Method for solving nonlinear goal programming tasks with interval coefficients using GA

In this study, the authors proposed a solution for directly using genetic algorithms without linear conversion for solving nonlinear goal programming problems involving interval coefficients. The proposed technique involves directly computing left-side interval numbers in the goal constraint equation from the genetic algorithm solution and comparing their size with those of the right-side numbers, for obtaining new difference variables. These difference variables vary depending on the goal type; they are defined using weights which express interval order evaluation criteria. In addition, as an example of application field for this method, the authors studied a large-scale problem for optimal design of system reliability involving interval coefficients, in order to clarify localization of this method. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(4): 55–65, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1145