When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.
Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.
Industrial electropolishing facilities can enhance their competitiveness in the market by embracing the concept of mass personalization, which involves closely integrating customers in the design process to accommodate their specific preferences. Traditional electropolishing operations rely heavily on skilled operators and costly trial-and-error experiments to determine the suitable process parameters for obtaining their target surface quality. The ongoing deterioration of the polishing bath further complicates achieving personalized outcomes. To adopt the mass personalization trend, electropolishing operations need to implement approaches that ensure consistent polishing results while enabling the rapid identification of optimal process parameters for achieving personalized surface finishes.
The goal of this work is to develop a prediction tool to determine the suitable polishing parameters in a deteriorating polishing bath to attain the target surface quality. The study uses surface roughness and brightness measurements, as well as SEM and laser confocal microscopy images of the parts to investigate the effect of bath conditions on the final surface finish of electropolished parts. Multiple explanations are suggested to justify the compromised performance of heavily-used baths. The study also proposes a current-voltage plot that can be used as a tool to determine the polishing voltages associated with the emergence of some surface defects under varying bath conditions.
A substantial dataset is next compiled from electropolishing experiments conducted under different bath states and using various polishing parameters. This dataset is later subjected to several machine learning algorithms to determine the model that best represents the data. The Random Forests model is selected as the foundation for the target prediction tool, which demonstrates excellent performance on both the dataset and previously unseen scenarios. The constructed tools offer a comprehensive perspective on electropolishing outcomes across different bath and part conditions and polishing process parameters.
Moreover, by facilitating the attainment of the desired surface finish even in heavily-utilized electrolytes, the developed tools diminish the need for frequent electrolyte replenishment and the expensive disposal of hazardous waste, thereby benefiting the environment.