System parameter identification with missing outputs

This paper develops three approaches for directly identifying system model parameters with missing outputs. The first is based on missing output estimation, the second is the generalized Yule-Walker algorithm and the third is a two-stage algorithm based on correlation technique. The basic idea is to directly identify the parameters of underlying single-rate models instead of the lifted models of dual-rate systems from the dual-rate input-output data, based either on the replacement of the missing outputs with the outputs of the estimated single-rate models or on the stationarity and ergodicity assumptions of system input-output data. An example is given