Skip to Main Content
This paper reports on an efficient algorithm for locating the `optimal' solutions for multi-objective optimization problems by combining a state-of-the-art optimizer with a fitness model-estimate. This hybrid framework is introduced to illustrate how to make sufficient use of an approximate model, which includes a `controlled' process and an `uncontrolled' process during the search. With the inclusion of such approximate model in the optimization block, a global reseeding strategy based on previous data is also applied to improve the ability of the multi-objective optimizer to find global set of solutions (`pareto' solutions). To this effect, the popular algorithm, NSGA-II, and a Multi-Layer Perceptron Neural Network (MLP) are combined synergetically to show details of such processing. Furthermore, a simple (but no simpler) method for selecting the `training' data necessary for eliciting the fitness landscape model is suggested to address what are now a common engineering problems, in particular those associated with sparse data distributions and objectives converging at significantly different speeds. To test the validity of the proposed multi-objective scheme, a series of simulation experiments, using well-know benchmark functions, are conducted and are compared to those carried-out while using the original NSGA-II and SPEA-2, under similar conditions. The proposed method is also applied to the `optimal' design of alloy steels in terms of chemical compositions and processing conditions and is shown to perform very well.