Thermal simulations in additive manufacturing using machine learning

Thermal simulations are critical in additive manufacturing to ensure product quality, structural integrity, and optimized designs without the need for repeated real experiments. Conventionally, these simulations utilize numerical-based methods, which can often be time-consuming and resource-intensive, especially as the problem domain expands. To address these limitations, a data-driven framework is proposed. In this work, a methodology to train and validate machine learning-based models for thermal simulation tasks is outlined. For this purpose, an exemplary 2D simulation scenario in laser-based powder bed fusion is considered. Furthermore, the developed framework is utilized in parameter selection with the aim to obtain an energy-efficient process.