Chaotic neurons for on-line quality control in manufacturing
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
Given today's trend towards globalisation of markets, on-line quality control of manufacturing processes is deemed essential. We describe the use of neural networks and chaos theory to implement the idea of intelligent integrated diagnostics (IID) for this purpose. Our efforts are specifically concentrated on implementing HD in the turning process - a ubiquitous manufacturing process. We propose and develop two types of chaotic neurons - neural network architectures trained to capture the underlying chaotic dynamics of the turning process - to address the common problems of tool wear and chatter. The first, called the fractal estimation continuously estimates tool wear; the second, called the COPAVAS, initiates optimal chatter control.