Optimization for CFRP pultrusion process based on genetic algorithm-neural network

The temperature and the degree of cure of carbon fiber reinforced polymer (CFRP) are coupled during pultrusion. In this paper, the governing equations for heat transfer and resin curing are solved by the combination of finite element method, finite different method and indirect decoupling method. The kinetic parameters needed for simulation are obtained from differential scanning calorimetry (DSC) measurement. The temperature of composites on real time during pultrusion is monitored by the fiber Bragg grating (FBG) sensor. And the final degree of cure of composites is also measured through Sorbitic extraction. It shows that the simulation procedure is effective and reliable and predicts temperature and degree of cure which are in good agreement with the experimental results. On the basis of the simulated results, the relationship between processing parameters and degree of cure is formed by artificial neural network. And genetic algorithm combines with the artificial neural network to solve the bi-objective optimization for CFRP pultrusion. It shows that there are considerable improvements in pull speeds and die temperatures after optimization by ANN and GA.

[1]  Jeff Martin,et al.  Pultruded composites compete with traditional construction materials , 2006 .

[2]  X. L. Liu,et al.  Heat transfer and cure analysis for the pultrusion of a fiberglass-vinyl ester I beam , 1999 .

[3]  Richard Stewart.  Pultrusion industry grows steadily in US , 2002 .

[4]  Xiang Ling,et al.  Optimal design approach for the plate-fin heat exchangers using neural networks cooperated with genetic algorithms , 2008 .

[5]  Hao Zhang,et al.  Temperature-independent FBG pressure sensor with high sensitivity , 2007 .

[6]  Ugur Kesgin,et al.  Genetic algorithm and artificial neural network for engine optimisation of efficiency and NOx emission , 2004 .

[7]  Sunil C. Joshi,et al.  Three-dimensional finite-element/nodal-control-volume simulation of the pultrusion process with temperature-dependent material properties including resin shrinkage , 2001 .

[8]  B.H.M. Sadeghi,et al.  A BP-neural network predictor model for plastic injection molding process , 2000 .

[9]  J. T. Liu,et al.  PREDICTION OF FLOW STRESS OF HIGH-SPEED STEEL DURING HOT DEFORMATION BY USING BP ARTIFICIAL NEURAL NETWORK , 2000 .

[10]  Yee Cheong Lam,et al.  Simulation of heat transfer and cure in pultrusion with a general-purpose finite element package , 2000 .

[11]  Amanda Jacob,et al.  Globalisation of the pultrusion industry , 2006 .

[12]  Yi He.  DSC and DEA studies of underfill curing kinetics , 2001 .

[13]  Sunil C. Joshi,et al.  Curing optimization for pultruded composite sections , 2002 .

[14]  Mohammad Safdar Baloch,et al.  Dynamic optimization of watering Satsuma mandarin using neural networks and genetic algorithms , 2007 .

[15]  Yanbiao Liao,et al.  Differential FBG sensor for temperature-compensated high-pressure (or displacement) measurement , 2004 .

[16]  J. A. Roux,et al.  Die and post-die temperature and cure in graphite/epoxy composites , 1996 .