Machine learning for performance modelling of high temperature coatings for steels

In several renewable energy technologies, steels are exposed to high temperatures, such as in heat transfer fluids in concentrated solar power, steam turbines, or hydrogen combustion. Aluminide diffusion coatings are a highly efficient and economical solution for their protection against corrosion.

“We are exploring the potential of machine learning to model these coatings, investigating where it can be most beneficial, identifying promising algorithms, and understanding its limitations,” explained Vladislav Kolarik from Fraunhofer ICT. He added, “Adaptation of the coating design to particular operati conditions as well as lifetime assessment could be facilitated by machine learning.”

Machine learning approaches rely solely on data and do not require physical models to describe dependencies. This is especially advantageous for systems influenced by multiple parameters, where the extent of each parameter's impact on the system is not well known. Decision tree-based algorithms, such as XGBoost, can determine the degree of influence of each single parameter. Currently, the potential of machine learning for lifetime modelling is under investigation.

Text: Martina Šaradínová, PR specialist for R&D
Image: VK archive