RLS Performance Analysis for a Synchronous Generator Model Identification

This paper presents a performance analysis of the recursive least squares (RLS) method, which is used for a synchronous generator (SG) model parameter identification. The RLS identification method includes measurements of the terminal voltage, electrical power, rotor angle or rotor speed following a perturbation in the reference of the terminal voltage of the SG. The performance analysis is carried out on a Heffron-Phillips (H-P) linearized model of a SG with an additional regulation loop of the terminal voltage. Such linearized H-P model is very convenient for the RLS performance analysis, as fourth order transfer functions with known parameters describe the system dynamics in the close vicinity of the chosen operating point of the SG. The performance analysis includes testing of different perturbation signals, sampling times, forgetting factors and other parameters of the RLS method. The influence of the discussed method parameters on the convergence and accuracy of the determined model parameters, as well as on the identification time, is presented. Such performance analysis is a good basis for the design of the online identification of SGs in power plants, which can be used for adaptive power system stabilizers.