Supervised classification model for estimation of wear in sisal fibre-epoxy composites

Abstract

The paper models the high stress abrasive wear behaviour of unidirectional epoxy composite reinforced with sisal fibre through supervised classification model. Rigorous experimentation initially involved the making of composites using polysulphide modified epoxy resin with varied sisal fibre concentrations and orientation (longitudinal, traverse, and normal). Later the abrasive wear behaviour of the composite fibre under various functional parameters was studied and recorded. A supervised classification model was then built to describe the effect of the various functional parameters on the abrasive wear behaviour of sisal fibre reinforced epoxy composite. The paper employs classification through decision tree and sequential covering rule induction. The results successfully show that decision tree classification has high ability to model the nonlinear dependence of the wear behaviour on the parameters describing the characteristics of the materials and conditions during the experimental work (like sliding distance, fibre weight, grit size and applied load). Once in operation the model can then be used for prediction of the wear behaviour in place of analytical investigations, thus reducing complexity and effort. Predicted values obtained through model testing were in good agreement with the experimental data.