ABSTRACT
The issues of modeling, simulation and optimisation for materials systems using different computational intelligence (CI) tools were discussed in Chapters 4 through 6. In Chapter 7 the issue of optimisation was dealt with using evolutionary algorithm tools. It was also emphasised that optimisation tools play a pivotal role in materials design, as they provide the solutions, in the form of composition or processing routes, to achieve the targeted performance of the materials. The prerequisite for using any optimisation tool for design purposes is one or more dependable objective function(s) mapping the input-output space of the materials system, which is nothing but good predictive model(s). As discussed in Chapters 1 and 2, for complex materials systems good models, developed following the fundamental science principles of the system, are sparse. The available models are, in many cases, not sufficiently capable of mapping the independent variables of the system to its final performance, which is most essential for a successful design of materials. Sometimes the system is assumed to be simple for the sake of modelling it, which does not have much practical use. In such cases data-driven models or imprecise knowledge-driven models using CI techniques can be utilised for modelling and subsequently as objective functions for optimisation. Thus artificial neural network (ANN) or fuzzy models could be utilised as objective functions for optimisation using evolutionary algorithms or other CI-based optimisation techniques. In this chapter such applications of more than one CI tool in tandem are discussed. Examples are provided of ANN models and fuzzy inference systems (FISs) as objective functions and the Genetic Algorithm (GA) as an optimisation tool in both single-objective and multiobjective modes.
