Process Parameters Analysis and Artificial Neural Network Prediction for Color Variation after Laser Induced Ti-6Al-4V Surface Coloring

Abstract

Metal surface coloring process is considered as a process with strong technical influence because its’ advantage in improving the appearance of metal, identifying metal products with information, preventing corrosion of metal substrate, excellent color stability and other characteristics. Laser induced metal coloring is expected to become a metal coloring process with stability and efficiency for its quickly response in producing bright colors on the metal matrix by simply setting the process parameters. The process is simple and with excellent efficiency and accuracy. At present, in the field of laser induced coloring, the main method is to find the corresponding relationship between laser energy density and color formation with a large number of laser parameters, the main limit in this method comes from lacking convenience to reproduce color s within costing a great deal of time. How to establish the nonlinear relationship between metal coloring process and specific color formation is still a challenge. Here, we observed thousands of laser-induced Ti-6Al-4V coloring test data in efficient way, based on the color evaluation index of CIE1976 color system and the theory of Multilayer Perceptron algorithm. It is proved that the color produced by the laser is highly matched with the color predicted by training of Multi-layer Perceptron algorithm. The precision difference brought by different algorithm strategies is compared, attempted to find the algorithm strategy that is capable of predicting color formation with high accuracy. This study provides a feasible method for the application of laser coloring to the industrial field.

Keywords: Laser-Induced Metal Oxidation, Artificial Neural Network, Laser Color Marking, Colorization