Physics-Informed Machine Learning for Transformer Condition Monitoring -- Part II: Physics-Informed Neural Networks and Uncertainty Quantification

TL;DR

This paper advances transformer condition monitoring by integrating physics-informed neural networks and uncertainty quantification, enhancing robustness and reliability in predictive maintenance of electrical transformers.

Contribution

It introduces Bayesian PINNs for transformer health assessment, combining physics-based modeling with uncertainty quantification for improved predictive accuracy.

Findings

Bayesian PINNs effectively quantify epistemic uncertainty.

Physics-informed models improve prediction robustness with sparse data.

Application to thermal and insulation aging models demonstrates practical utility.

Abstract

The integration of physics-based knowledge with machine learning models is increasingly shaping the monitoring, diagnostics, and prognostics of electrical transformers. In this two-part series, the first paper introduced the foundations of Neural Networks (NNs) and their variants for health assessment tasks. This second paper focuses on integrating physics and uncertainty into the learning process. We begin with the fundamentals of Physics-Informed Neural Networks (PINNs), applied to spatiotemporal thermal modeling and solid insulation ageing. Building on this, we present Bayesian PINNs as a principled framework to quantify epistemic uncertainty and deliver robust predictions under sparse data. Finally, we outline emerging research directions that highlight the potential of physics-aware and trustworthy machine learning for critical power assets.