Abstract
This paper presents a simplification of a large reducible stochastic system. First, by using a modal approach for a simplification of a linear time-invariant dynamical system, the equations to express the mean and covariance of the state variales of a large reducible stochastic system are simplified. Second, a technique of deriving a simplified model from the above two simplified equations with respect to the mean and the covariance is denoted. The errors between the simplified models and the original systems are evaluated quantitatively. Finally, a numerical example of a second order system is shown to demonstrate the effectiveness of the proposed technique.
