Hot Off the Press: Soft computing methods in the solution of an inverse heat transfer problem with phase change

Hot Off the Press: Soft computing methods in the solution of an inverse heat transfer problem with phase change

Paper in proceedings (conference paper)

This Hot Off the Press paper summarizes our recent work “Soft computing methods in the solution of an inverse heat transfer problem with phase change: A comparative study” published in Engineering Applications of Artificial Intelligence [5]. In the paper, we study inverse heat transfer problems with phase change, where the boundary heat flux is estimated. Such problems are ill-posed and their solution is challenging. Although there were conventional developed for this problem in the past, they are not well-suited for cases including phase change, as these contain strong nonlinearities that bring additional computational difficulties. For such problems, soft computing methods provide a promising approach. Four methods from distinct categories of techniques are applied to this problem and thoroughly compared - the conventional gradientbased method, a fuzzy logic-based method, a population-based meta-heuristic, and a surrogate-assisted method. A reformulation of the problem utilizing dimension reduction and decomposition schemes was developed, bringing extensive computational improvements. The metaheuristic and the surrogate-based methods showed superior performance. Their performance was also rather stable and insensitive to the location of the temperature sensor (the source of data for the inverse estimation). A Zenodo repository with the complete implementation of all considered problems and methods is available

This Hot Off the Press paper summarizes our recent work “Soft computing methods in the solution of an inverse heat transfer problem with phase change: A comparative study” published in Engineering Applications of Artificial Intelligence [5]. In the paper, we study inverse heat transfer problems with phase change, where the boundary heat flux is estimated. Such problems are ill-posed and their solution is challenging. Although there were conventional developed for this problem in the past, they are not well-suited for cases including phase change, as these contain strong nonlinearities that bring additional computational difficulties. For such problems, soft computing methods provide a promising approach. Four methods from distinct categories of techniques are applied to this problem and thoroughly compared - the conventional gradientbased method, a fuzzy logic-based method, a population-based meta-heuristic, and a surrogate-assisted method. A reformulation of the problem utilizing dimension reduction and decomposition schemes was developed, bringing extensive computational improvements. The metaheuristic and the surrogate-based methods showed superior performance. Their performance was also rather stable and insensitive to the location of the temperature sensor (the source of data for the inverse estimation). A Zenodo repository with the complete implementation of all considered problems and methods is available

Inverse heat transfer; Soft computing; Machine learning; Metaheuristics; Surrogate model; Fuzzy logic

Inverse heat transfer; Soft computing; Machine learning; Metaheuristics; Surrogate model; Fuzzy logic

MAUDER, T.; KŮDELA, J.; KLIMEŠ, L.; ZÁLEŠÁK, M.; CHARVÁT, P.

2025

01.08.2024

Association for Computing Machinery, Inc

979-8-4007-0495-6

2024 Genetic and Evolutionary Computation Conference Companion

47

48

2

https://dl.acm.org/doi/10.1145/3638530.3664073

http://hdl.handle.net/