Estimating Parameters in Physical Models using MathModelica
Jonas Sigfrid Sjoberg, Fredrik Fyhr, Tomas Grönstedt · 2002
This paper describes a program which extends MathModelica so that model parameters can be estimated using measured data. Given initial values of the parameters, the parameter estimates are iteratively changed so that the sum of squared errors of the difference between the model output and the data is minimized. In each iteration an extended differential equation has to be simulated. The developed program imports the Modelica model into Mathematica and derives a symbolic expression for this extended differential equation. The extended model i convened to Mod- elica format and MathModelica is used to simulate it efficiently.