Fitting model parameter from experimental data's

Hello,

I would like to fit model parameter to explain and modelise my system. It is chemical reaction, and the simplest exemple I can write is : A + B --> C + D

To have some think more visual : glucose + acetone ---> acetonide + water. Rate of the reaction would be r = A * exp(-Ea/RT) * [A] * [B]

I have all analytical data from several experiments at various temperature. Best way is to perform essai at a constant temperature, to have k = A * exp(-Ea/RT) and then to easily calculate A and Ea.

In the cas of my model I have 6 equations (r1 to r6) with 8 parameters. After integration of the model with an ode, I am able to compare real valu with calculated one. By simplification I am able to have some parameters (3) with a good certitude. Then I am able to optimise other parameters (done on (Experimetal data - Calculated value) ^ 2) .

I have a quit good result with ga algorithme, but I am unable to fit some think with fmincon : it is like the variation imposed on the fitted parameter are too small.

I am not able to find the way to modify the size of parameter variation. Any help would be welcome !

Thanks for your help Jean-Marie