Model error bounds for identification of stochastic models for control design

摘要: A method for computing model (Fig. 3) error bounds 1, 50) system identification 30) of stochastic systems is disclosed. The take the form additive frequency-weighted singular value such that they are directly used in H and u - synthesis robust control design methods 60). largest uncertainty bound determined by performing a high number simulations uncertainty. Simulated values computed large data population, each candidate entry simulated lies on 3-sigma ellipsoids defined covariance functions 4, 48, 50). For uncertainty, maximum then determined. In order to determine scalar function needed 60), over population computed.