Data‐Driven Pareto Optimization for Microalloyed Steels Using Genetic Algorithms

Abstract

A data base was put together for the mechanical properties of microalloyed steels, which contained about 800 entries for ultimate tensile strength (UTS), yield strength (YS), and elongation. Using an evolutionary neural network, based upon a predator–prey genetic algorithms of bi‐objective type, this information was used to construct data‐driven models for UTS, YS, and elongation. The optimum Pareto tradeoffs between these properties were obtained using a multi‐objective genetic algorithm. The results led to some hitherto unexplored steel compositions with optimum properties. Some such steels were actually cast and the experimentally observed property values were found to be well in accord with the predicted results.

Critical Assessment 3: The unique contributions of multi-objective evolutionary and genetic algorithms in materials research

“…20 Several subsequent publications have dealt with diverse materials including iron and steel, 21,22 residual stress minimisation during machining, 34 semiconductor solar cell design, 35 iron ore sintering, 36 etc. have come under its purview, in addition to materials studies of more fundamental nature, 37 providing vital insights to materials design, fabrication and processing.…”

“…[42][43][44] Such meta-models enabled prediction of newer steels with superior and optimised properties, which were also verified experimentally. 22 It made it possible accurately to predict the yield point in steel, through calculation of the Hill's coefficients, 45 without making customary ad hoc assumptions regarding the strain offset level. 37 Such calculations were not possible previously but are feasible now, simply because in recent years the evolutionary multi-objective optimisation algorithms have become mature and advanced.…”

Evolutionary Data-Driven Modeling

“…In principle, any multi-objective optimization routine can be used in EvoNN; the very first paper on this algorithm , however, was based upon a modified predatorÀprey genetic algorithm (Li, 2003) and the practice subsequently continued (Bhattacharya et al, 2009;Govindan et al, 2010;Kumar et al, 2012). In this algorithm two species, the predators and the prey, are introduced in a torroidal computational grid that emulates a forest.…”

“…The bi-objective genetic algorithm-based modeling strategy, as discussed here, has been effectively tested for a number of problems in the materials domain that deal with noisy data (Bhattacharya et al, 2009;Kumar et al, 2012). A series of recent analyses of the FeÀZn system is presented as a paradigm case.…”

Evolutionary Design of Nickel-Based Superalloys Using Data-Driven Genetic Algorithms and Related Strategies

“…GAs were combined with molecular dynamics [8] and phase diagram information [9] in other studies dealing with similar designs. For relevant alloy systems, steel, for example, it was clearly demonstrated [10] how hitherto unknown alloys with superior properties could be designed using evolutionary data-driven models. In the present work, we will extend this approach to Ni-based superalloys, further strengthening it with a number of emerging evolutionary paradigms and supporting thermodynamic calculations.…”