An Expert System for Life Prediction of Woven-roving GFRE Closed end ThickTube Subjected to Combined Bending Moments and Internal hydrostatic Pressure Using (ANN)

Abstract

As the study of fatigue failure of composite materials needs a large number of experiments as wellas long time, so there is a need for new computationaltechnique to expand the spectrum of the results and tosave time. The present work represents a new techniqueto predict the fatigue life of Woven Roving Glass FiberReinforced Epoxy (GFRE) subjected to combinedcompletely reversed bending moments and internalhydrostatic pressure, at different pressure ratios(Pr), PPrr= 00, 00.2222, 00. 55, 00. 7777 (i.e. pressures amountingto 0%, 25%, 50% and 75% of the burst pressure). Twofiber orientations, [0o,90o]3s and [±45o]3s are considered.Two neural network structures, feed-forward (FFNN)and generalized regression (GRNN), are applied,trained and tested. The groups of data considered, arethe maximum stress and the Pressure ratio with thefiber orientation. On the other hand, more accurateprediction method is obtained by using a useful expertsystem which is designed to aid the designer to decidewhether his suggested data for the composite structureis suitable or not. In this expert system a neuralnetwork is designed to consider the design data as inputand to get yes or no as output. The results showimprovement when using the one input life (N) of themaximum stress (σmax) and the pressure ratio (Pr). Alsothe feed-forward neural network shows better resultsthan that given by the generalized regression network.The designed expert system helped the designer witha100% correct conclusions about his decision of thecombination of the proposed data.