Title :
Choice of Weighting for Averaged Nonparametric Transfer Function Estimates
Author_Institution :
Univ. of Manchester, Manchester
Abstract :
The statistical properties of nonparametric transfer function estimates formed via the weighted sum of empirical transfer function estimates are considered. In the case that the system input is considered deterministic, it is shown that the weighting corresponding to the classical averaged periodogram estimate is optimal. A similar asymptotic result is derived for the case where the input is considered stochastic.
Keywords :
estimation theory; frequency response; spectral analysis; statistical analysis; transfer functions; frequency response; nonparametric identification; nonparametric transfer function estimation; periodogram estimate; spectral estimation; statistical properties; weighted sum; Control systems; Discrete Fourier transforms; Equations; Fourier transforms; Frequency; Gaussian noise; Heat transfer; Stochastic processes; Stochastic resonance; Transfer functions; Frequency response; nonparametric identification; spectral estimation;
Journal_Title :
Automatic Control, IEEE Transactions on
DOI :
10.1109/TAC.2007.906216
