Analysis of thermal conductivity nonuniformity in thermal energy storage systems using convolutional neural networks

Thermal energy storage (TES) systems play a vital role in improving energy efficiency and driving the integration of renewable energy. However, the non-uniformity of heat conduction in TES systems has a significant impact on its performance and reliability. In this work, we propose a method for high-precision analysis and prediction of heat conduction distribution in TES systems. Specifically, the proposed model adopts a combination of physical and data-driven design, combined with the heat conduction equation, introduces a physical constraint loss function, and embeds the thermophysical laws into the optimization process of CNN to ensure that the prediction results of the model are thermodynamically reasonable. Secondly, a multi-scale convolutional structure is designed in the model, which can not only capture the detailed features of local heat conduction, but also identify large-scale thermal inhomogeneity patterns through global context information integration. In addition, in order to further improve the accuracy and adaptability of the model, high-resolution thermal imaging and finite element simulation data were constructed in data preprocessing, and the contributions of different loss terms were dynamically adjusted through the adaptive weighting mechanism, so as to achieve an optimal balance between data-driven and physical constraints. Experimental verification uses synthetic data and actual thermal imaging data, and the results show that the proposed method is superior to the existing models in terms of prediction accuracy, thermophysical consistency and computational efficiency.