Experimental Analysis and Identification of Thermodynamic Transfer Functions of the Laser Remelting Process Towards Developing Process Control

Abstract

Laser remelting (LRM) is a laser-based technique for surface modification, capable of polishing and restructuring surfaces at high speed and with full automation. However, LRM is sensitive to thermodynamic disturbances, making real-time monitoring and process control essential for achieving consistent results. This thesis investigates thermographic (TG) monitoring as a basis for closed-loop control of LRM. A coaxial thermographic camera was used to record transient responses of the molten pool under variations in laser power, scanning speed, and exposure time. The resulting image sequences were reduced to scalar signals, enabling identification of first-order transfer functions (TFs) that capture the dynamic relationship between process inputs and thermographic outputs. In addition to classical image signatures, a novel Gaussian parameterization strategy was introduced to describe thermographic distributions. These Gaussian descriptors not only allow reconstruction of missing image regions but also provide physically interpretable features, such as intensity and distribution width, parameters that were shown to align with conventional signatures. The findings demonstrate that exposure time is as influential as laser power or scanning speed in shaping thermographic response, and that Gaussian parameters offer an expanded set of feedback variables for process control. Together, these results establish a foundation for future closed-loop strategies in LRM, where real-time thermographic feedback can be systematically linked to surface quality outcomes.