Abstract
The wear resistance of magnesium alloys is one of its key technological properties that could limit their practical application. In accordance with ASTM G99–95a standard, this study used a pin-on-disc method to analyze the wear behavior of ascast AZ31 magnesium alloy under dry-sliding conditions. With a track radius of 37.5 mm, varied sliding velocities of 0.25, 0.5, 1, and 1.5 m/s, and normal loads of 40, 60, and 90 N were employed to quantify wear rate over a fixed sliding distance of 600 m. The surface morphology of the alloy's corroded surface was investigated using a SEM/EDS. The effectiveness of two Evolutionary Computing integrated machine learning algorithms, Particle Swarm Optimization coupled Decision Tree (PSO-DT) and Particle Swarm Optimization coupled Gradient Boosting Regressor (PSO-GBR), is also compared in this study in predicting the particular wear rate of AZ31 magnesium alloy. The experimental observations of wear behavior at various sliding velocities and normal loads make up the dataset used in this study. The algorithms' prediction performance was assessed using the coefficient of determination (R2), mean squared error (MSE), and mean absolute error (MAE). The results show that when it comes to foretelling the precise wear rate of AZ31 magnesium alloy, the PSO-GBR algorithm works better than the PSO-DT algorithm resulting in the R2 value of 0.99970. The PSO-GBR algorithm's successful integration of Particle Swarm Optimization and the gradient-boosting regressor model is responsible for this higher performance. The PSO-GBR algorithm improved accuracy and better captured nonlinear patterns in the data by improving the algorithm's parameters and capturing complex wear mechanisms.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call