Recursive parameter method for computing the predicting function of the multivariable ARMAX model (Q2712631)

A set of schemes for recursive calculation of linear minimum-variance \(d\)-step ahead predictor parameters is constructed. The derivation concentrates on single-input single-output and multi-input multi-output systems described by general autoregressive moving average input-output (ARMAX) models as well as by autoregressive (ARX) and moving average (MAX) models with exogenous variable. The proposed recursive-parameter method and the stepwise recursive method suggested earlier by \textit{G. Favier} and \textit{D. Dubois} [Automatica 26, 75-84 (1990; Zbl 0708.60034)] are compared on the ARX, MAX, ARMA and ARMAX basis. It is argued that the proposed method is more convenient.