Use of artificial neural network to fit creep behavior of polyetherimide/carbon fiber composite under low-stress load

Author(s)

Ornaghi Jr, Heitor Luiz ; Monticeli, Francisco Maciel ; dos Reis, Ana Karoline ; Neves, Roberta Motta ; Santos, Luis Felipe de Paula ; Botelho, Edson Cocchieri

• Document type: Journal article
• Source: POLYMER BULLETIN; v. N/A, p. 12-pg., 2023-08-07.

Abstract

High-performance composites are subjected to different mechanical forces and stresses during their lifetime. Hence, knowledge of the time-dependent deformation at different temperatures is crucial for engineering applications. In this context, a polyetherimide/carbon fiber composite was analyzed via creep behavior from the glassy to the elastomeric state (30-250 & DEG;C). Creep tests were carried out under low-stress loads. An artificial neural network (ANN) was used to fit the curves for all tested temperatures. The ANN allowed excellent fit for all conditions independently of the curve shape, mainly for the glassy region, in which most of the analytical models did not fit well. This study is a growing and promising field in the materials science field due to the fact that it uses ANN as an alternative to fit creep behavior not fitted by usual analytical equations, which utilize closed-form solutions. Furthermore, it is expected that this study will allow the formulation of new equations and methods to determine the creep behavior at high stresses, for example. (AU)

• FAPESP's process: 17/16970-0 - Obtaining and Characterization of Nanostructured Thermoplastic Composites for Aeronautical Application
• Grantee: Edson Cocchieri Botelho
• Support Opportunities: Regular Research Grants
• FAPESP's process: 18/07867-3 - Processing and characterization of multifunctional thermoplastic composites reinforced with carbon fibers and buckypaper of carbon nanotubes
• Grantee: Luis Felipe de Paula Santos
• Support Opportunities: Scholarships in Brazil - Doctorate