Two new types of bias-eliminated least-squares (BELS) based algorithms are proposed for consistent identiÿcation of linear systems with noisy input and output measurements. It is shown that estimation of the noise variances can be implemented through one-dimension over-parametrization of the system
Transfer function matrix identification from input—output frequency response data
✍ Scribed by Zhiqiang Gao; Bruce Tabachnik; Razvan V. Savescu
- Publisher
- Elsevier Science
- Year
- 1994
- Tongue
- English
- Weight
- 942 KB
- Volume
- 331
- Category
- Article
- ISSN
- 0016-0032
No coin nor oath required. For personal study only.
✦ Synopsis
A new formulation of transfer function matrix identtj?cation in frequency domain is introduced. It reduces the problem to a simple linear least square problem. It is shown that such a system identtjication problem is a special case of a matrix interpolation problem and much insight can be obtained by examining its algebraic characteristics. A new approach is proposed to determine the transfer function matrix of a multi-input and multi-output system from the input-output data. It eliminates the common assumption in the literature that the ,frequency response of the system is given. Its ef$ciency and practicality is superior to the existing methods, where the solution is obtained by solving a nonlinear least square problem using mathematical programming techniques. The simplicity of the new procedure makes it a viable candidate for real time implementation where systems can be identt$ed on-line. Unmodeled dynamics can also be better characterized.
📜 SIMILAR VOLUMES