Duan Li, Xiaoling Sun and Ken McKinnon
Technical Report MS 00-018
Systems reliability can be improved by adding redundant components or increasing the reliability levels of subsystems. Determination of the optimal amount of redundancy and reliability levels among various subsystems under limited resource constraints leads to a mixed-integer nonlinear programming problem. The continuous relaxation of this problem is a nonconvex nonseparable optimization problem with certain monotone properties. In this paper a convexification method is proposed for solving this class of continuous relaxation problems. Combined with a branch-and-bound method, our solution scheme provides an efficient way to find an exact optimal solution for reliability optimization in complex systems.
Reliability optimization, global optimization, convexification method,
outer approximation, branch-and-bound.
PDF MS 00-018.pdf (~138K).
postscript MS 00-018.ps (~360K).
G-Zipped postscript MS 00-018.ps.gz (~175K).