Reduction of order of linear state equations

A method is proposed for calculating time-varying coefficients of a low-order linear model which approximately describes the behaviour of a known high-order linear plant. In the expression obtained for the coefficients, use is made of the covariance matrix of the combined state and control vectors obtained from a simulation of the response of the high-order plant to random-signal inputs. The procedure is based on an approximate minimisation of the difference between the time rate of change of the variables of the low-order model and of the high-order plant.