What are the 7 states for the model?

I am using System Identification tool box. I am using 2 signals as input and 3 signals as output. the system identification is giving as result a state-space model n4s7. Which has an A = [ 7 X 7 ] B = [ 7 X 2 ] C = [ 3 X 7] and D = [ 3 x 2]. How can I define the 7 states in the system? In the book for system identification toolbox, it is mentioned that the state can be reconstructed from the model. Anyone has an example that I can follow to define my states.

Thanks.

Answer 1

[Jose]

I highly appreciate your help Rajiv Singh. I want to clarify the question: I have used Matlab to generate a block box model of my system using the Sys. ID toolbox. My initial system is quantified by 3 state variables u,v and psi. The state vector output of the black box model has 7 elements in it. How can I determine what they physically represent or relate them to the original u, v and psi states of my system. Thanks

Answer 2

[Rajiv Singh]

Can you elaborate on what you mean by "defining the states"? If you want to calculate the state-trajectory (as a function of time), use SIM or LSIM command. If you want to attach name/units for state variables use the StateName, StateUnit model properties.

[Jose]

I highly appreciate your help Rajiv Singh. I want to clarify the question: I have used Matlab to generate a block box model of my system using the Sys. ID toolbox. My initial system is quantified by 3 state variables u,v and psi. The state vector output of the black box model has 7 elements in it. How can I determine what they physically represent or relate them to the original u, v and psi states of my system. Thanks

[Rajiv Singh]

The generated states cannot be related to your physical states you desire. If you had estimated a 3 state model, the estimated states would be related to the physical ones through a state coordinate transformation matrix T which is again impossible to determine in absence of additional information. For physically meaningful realizations, you have to build a structured (grey-box) representation. This basically means that you have to know the equations of motion explicitly and use the software simply to estimate the coefficients of those equations. See:

http://www.mathworks.com/help/ident/ug/estimating-linear-grey-box-models.html