A Novel Approach to Estimate the Radius Ratio of an Elastic Tube from Its Form Function Using Soft Computing Techniques

Abstract

A number of researches in different areas has been conducted on using the Principal Component Analysis PCA combined with Artificial Neural Networks ANN. In this current research, we use these two techniques to estimate the radius ratio b/a (a: outer radius, b: inner radius) of an elastic tube based on its form function. The PCA technique is used to estimate the component loadings corresponding to form function. Then, the component loadings it's used in the ANN technique to estimate the radius ratio of the tube. To get the optimal network, several configurations are implemented and tested. The optimal configuration selected is a network with 16 inputs, 2 hidden layers composed of 4 and 1 neurons respectively, and trained by the back-propagation algorithm. This configuration is able to estimate the radius ratio �� /�� with a mean absolute error MAE of about 0.0024 and a mean square error MSE of 0.0008. This study reveals benefits of the combination between PCA and ANN, and also it provides some new ideas for further researches. This current work can be used as a novel approach for the characterization of an elastic tube.