The model parameters were estimated using a refined instrumental
variable approach with the Captain toolbox in Matlab
(Young, 2011). In order to build the model, different combinations
for na, nbd and ntd were calculated. More specifically, in the SISO
model which has only one input, na ranged from 1 to 3, nbd from
1 up to 3 and ntd from 0 to 2. Therefore, to identify the best fitting
TF model parameters of a total of 48 (4 *4* 3) possible models
were calculated. The resulting models were evaluated by the coefficient
of determination R2T
(Young and Lees, 1993) and an identification
procedure was used to select the most appropriate model
order based on the minimization of the Young Identification Criterion
(YIC) explained by Young and Lees (1993). The smaller the variance
of the model residuals in relation to the variance of the
measured output, the more negative this term becomes.