Robust identification on limited measurable frequency band

The paper solves the problem of robust identification when system frequency posteriori data can only be measured on some frequency band. Robust convergency of traditional robust identification algorithm is based on the assumption that system frequency samples can be obtained on the whole frequency band, which is hard to satisfy in many applications. First, we reconstruct frequency samples on unmeasurable frequency band through system samples obtained on measurable frequency band. Then we propose a robust identification algorithm which deals with frequency samples not equally distributed on the whole frequency band. So, we transform identification problem on limited frequency band to the one oil whole frequency band.