Designing dual-phase steels with improved performance using ANN and GA in tandem

In this study, artificial neural network (ANN) and multi-objective genetic algorithm (GA) are employed in tandem to design dual-phase (DP) steel with improved performance. Six different mechanical properties are modeled and optimized for simultaneous enhancement of strength and ductility. The existing database on DP steels is utilized to create ANN based models for yield strength, tensile strength, uniform elongation, total elongation, yield ratio and strain hardening exponent considering thirteen input variables related to composition and processing parameters. The developed models are employed to achieve better understanding of complex correlations of composition-processing-property, and are also used as objective functions for the multi-objective optimization of conflicting properties by GA. The generated Pareto solutions are explored for concurrent improvement of both strength and ductility, and finally, the optimum composition and process variables of DP steels are identified. The special roles of some alloying elements and heat treatment schedule are recognized for further exploration in achieving improved performance of DP steels.