Robust System Identification with Noisy Time/Frequency Response Experimental Data: Projection Operator and Linear Algorithms

Er-Wei Bai; Sundar Raman

doi:10.23919/ACC.1993.4792827

Back

Conference proceeding

Robust System Identification with Noisy Time/Frequency Response Experimental Data: Projection Operator and Linear Algorithms

Er-Wei Bai and Sundar Raman

1993 American Control Conference, pp.156-160

06/1993

DOI: 10.23919/ACC.1993.4792827

View Online

Abstract

In this paper we consider the problem of robust system identification with noisy time or frequency response measurement data. It is shown here that any linear identification algorithm which is convergent in the noise free case can be made robustly convergent in the presence of noise by incorporating a simple projection operator into the algorithm. Specific algorithms using this projection operator along with corrupted frequency/time response data are analyzed in detail. The computational simplicity and faster rate of convergence distinguish this approach from other existing robustly convergent non linear identification techniques.

Algorithm design and analysis

Convergence

Frequency measurement

Frequency response

Noise robustness

Robust control

System identification

Tellurium

Time factors

Time measurement

Details

Title: Subtitle: Robust System Identification with Noisy Time/Frequency Response Experimental Data: Projection Operator and Linear Algorithms
Creators: Er-Wei Bai - University of Iowa
Sundar Raman - University of Iowa
Resource Type: Conference proceeding
Publication Details: 1993 American Control Conference, pp.156-160
DOI: 10.23919/ACC.1993.4792827
Publisher: IEEE
Language: English
Date published: 06/1993
Academic Unit: Electrical and Computer Engineering
Record Identifier: 9984197284002771

Metrics

21 Record Views