Prediction of built-up edge formation in machining with round edge and sharp tools using a neural network approach

Abstract

Built-up edge (BUE) formation in machining has a profound effect on the cutting forces and vibrations, the quality of machined surfaces, etc. Prediction of BUE formation is important for machining optimisation and tool condition monitoring. This article presents a neural network approach to predicting BUE formation in orthogonal machining of 2024-T351 aluminium alloy with round edge and sharp tools. Extensive cutting experiments within and beyond the range of BUE formation were conducted. The cutting forces and vibrations were measured. The experimental data were employed to develop the Resource Allocation Network (RAN) models and the Multilayer Perceptron (MLP) network models for round edge and sharp tools. The inputs to the models include the cutting speed, the feed rate, the cutting force, the thrust force and the vibration amplitude. The output is a 3-bit binary code that represents the three BUE states corresponding to different cutting conditions. The results show that the overall prediction accuracy of the RAN models is 4.5% higher than that of the MLP models for round edge and sharp tools. Not only do the RAN models learn faster, but they also make a more accurate prediction of BUE formation than do the MLP models.

Keywords:

• machining processes
• neural network applications
• cutting tools
• built-up edge formation
• cutting forces and vibrations

Acknowledgements

The authors thank Dr. T.N Nagabhushana, Professor and Head of the Department of Information Science and Engineering, S.J College of Engineering, Mysore, India, in helping with the C++ computer coding of neural network models developed in this work.