Recursive generalized extended least squares method and its application in a power plant

The recursive generalized extended least squares method (RGELS) is presented. It can identify the parameters of scalar systems in the presence of color noises, and simultaneously estimate all parameters of the whole multivariable system. The problem of identification in the presence of interactive noises at different outputs is solved. The convergence of the RGELS algorithm is analyzed by means of stochastic process theory. Compared with subsystem identification algorithms, the RGELS algorithm requires much less calculation. The RGELS algorithm has been successfully applied to estimate the parameters of simulated plants and a superheated steam temperature control system in a power plant