Rough-Fuzzy-GA-based design of Al alloys having superior cryogenic performance

Multi-objective genetic algorithm (GA) is employed for the optimal design of novel heat-treatable aluminum alloys with superior performance at cryogenic temperatures. Existing database on age-hardenable aluminum alloys is utilized to create a learning model. Composition and processing parameters of the alloys are considered as the inputs, whereas mechanical properties, viz. YS (Yield Strength), UTS (Ultimate Tensile Strength) and %Elongation tested at subzero temperatures, are used as the outputs, which characterize the performance of the alloy. Data-driven models are developed using the hybrid rough-fuzzy approach. While rough set brings out the most significant variables and formulates if-then rules to explain the relationships between inputs and outputs, fuzzy inference system (FIS) serves as the predictive model. Strength and ductility of the Al alloys at low temperature being conflicting in nature are simultaneously optimized using multi-objective GA to design alloys having an optimal blend of the two properties.