Extreme learning machine and support vector regression wear loss predictions for magnesium alloys coated using various spray coating methods

Abstract

Magnesium alloys are popular in the aerospace and automotive industries due to their light weights and high specific strengths. The major disadvantages of magnesium alloys are their weak wear and corrosion resistances. Surface coating is one of the most efficient methods of making material surfaces resistant to wear. Experimental determination of wear loss is expensive and time-consuming. These disadvantages can be eliminated by using machine learning algorithms to predict wear loss. This study used experimentally obtained wear loss data for AZ91D magnesium alloy samples coated via two different spray coating methods (plasma and high velocity oxy-fuel spraying) using various parameters. Support vector regression (SVR) and extreme learning machine (ELM) methods were used to predict wear loss quantities. In models tested using 10-k cross-validation, R2 was calculated as 0.9601 and 0.9901 when the SVR and ELM methods were applied, respectively. The ELM method was more successful than SVR. Thus, the ELM method has excellent potential to support the production of wear-resistant parts for various applications via spray coating.