作者: R. Doraiswami
关键词:
摘要: A two-stage identification scheme is proposed for multivariable systems applications including spectral estimation, and signal system model estimation. The statistics of the corrupting noise are taken as unknown, except that assumed to have a rational spectrum. First, very high-order estimated then reduced-order derived from higher-order model. An algorithm based on theory heuristics developed select set frequencies where signal-to-noise ratio high. obtained best weighted least-squares fit at selected frequencies.