Reliability-based optimization for multiple constraints with evolutionary algorithms

Abstract

Abstract-In this paper, we combine reliability-based optimization with a multi-objective evolutionary algorithm for handling uncertainty in decision variables and parameters. This work is an extension to a previous study by the second author and his research group to more accurately compute a multiconstraint reliability. This means that the overall reliability of a solution regarding all constraints is examined, instead of a reliability computation of only one critical constraint. First, we present a brief introduction into this so-called 'structural reliability' aspects. Thereafter, we introduce a method for identifying inactive constraints according to the reliability evaluation. With this method, we show that with less number of constraint evaluations, an identical solution can be achieved. Furthermore, we apply our approach to a number of problems including a real-world car side impact design problem to illustrate our method.

Novel frameworks for creating robust multi-objective benchmark problems

“…For instance, there are specific sets of constrained test functions for constrained multi-objective meta-heuristics [19,48,53,55], reliability-based optimization [10,17,18], and dynamic multi-objective test problems [1,9,25,26,33,36,38,41]. In the field of robust multi-objective optimization, however, there is little in the literature on the development of robust multi-objective problems.…”

Perception-based evolutionary optimization: Outline of a novel approach to optimization and problem solving

“…Someone may consider PEO as a method for uncertainty handling in optimization such as the work of Deb et al in [13][14] that proposed an evolutionary framework for reliability-based optimization in which uncertainties in design variables and problem parameters are considered in optimization and a reliable optimal solution is achieved. The main differences between PEO and other works like the one of Deb [13][14] are as follows. First, PEO is inherently granular and perception-based even for certain problems where there is no uncertainty in decision variables and parameters.…”

Hybrid particle swarm algorithm in the application of optical fiber sensor with ZnO coating

“…The reflection coefficient of it is an important parameter for measuring the device quality, which is a multi-dimensional constraint function. At present, a lot of optimization algorithms can be used for multi-constraint function, such as genetic algorithm, heuristic algorithm, evolutionary algorithm etc [3][4][5], among which genetic algorithm has been used more often. However, the genetic algorithm has some well known limitations such as easy to premature, slow convergence and parameters dependency.…”