Fuzzy logic model for prediction of fusion characteristics in rigid polyvinyl chloride nanocomposite

Abstract

Fusion is a key parameter in achieving polyvinyl chloride (PVC) nanocomposites with desired properties. In the present research, a fuzzy logic (FL)-based model is developed to predict fusion time (FT) for the contents of nanoclay, processing aid, and calcium stearate in PVC processing. In order to have precise rules for the FL model, data mining algorithm RepTree is employed to detect dominating patterns among experimental data. The model parameters are then well adjusted using genetic algorithm. The modeling results show a correlation of 0.86 between predicted and observed values for FT. So, it proved reliability of the idea of employing the decision tree resulted from a data mining algorithm as the base knowledge of FL models. Also, applying genetic algorithm optimization, the correlation coefficient increased from a value of less than 0.83 to 0.86. The calculated correlation coefficient for the test data was 0.88, which denotes good model universalizing ability.