Optimal Design of Filament Winding Using Neural Network Experimental Design Scheme

Abstract

The goal of this paper is to design the filament winding in composites materials based on our previous developed technique. The most significant parameters in design and construction of composites prepared by filament winding are resin temperature, fiber tension and winding angle. The three variables depict nonlinear relationship; thus a nonlinear modeling technique is required. The proposed methodology enjoys many of the advantages claimed for the artificial neural network (ANN), random search optimization, fuzzy classification and information theory for the sequential design filament winding. The neural network is used to construct a model based on the currently available experimental data. Random search generates a number of candidates of the next batch of experiments. Fuzzy classification and information analysis are defined to balance the need of better classification and the relevance of each class in optimization. The test results of the proposed method show that the abilities of the proposed methodology handle multivariable experimental design and also help experimenters discuss complex trade-offs between practical limitation and statistical preferences in the experiment.