Simulink: Optimize model parameters to match measured values

Dear Community,

I am mechanical engineer and I am using Matlab / Simulink r2015b as tool to solve engineering problems. I'd like to as for help for the task described here:

1. In Simulink, I have a model of a component which calculates an output temperature of the system
2. The model is based on several key parameters
3. I also have measurements taken from the real component (measured output temperature values to be exact)
4. The model was verified with a known set of parameters and measurement values
5. I'd like to use the model to optimize the parameters, so that I can simulate every possible component

Fig. 1: Simulink model

In short: The optimization goal is to find parameters which minimize the deviation between the simulated temperature values and the measured temperature values.

For this I'd like to use lsqnonlin from the optimization toolbox. I understand the basic concept of this optimization function and I used it in a purely matlab context, but never in combination with a Simulink model.

Can anyone give me a "jumpstart" here?

Some pseudo-code in order to show where I stand:

fun = (simulated temperature values) - (measured temperature values)
parameterSet = lsqnonlin(fun,paramterSet0)

Thank you all in advance.

Best regards, Michael

Answer 1

[Alan Weiss]

Perhaps this answer will help. I do not know anything about Simulink, so will not be able to help further.

Alan Weiss

MATLAB mathematical toolbox documentation