The use of exponential weighting in the excitation dependent least squares estimator

The excitation dependent least squares estimator represents an alternative solution to the problem of parameter drifting in the parameter estimation of time-varying dynamic systems. In order to improve the technique´s parameter tracking ability we incorporate to it a forgetting factor that depends on the chosen data block length. This makes the algorithm more suitable for real time applications. Simulation results that illustrate the capabilities of the new algorithm are presented