System identification that is reconstruction of mathematical model from experimental data is one of the important areas of system dynamics and general system analysis. Serious problems in solving the problem arising in the case of identification when data corrupted by the errors. In practice, this case is most realistic because it is impossible to obtain accurate data under measurements or in computer calculations.
The presence of errors in the available data mean that in many cases identification problem become ill-posed. Under certain assumptions about the stochastic uncertainty it become possible using covariance analysis come away from ill-conditionality and receive unbiased estimates for model parameters, that provide consistence of evaluation.
In proposed research direction ill-posed identification problem will be solved using the method of regularization. It is planned to use in research approaches numerical simulations.
System Analysis and Control
Topics:
1. Finding approximate models with states space description using methods based on regularization and realization theory
2. Researches of nonparametric identification methods for complex systems
3. Structural and parametric identification methods using finite-difference models classes
4. System identification on frequency parameters
5. System identification based on the direct variational methods