Probabilistic robust design for covariance minimization of nonlinear system

A linear or second-order model is usually used to approximate a practical nonlinear system in probabilistic robust designs. This approximation often makes these designs less effective when the system has strong nonlinearity as well as large uncontrollable variation. In this paper, a novel probabilistic design approach is proposed to design a nonlinear system to be robust under circumstance of large random variation. First, a variable sensitivity approach is employed to place the nonlinear influence under the sensitivity matrix of the covariance model. On this basis, a robust design approach is proposed to minimize the influence of random variation in relation to the performance covariance. Since this proposed approach considers both nonlinear influence and probabilistic distribution in a large uncertain region, it can effectively ensure robustness of nonlinear systems even if large random variation exists.

► We design the robustness of nonlinear system under large random variation. ► The variable sensitivity approach is used to model the nonlinear system. ► The influence of nonlinearity and large random information is considered in design. ► The sensitivity design method is developed to handle the variable sensitivity matrix.