Analysis of prediction capability for internal micro-channels fabrication in polycarbonate and PMMA


Keywords: , , , , , ,

This paper presents a 3^3 factorial Design of Experiment (DoE) and Artificial Neural Networks (ANN) for the prediction of two characteristics of the pulsed Nd:YVO4 laser machined internal micro-channels in Polycarbonate and PMMA glass. Power, P, pulse repetition frequency, PRF, and translation speed, U, were set as control parameters. These and the corresponding channel results were used to construct the DoE and ANN predictive models. The responses chosen were the width and process cost for micro-channel fabrication. Two multi-layered feed-forward, back-propagation ANN models with different training data were generated within LabVIEW code. The prediction results of both the ANN models and the DoE models formed with the same input data were compared in terms of absolute prediction error. The ANN model showed better predictive capability within the examined range over the DoE model.

PDF of paper:

Journal: TechConnect Briefs
Volume: 2, Nanotechnology 2010: Electronics, Devices, Fabrication, MEMS, Fluidics and Computational
Published: June 21, 2010
Pages: 492 - 495
Industry sector: Sensors, MEMS, Electronics
Topic: Micro & Bio Fluidics, Lab-on-Chip
ISBN: 978-1-4398-3402-2