Abstract
Input-error concept is used to develop the optimal projection equations (OPEQ) for obtaining optimal reduced-order model (ROM) of both known and unknown parameter systems. In the case of a known system, the OPEQ are expressed in terms of the system parameters and in the other case, on escaping the estimation of system parameters, these are expressed in terms of the parameters of an assumed model of an order same as that of the ROM. Introduction of the input-error concept in the reduction is found to ensure a match between the outputs of the ROM and the system and to remove the existing restriction of the persistency requirement of the system input in the model reduction