A new approach of frequency domain points estimation for robust identification

In H_∞ robust identification, the posteriori information to be utilized is system frequency domain point response which should be estimated from measured time domain data already contaminated by noise. Traditional method to obtain such frequency data is based on many separate time domain experiments according to each frequency point, which is offline and batch processing. The paper proposes a new recursive and online algorithm to get system frequency domain point response estimate. The method boasts such advantages: algorithm is only based on one time domain experiment; algorithm is recursive and online; system priori information can be well made use of in the algorithm