Research

Intelligent Texture Tool for Micro-Surface Engineering

Project Details

Abstract

The research investigates the application of Ultrasonic Vibration-Assisted Turning (UVAT) in a radial direction. It examines the mechanisms and experimental setups through vibrating micromachining driven by a piezo-actuated tool. Previous literature has demonstrated that UVAT is effective by improving the surface quality, reducing cutting forces, reducing heat, extending the tool life, and several other benefits. This research gives an overall overview of the process of manufacturing the piezoelectric tool, preparing the mechanical and electrical experimental set-up to run these types of tools, and testing it on surfaces. Moreover, this research focuses on understanding the nuances of the power supply, wiring, connections, and electrical configurations. After running the experiment and generating micro-textures, artificial intelligence will be employed to detect and categorize these features. The creation of micro-textures and achieving precise accuracy on the micro-scale has become primordial for the functioning of modern devices like sensors, semi-conductors, and silicon chips. And so, Convolutional Neural Networks (CNNs), specifically GoogleNet and Resnet-50, will be used to pinpoint the most suitable model in the industries. An experimental comparison will take place between both pre-trained neural networks to determine the accuracy, processing speed, and adaptability to changes in the dimple geometry. These various micro-texture geometries were simulated and trained to test the CNNs when detecting different shapes and textures. In future studies, the experimental results will be compared to these simulated results.