Pareto Frontier Exploration of Adaptive Systems Using the Selection-Integrated Optimization (SIO) Methodology

Abstract

Recently, we proposed the Selection-Integrated Optimization (SIO) methodology for designing adaptive systems. An important feature of the SIO methodology is that it integrates the two key processes in the adaptive system optimization: (1) the selection of the adaptive and fixed design variables, and (2) the optimization of the adaptive system, thereby leading to an optimum design. This is a significant departure from most existing approaches that separate those two key processes, which is likely to introduce a significant source of sub-optimality. In the previous SIO methodology formulations, we had combined the two design objectives involved in the SIO methodology, performance and penalty, in one aggregate objective function (AOF). In this paper, we develop the Pareto frontier exploration technique for adaptive systems using the SIO methodology. Exploring Pareto frontiers of adaptive systems is particularly challenging because the optimal sets of fixed and adaptive variables may change along the frontier. This change is likely to make the Pareto frontier discontinuous in nature. In this paper, we develop a SIO methodology base Pareto Exploration (SIO-PE) method that allows the designer to effectively explore the Pareto frontiers of adaptive systems. In particular, the SIO-PE method integrates the Normal Constraint (NC) Pareto exploration strategy in the SIO methodology. We show the application of the SIO-PE method through a case study of an adaptive ten-bar truss.