Prediction and optimization of fireproofing properties of intumescent flame retardant coatings using artificial intelligence techniques

A inulti-structured architecture of artificial intelligence techniques including artificial neural network (ANN), adaptive neuro-fuzzy-inference-system (ANFIS) and genetic algorithm (GA) were developed to predict and optimize the fireproofing properties of a model intumescent flame retardant coating including ammonium polyphosphate, pentaerythritol, melamine, thermoplastic acrylic resin and liquid hydrocarbon resin. By implementing ANN on heat insulation results of coating samples, prepared based on a L-16 orthogonal array, mean fireproofing time (MFPT) values were properly predicted. The predicted data were then proved to be valid through performing closeness examinations on fuzzy inference systems results regarding their experimental counterparts. However, the possible deviations tapped into phenomena like foam detachment and char cracking were alleviated by ANFIS modeling embedded with pertinent fuzzy rules based on the sole and associative practical role of used additives. The contribution of each intumescent coating component on the formulation with optimized fireproofing behavior was then explored using GA modeling. A similar optimization procedure was also conducted using conventional Taguchi experimental design but the GA based optimized intumescent coating was found to exhibit higher MFPT value than that suggested by the Taguchi method. (C) 2013 Elsevier Ltd. All rights reserved.