MISO fractional systems identification with fractional models in the EIV context

This paper proposes a new multi-input-single-output (MISO) system identification methods with fractional models in the errors-in-variables context. The developed methods are based on the instrumental variables and use the Higher-Order Statistics (HOS), such as the third-order cumulants, to obtain an unbiased estimate. Two different cases are established : the first supposes that the fractional orders of the single input-single-output (SISO) systems decomposing the MISO system are known a priori and only their linear coefficients are estimated. In the second case, the fractional orders are optimized along with linear coefficients. A Monte Carlo simulations are used, in a numerical example, to analyze the consistency of the developed estimators.