On-line tuning procedure of a recursive parameter estimator used for a synchronous generator adaptive control

The paper presents some issues regarding an online tuning procedure for a recursive last mean square (RLMS) parameter estimator embedded into an adaptive controller used to control the terminal voltage of a synchronous generator. The analysis is mainly focused on the problematic regarding the reinitialization of the covariance matrix within the RLMS algorithm. A study case for a dynamical model, with variable parameters (obtained through a linearization, around a functioning point, of a synchronous generator nonlinear model), is performed and some reinitialization rules are highlighted in order to improve the performance of the parameter estimator.