Data-driven Power Loss Identification through Physics-Based Thermal Model Backpropagation

2025-03-31

AI Summary

Direct measurement of power losses in power electronics digital twins remains infeasible due to hardware and operational constraints. Method: This paper proposes a physics-informed, temperature-sensor-only power loss inversion method. It introduces a hybrid framework integrating neural networks with a reduced-order thermal model, enabling differentiable backward propagation through thermal dynamics for the first time. A physics-guided loss function and adaptive normalization strategy are further designed to jointly optimize accuracy, stability, and real-time performance. Results: Experimental validation demonstrates significant improvements: temperature prediction error decreases from 7.2±6.8°C to 0.3±0.3°C, and power loss estimation error drops from 5.4±6.6 W to 0.2±0.3 W. The proposed approach substantially outperforms conventional pure-physics models, establishing a high-fidelity, deployable paradigm for loss-aware digital twins in power electronics systems.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems

Abstract

Digital twins for power electronics require accurate power losses whose direct measurements are often impractical or impossible in real-world applications. This paper presents a novel hybrid framework that combines physics-based thermal modeling with data-driven techniques to identify and correct power losses accurately using only temperature measurements. Our approach leverages a cascaded architecture where a neural network learns to correct the outputs of a nominal power loss model by backpropagating through a reduced-order thermal model. We explore two neural architectures, a bootstrapped feedforward network, and a recurrent neural network, demonstrating that the bootstrapped feedforward approach achieves superior performance while maintaining computational efficiency for real-time applications. Between the interconnection, we included normalization strategies and physics-guided training loss functions to preserve stability and ensure physical consistency. Experimental results show that our hybrid model reduces both temperature estimation errors (from 7.2+-6.8{deg}C to 0.3+-0.3{deg}C) and power loss prediction errors (from 5.4+-6.6W to 0.2+-0.3W) compared to traditional physics-based approaches, even in the presence of thermal model uncertainties. This methodology allows us to accurately estimate power losses without direct measurements, making it particularly helpful for real-time industrial applications where sensor placement is hindered by cost and physical limitations.

Problem

• Identify power losses using temperature measurements only
• Combine physics-based and data-driven models for accuracy
• Reduce errors in temperature and power loss estimation

Innovation

• Hybrid physics-data model for power loss
• Neural network corrects thermal model outputs
• Normalization and physics-guided loss functions