Error using 'pem' and 'greyest' in System Identification of a MIMO system.

4 vues (au cours des 30 derniers jours)
Pranjal Biswas
Pranjal Biswas le 13 Mar 2017
Commenté : Gergely Takács le 7 Déc 2018
I am have been working on System Identification of a MIMO system. While using the commands:
sys_grey_out = greyest(data, sys_grey, opt);
or
sys_pem = pem(data, init_sys);
I am getting the following the error:
The initial computation of the loss function failed. The initial model, if
specified, may be unstable. Consider setting estimation focus to "stability" or
"simulation". Also make sure that the parameter bounds do not make the model
unstable.
Setting the focus to Simulation or Stability isn't helping. Can anyone help me with the error?
  2 commentaires
Gillian Rosen
Gillian Rosen le 15 Mar 2017
Hi Pranjal, 
I understand that you are trying to use the functions 'pem' and 'greyest' for system identification of a MIMO system, but that you are encountering an error that states that "the initial computation of the loss function failed". 
To ensure that this issue is specific to your particular system, have you tried running examples from the 'greyest' and 'pem' documentation pages (linked below)? Were you successful? 
You mentioned that you have tried setting the focus to 'Stability' and 'Simulation', as suggested by the error message. Have you also made sure that the parameter bounds for your model do not make the model unstable? 
Changing other 'greyest' options to better fit your system may also help in this situation. See the following documentation link for a full list of options: 
If you need more assistance, it would be very helpful if you posted a code snippet that would allow others to recreate the error that you are seeing. 
Gergely Takács
Gergely Takács le 7 Déc 2018
Gillian;
I am getting the same error message for a
nlgreyest()
problem. The error message seems to suggest that nlgreyest() can only handle open-loop stable models, is that correct? (I know that my model is unstable...)

Connectez-vous pour commenter.

Réponses (0)

Catégories

En savoir plus sur Linear Model Identification dans Help Center et File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by