Joint deep reversible regression model and physics-informed unsupervised learning for temperature field reconstruction

Temperature monitoring over heat source components in engineering systems, such as the energy system, electronic equipments, becomes essential to guarantee working performance of these components. However, prior methods, which mainly use interpolate estimation reconstruct overall temperature field from limited points, require large amounts tensors for an accurate estimation. This may affect availability and reliability system. To solve problem, this work develops a novel reconstruction method joints deep reversible regression model physics-informed unsupervised learning heat-source systems (TFR-HSS). Firstly, we define TFR-HSS mathematically, numerically system with discrete grids, hence transform task image-to-image problem. Then, can better learn physical information, especially area near boundaries Finally, proposes loss characteristics learns without labelled samples. Experimental studies have conducted typical two-dimensional validate effectiveness proposed method. Under method, mean average error constructed achieve about 0.1K, 50% lower than other methods. Besides, takes 5.2 ms per sample inference provide real-time predictions.