Revisiting Optical Scattering with Machine Learning (SPARKLE)
This collaboration with the University of Nottingham concerns the development of optical instrumentation to determine the surface quality of the machined parts during the manufacturing process.
The surface topography of a component part can have a profound effect on the function of the part. It is estimated that surface effects cause 10% of manufactured parts to fail. In tribology, it is the surface interactions that influence such quantities as friction, wear and the lifetime of a component. In fluid dynamics, it is the surface that determines how fluids flow and it affects such properties as aerodynamic lift, therefore, influencing efficiency and fuel consumption of aircraft.
The overall aim of this project is to demonstrate a simple and cost-effective measurement system using artificial intelligence (AI) to classify a machined surface according to its scattered light signature.
In this collaborative project we are providing a rigorous vector solution to surface scattering that can model sensor output and provide an accurate and complete data set for AI training purposes.
This project will deliver instrumentation that can be cited on machine tools and can be used to control the manufacturing processes and optimise surface form and finish. Perhaps the greatest impact of this technology will be in the field of Additive Manufacturing where materials and processes are developing rapidly.
Our code is developed in MATLAB running on high-specification (25 Core, 256GB) computer servers within our Optical Laboratories in the School of Mechanical, Electrical and Manufacturing Engineering. We also expect to develop code that exploits the high-performance GPUs that are also available on these machines.
Our solution expresses the vector fields describing the scattered (and transmitted) fields as an expansion of dipole sources placed at the boundary surface. To calculate the strength of these sources we apply the extinction theorem of Ewald and Oseen at points placed a small distance inside and outside of the boundary surface according to the Nyquist sampling criteria. The method that we call 3sBSM properly accounts for polarisation, multiple scattering and surface plasmon resonances when they occur. The method and its relationship to Huygens’ principle is explained in our recent article.
Professor Jeremy Coupland, Professor of Applied Optics and Associate Dean (Research)
“We proved that by using Ewald and Oseen extinction theorem, a 3D vector formulation of Huygens’ principle can be used to calculate light scattered from the surface, without any significant mathematical complication.”