TS-RLS algorithm for pseudo-linear regressive models

This paper presents a two-stage recursive least squares (TS-RLS) algorithm for pseudo-linear regressive models corresponding to the Box-Jenkins models by combining the auxiliary model identification idea and the decomposition technique. The basic idea is to decompose a system into two subsystems with the system model parameters and the noise model parameters, respectively, and then to identify the parameters of each subsystem. Compared with the auxiliary model based recursive generalized extended least squares algorithm, the TS-RLS algorithm has less computational burden. The simulation results confirm these conclusions.