MODEL-BASED CONTROL OF VOIDS AND PRODUCT THICKNESS DURING AUTOCLAVE CURING OF CARBON/EPOXY COMPOSITE LAMINATES

In this article we present a novel feedback control strategy for the autoclave curing of carbon/epoxy laminates. The strategy utilizes an on-line model of the process to control the quality by adjusting the cure cycle. On line measurements are used to correct for modeling errors. The quality variables of interest here are the thickness of the product laminate and the maximum void size present in the laminate. Artificial neural networks are utilized to regress past operational data to generate nonlinear models for predicting the product quality. These models utilize properties of the raw material prepreg and secondary measurements taken while the process cycle is in progress. Some results of a control study done using a simulated model of a process for autoclaving a carbon fiber/epoxy resin system are also reported. The main contribution of this paper is the successful demonstration of a neural network model based feedback control strategy for autoclave curing. While discussion in this paper is Limited to control of the autoclave process, the methodology suggested is also equally applicable to other composite manufacturing processes.