Machine Learning Analysis of Ni/SiC Electrodeposition Using Association Rule Mining and Artificial Neural Network

Abstract

Due to their advanced tribological and mechanical properties, nickel/silicon carbide (Ni/SiC) composites have gained significant attention in recent years. Electrodeposition is a cost-effective method to produce the Ni/SiC composites with high uniformity. However, materials and process parameters of Ni/SiC electrodeposition have immense impact on the amount and the uniformity of the co-deposited SiC particles. In this study, machine learning algorithms are used to investigate the effect of electrodeposition parameters and materials on the Ni/SiC composite. Association rule mining (ARM) is used to determine the important factors leading to high SiC incorporation and artificial neural network (ANN) is used to build a model that can predict the amount of SiC particles in the deposit. ARM results clearly present that the use of cationic dispersants, especially AZTAB and TMAH, at concentrations higher than 1 g l−1 is highly beneficial for high SiC incorporation into the deposit. Moreover, the ANN model shows that the estimation of SiC vol.% in the composite is possible with high prediction accuracy; the RMSE values of the training and the testing set are calculated as 5.49 and 6.61, respectively. This thorough analysis confirms that machine learning is a highly effective method, especially for such well-defined systems with many parameters.