Optimization and prediction of AZ91D stellite-6 coated magnesium alloy using Box Behnken design and hybrid deep belief network

Abstract

Nowadays, the utilization of magnesium (Mg) alloys is increased due to its various beneficial characters such as better thermal conductivity, minimum density, less weight, superior strength, good dimensional, stability nature, etc. They are used in both mechanical as well as electrical industries. In the same way, it is used in electronic industries for its perfect capability of damping and electromagnetic shielding offers noise reduction. Although, they have a lot of advantages some disadvantages surround them. Mg alloy easily corroded when used at certain wet zones and the poor wear behavior may suppress the service life of Mg alloys. To overcome such issues, a surface coating approach is to be carried out on Mg alloy AZ91D using stellite powder to assist with NiCr pre-coating. The coated Mg alloys are studied for their porosity (%), corrosion rate (mm/year), and wear loss (mg). The experimental study is planned by Response Surface Methodology (RSM) using Box Behnken Design (BBD). The outcomes are validated by Hybrid Deep Belief Network (DBN) based Mayfly optimization approach in MATLAB version 2020a. The minimum porosity, corrosion rate and wear loss obtained for the coated AZ91D Mg alloy is 0.588%, 1.302 mm/year, and 0.719 mg. Besides, the hybrid DBN-based predicted outcomes are closer to the experimental outcomes.

Keywords

• AZ91D
• Stellite
• Response surface methodology (RSM)
• Box Behnken design (BBD)
• Hybrid deep belief network (DBN) based mayfly optimization