Simplified model parameter estimation

hello

I have different variables measured in a diesel engine. I want to estimate transfer functions that best fit the behaviour. for example input power and output exhaust temperature. What should I use to calculate parameters of transfer function. Estimate parameters o system identification tool?

thanks

[Miro]

Do you know, how the output shoud behave theoretically? If yes you could do a simple fitting of the Data to the curve. If there are many parameters in the functio you should use an optimisation function like LSQNONLIN.

Answer 1

[Rajiv Singh]

Try the various linear and nonlinear model identification tools in the System Identification Toolbox product. There are dedicated commands for each type of model; all estimations can be carried out in the GUI too.

Transfer function (linear): tfest (new in R2012a), oe

State-space: ssest (new in R2012a), n4sid, pem

Process models (low order continuous-time ZPK models): procest (new in R2012a), pem

Nonlinear ARX models: nlarx

Hammerstein-Wiener models: nlhw

Grey box models: greyest (new in R2012a), pem; see also idgrey, idnlgrey.

You will find plenty of user examples here: http://www.mathworks.com/products/sysid/examples.html