“…The objective of this control design is to follow a time‐varying reference signal and to reduce the noise effects acting on each system output. In this sense, we retain the following assumptions: - the polynomials A i ( q −1 , k ), B i ( q −1 , k ), B ij ( q −1 , k ), A ij ( q −1 , k ) and C i ( q −1 ) have the same order in order to facilitate the problem formulation;
- the approximated errors , and are negligible;
- the structure of the submodel (1) is fixed, the delays d i , d ij and t ij are known, in such a way that d i < d ij and d i < t ij ;
- the parameters b i , 1 ( k ), b ij , 1 ( k ) and γ j , 1 are well‐known at each discrete‐time k , in such a way that b i , 1 ( k ) = b ij , 1 ( k ) = γ j , 1 = 1, ∀ k , in order to overcome the implementation difficulties of the parametric estimation ;
- the sequence { e i ( k )} constituted of independent random variables and generated as Gaussian distributed with zero average and constant variance ;
- the sequences { e i ( k )} and { u i ( k ), y i ( k ), u j ( k ), y j ( k ); i , j = 1, … , N ; j ≠ i } are independent.
…”