Development of an artificial neural network for forecasting energy absorption capability of thermoplastic commingled composites: processing, characterization, and crashworthiness

Abstract

The energy absorption capability of structural thermoplastic composites is governed by the properties of the polymer matrix and the processing parameters. However, previous studies have been investigated the effect of thermal degradation kinetics on mechanical behavior of the material and its crashworthiness. Understanding the kinetics of thermo-oxidative degradation as well as temperature and time thresholds allows the manufacture of high-performance composite components with greater impact resistance. This fact increases the passenger's safety in an automobile collision. In addition, there is currently a scientific interest regarding the characterization and the processing of thermoplastic commingled composites. This material combines reinforcing fibers and yarns of polymer matrix in the same tow or fabric. The commingled technology portrays possibilities for manufacturing high performance structural composites, aiming for high productivity, reduction of time and operational costs. The objective of this study is to develop an integrated modeling, calculation, prediction and process optimization of thermoplastic composite materials by an artificial neural network (RNA) trained by inputs of (i) matrix thermal properties, (ii) viscoelastic properties of the polymer, (iii) thermo-oxidative degradation kinetics and (iv) processing parameters. This project is associated to a FAPESP regular proposal (nº 2017/16970-0), under responsibility of the supervisor of this proposal. (AU)