Inverse modelling for optimal metal design using fuzzy specified multi-objective fitness functions

This paper presents the results which relate to the development and application of evolutionary multi-objective optimisation algorithms for the design of alloy steels using crisp as well as fuzzy logic based objective functions. The applied optimisation algorithms aim at determining the optimal heat treatment regime(s) and the required weight percentages for the chemical composites to obtain the pre-defined mechanical properties of steels, such as the ultimate tensile strength (UTS) or better known as tensile strength (TS) and elongation (ELO). During this process, the targeted mechanical properties and the reliability of their predictions are considered simultaneously in the above objective functions. Results show that for the multi-objective case the use of fuzzy logic based functions, as opposed to crisp ones, is beneficial especially when the application can tolerate a relatively low degree of discrimination between the so-called multi-objective ‘pareto’ solutions.