Analysis of surface defects on metal coatings by artificial intelligence methods

Metallic coatings are commonly used to enhance the physical properties and resist corrosion of metal materials. This is especially critical in military and defense industries, where the equipment such as tanks, aircrafts, and weapons must be strong, heat-resistant, and able to withstand wear. Different coatings are applied based on their intended purpose, with decorative coatings prioritizing appearance, while functional coatings focus on hardness and wear resistance. Zinc coatings can serve both purposes. For corrosion protection, the coating must be thick and include a passivation layer. Decorative coatings aim to have a bright surface. The brightness and thickness of the coating can be influenced by organic additives in the coating bath. However, it is challenging to control the quantity of these additives as they cannot be measured by traditional methods and their composition can change due to reduction and oxidation reactions. In this study, AI methods were used to examine the relationship between surface defects and organic additives in the coating bath. First, the Mask RCNN algorithm was used to classify the types of defects on the surface. Then, ML algorithms were applied to determine the relationship between the surface defects and organic additives, and RF was achieved the highest accuracy. Therefore, the RF model is used as the objective function in genetic algorithm. Finally, the NSGA-II genetic algorithm was used to optimize the quantity of organic additives in the coating bath, taking into account that the optimal values can change as the coating bath ages.