Improving predictions in LPV-PBSIDopt

System identification of linear parameter varying systems has been the subject of numerous recent developments. One such method, LPV-PBSIDopt presented in, is an extension of the PBSID method for time-invariant systems. In order to deal with time-varying systems, certain approximations are made in the output predictors. As shown in this paper, these approximations are not necessary, and the predictions can be greatly improved with no significant increase in computational cost. However, improving the predictions does not necessarily improve the resulting model. Therefore, simple guidelines for choosing between the two prediction methods are also presented.