The optimization of nickel electroplating process parameters with artificial intelligence methods

The scope of this study is to estimate the composition of the nickel electrodeposition bath using artificial intelligence method and optimize the organic additives in the electroplating bath via NSGA-II (Non-dominated Sorting Genetic Algorithm) optimization algorithm. Mask RCNN algorithm was used to classify the coated hull-cell panels based on their appearance. The coated plate was divided into 5 classes as “Full Bright”, “Full Fail”, “HCD Fail”, “LCD Fail” and “Metallic impurity”. The intersection over union (IoU) values of the Mask RCNN model was ascertained between 89 and 98%. Machine learning (ML) algorithms, MLP, SVR, XGBoost, GP and RF, were trained using the images classified by the Mask RCNN. The additives in the electrodeposition bath and operating parameters were specified as an input and the classes of the coated panels as an output in the ML study. Among the trained models, RF model imparted the highest F1 scores for all classes. The F1 scores of RF model for” Full Bright”, “Full Fail”, “HCD Fail”, “LCD Fail” and “Metallic impurity” are 0.99, 1.00, 1.00, 1.00 and 0.94 respectively. NSGA-II genetic algorithm was employed to optimize the content of the bath. The trained RF model was defined as the objective function in the optimization process. The organic additives and operating parameters for full bright coating surface were optimized and the direction and importance of features (factors) impacting the output were delineated. Graphical abstract