WEB Predicting laser-induced functional surface textures: A comparison of random forest and neural networksWednesday (23.09.2020) 10:55 - 11:10 F: Functional Materials, Surfaces, and Devices 2 Part of:
Due to its flexible applications, functional laser surface texturing (LST) has developed into a powerful tool to fabricate micro-textures which mimic natural surfaces, for example the lotus effect for self-cleaning properties or the texture of Collembola skin for antibacterial behaviour [1,2]. With the increasing capabilities in functional LST, the prediction of surface properties becomes more and more important in order to reduce the time-of-development of surface functionalities. Consequently, advanced approaches for the prediction of the properties of laser-processed surfaces – the so-called predictive modelling – are required .
This presentation will introduce the concept of predictive modelling with respect to LST by means of selected laser-based manufacturing techniques such as Direct Laser Writing (DLW) and Direct Laser Interference Patterning (DLIP). Two fundamental prediction approaches, namely Artificial Neural Network and Random Forest, were trained with experimental data for stainless steel surfaces. The modelling took into consideration the resulting topographic of the ablation, the used laser parameters and other process-relevant information for the prediction of the surface roughness. Statistical results indicate that both models can predict the desired surface functionality and topography with high accuracy, despite the use of a small dataset for the learning process. It is shown that from the forecasted surface roughness conclusions can be drawn about the resulting functionality, such as the wettability properties. This allows estimating the functional performance of a surface before the laser texturing process is initiated.