Hi Engr MK

I would approach this problem by first generating the 'T' and then loading it to the model. The matlab rand function computes a value between 0 and 1. We can rescale it:

r = (up-lb)*rand(1) + lb;

since your value is +-20% of the nominal value, you can obtain a valid T within this range by scaling it by [0.8,1.2].

lb = 1 - 0.2;

up = 1 + 0.2;

Tnominal = 1;

T = Tnominal * ( (up-lb)*rand(1) + lb );

If you are using Simulink. you probably want to place the code above in Simulation Start Function (File-> Model properties->Callbacks->StartFcn). A random 'T' will be generated every time you run your simulink model.

For instance, you can use the LTI system block, where tf(1,[1 T]).

Thank You. I am going to try this method. Rômulo T. Rodrigues

Thanks Hiva Nasiri

Hello, it is important to characterize your input uncertainty in order to perform the stochastic analysis proposed above. You can sample in this case the uniform PDF in the range [-20% +20%], otherwise, a normal distribution could be used. Look at your state of the art and find a precise definition for this uncertainty. Hiva Nasiri

Hello Engr MK,

there are several ways to simulate uncertain parameters in Simulink. The easiest way is of course a stochastic approach, but you have to think think about your problem and which stochastic distribution fits best. Maybe a chi square distribution or F-score distribution fits better than the standard normal or gaussian distribution.

Regards,

Thomas

T. Pursche Thanks I will take this into consideration.