Intelligent tool condition monitoring in high-speed turning of Titanium Ti-6Al-4V alloy

Abstract

Intelligent Tool Condition Monitoring (TCM) is an essential requirement in the drive towards automated machining operations. In this paper, a Multi-Layered Perceptron (MLP) neural network model has been developed for on-line condition monitoring of tool wear in high-speed turning of Titanium-based alloy (Ti-6Al-4V). Machining trials were conducted for typical rough and finish turning operations with cutting speed (90 – 120 m/min), feed rate (0.15 – 0.2 mm/rev), and depth of cut (0.5 -2.0 mm) using uncoated cemented carbide (K10 grade) inserts with Inter-national Standard Organization (ISO) designation “CNMG 120412”. The tool maximum flank wear (VBmax), cutting forces (feed force, Fx, and tangential force, Fz), and spindle motor power were measured during each machining operation. The cutting parameters (cutting speed, feed rate, and depth of cut), and cutting force and spindle power were used in isolation or in combination as input dataset in training the neural network to predict wear land on cutting tool at different stages of wear propagation (light, medium and heavy). The neural network model was designed using Matlab® neural toolbox. Accuracy of model for the prediction of tool wear at different wear stages were evaluated based on the Percentage Error (PE) for both roughing and finishing operations. Results showed that, the heavy wear stage (PE = ±5%) was predicted more accurately compared to the light (PE = +5 to -10%) and medium (PE = +25 to -30%) wear stages. The combination of the force, power signals and cutting parameters improved performance of the model.