To establish a statistical formulation of robust design optimization and to develop a fast optimization algorithm for the solution of the statistical design problem.
Existing formulations and methods for statistical robust design are reviewed and compared. A consistent problem formulation in terms of statistical parameters of the involved variables is introduced. A novel algorithm for statistical optimization is developed. It is based on the unscented transformation, a fast method for the propagation of random variables through nonlinear functions. The prediction performance of the unscented transformation is demonstrated and compared with other methods by means of an analytical test function. The validity of the proposed approach is shown through the design of the superconducting magnetic energy storage device of the TEAM workshop problem 22.
Provides a consistent formulation of statistical robust design optimization and an efficient and accurate method for the solution of practical problems.
The proposed approach can be applied to all kinds of design problems and allows to account for the inevitable effects of tolerances and parameter variations occuring in practical realizations of designed devices.
