Designing age-hardenable Al alloys using ANFIS and GA

Abstract

Age-hardenable aluminium alloys have higher strength compared non-heat-treatable alloys due to strengthening by the precipitates. Increasing the toughness of such alloys require improved balance between strength and ductility. To improve both ductility and strength through modification of composition and heat-treatment parameters of age-hardenable Al alloys (2XXX, 6XXX and 7XXX) an artificial intelligence based computational design approach is employed. Published data on the alloys are used for developing the data-driven models for tensile strength, yield strength, and %elongation. Fuzzy C means clustering is used to cluster the variables in the database into different levels and to generate fuzzy rule correlating those variables. Adaptive neuro-fuzzy inference system (ANFIS) used the fuzzy rules to develop the data-driven fuzzy predictive models for the said properties of the alloys. This models in turn played the role of objective functions for the multi-objective optimization using genetic algorithm (GA) for handling the conflicting objectives of improving ductility as well as strength to design alloys. The generated Pareto solutions are analyzed for finding suitable composition and process parameters fulfilling the purpose.