1. Characterizing surface roughness of β-titanium alloy machined using carbide tool by machine learning approaches
- Author
-
Abdul Md Ma, Tasadduq Imam, Kazi Badrul Ahsan, and Neamul Ahsan Noman Khandoker
- Abstract
By virtue of its high strength-to-weight ratio, preserving its toughness and other mechanical strength properties at high temperature, low wear, and corrosion rates even in eroding environment, and non-reactivity with human body fluid, the titanium α-β alloy Ti-6Al-4V is a supreme material for producing responsible machine parts. However, machining of such and other Ti-alloy based materials is difficult especially due to lower thermal conductivity, unprecedented high heat generation in cutting zone, chemical reactivity of Ti molecules with cutting tool materials (producing built-up edge), further to notoriously low tool life, low molecular density, and poor surface integrity sustained during machining. Development of optimum machining regimes is, hence, important to reduce machining time and consequently the production cost while maintaining product quality and surface integrity. A series of real-life machining investigations, using coated and uncoated carbide inserts, have been carried out to characterise the optimum ranges of individual cutting parameters (v, f, d) with respect of obtained surface roughness (Ra). While machining literature generally focuses on varying specific parameters keeping other parameter constant to characterise the impact of cutting parameters on surface roughness, we in this article process the machining data using machine learning (ML) approaches (Support Vector Machine, Random Forest, and Deep Learning) to characterise surface roughness with respect to varying all three cutting parameters together. Thus, this work makes an important contribution to the current machining literature. The achieved outcomes will facilitate sustainability, economic efficiencies and machining quality in future research and the manufacturing process of Ti alloy products.
- Published
- 2022
- Full Text
- View/download PDF