IST/ISR floating ball setup

MSc dissertation proposal 2014/2015

Nonlinear SISO System Identification

 

 

 

Introduction:

 

As noted by Fliess et al, in the work named "Non-linear estimation is easy"(2008), knowing the state of a continuous system and its derivatives as functions of the control and output variables, is equivalent to know the system itself. Commonly, one has sensors giving observations of the state but not of its time derivatives. In order to identify the system, one can than follow a system discretization approach, as proposed e.g. by Lennart Ljung or, more recently, try to estimate directly the derivatives using filtering kernels. We propose following the later approach, i.e. considering very simple, non-causal, first order, difference kernels to estimate the derivatives.

 

 

Objectives:

 

Given a nonlinear Single-Input-Single-Output (SISO) system, identify its parameters considering that the system is defined as a polynomial model in the state and the input variables. Explore simple derivative approximations to identify the system using few computations.

 

 

Detailed description:

 

Considering a SISO dynamic system model, having x as the system state, u the control input, y the measurements and f, g denoting nonlinear system and measurement functions, i.e. dx/dt(t) = f(x(t), u(t); p), and y(t)= g(x(t), u(t); p) our objective is to identify the system parameters, p usually unknown, or inaccurately known, in real systems. The functions f and g are assumed to be polynomials in x and u.

 

Work-proposal detailed steps:

 

- Create a Matlab/Simulink nonlinear SISO system, and use it to test the system identification methodology

 

- Compare the identification methodology with gold standard methodology by Fliess et al. [Fliess08]

 

- Test the identification methodology on a real system

 

 

References:

 

[Fliess08] Michel Fliess, Cedric Join, Hebertt Sira-Ramirez, ”Non-linear estimation is easy”, Int. Journal of Modelling,

Identification and Control, V4n1, pp12-27, 2008.

 

[Ljung99] Lennart Ljung, ”System Identification - Theory For the User”, 2nd ed, PTR Prentice Hall, Upper Saddle River,

N.J., 1999.

 

[Zehetner07] J. Zehetner, J. Reger, M. Horn, ”A Derivative Estimation Toolbox based on Algebraic Methods — Theory and Practice”. IEEE 2007 Multi-conference on Systems and Control (MSC 2007).

 

 

Requirements (grades, required courses, etc):

-

 

Expected results:

 

At the end of the work, the students will have enriched their experience in nonlinear system identification.

 

 

Place for conducting the work-proposal:

ISR / IST

 

 

More MSc dissertation proposals on Computer and Robot Vision in:

 

http://omni.isr.ist.utl.pt/~jag