equivalent series resistance from 25.8% to 10.2%—approaching the ±10% manufacturing tolerance—while achieving a 270-fold reduction in energy consumption and a 93% spike sparsity.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Always-on converter health monitoring demands sub-mW edge inference, a regime inaccessible to GPU-based physics-informed neural networks. This work separates spiking temporal processing from physics enforcement: a three-layer leaky integrate-and-fire SNN estimates passive component parameters while a differentiable ODE solver provides physics-consistent training by decoupling the ODE physics loss from the unrolled spiking loop. On an EMI-corrupted synchronous buck converter benchmark, the SNN reduces lumped resistance error from $25.8\%$ to $10.2\%$ versus a feedforward baseline, within the $\pm 10\%$ manufacturing tolerance of passive components, at a projected ${\sim}270\times$ energy reduction on neuromorphic hardware. Persistent membrane states further enable degradation tracking and event-driven fault detection via a $+5.5$ percentage-point spike-rate jump at abrupt faults. With $93\%$ spike sparsity, the architecture is suited for always-on deployment on Intel Loihi 2 or BrainChip Akida.

Problem

• neuromorphic computing
• power converter health monitoring
• spiking neural networks
• parameter estimation
• always-on inference

Innovation

• Spiking Neural Networks
• Neuromorphic Computing
• Physics-Informed Learning
• Power Converter Health Monitoring
• Energy-Efficient Inference