I am trying to estimate a second-order plus time delay (SOPTD) process model using MATLAB’s System Identification Toolbox. However, in some cases, the estimated model does not start from zero, even though my input data is normalized and clamped to ensure non-negative values. and somecases they start above zero
I have tried several steps to fix this:
Adding an offset to the normalized data to avoid negative values.
Setting the process gain (Kp) to 1 and fixing it.
Setting constraints on the time delay (Td) and ensuring it is non-negative
Despite these attempts, 5 out of my 20 plots still exhibit negative values at the start. I need help understanding why this happens and how to ensure the model never goes negative.I am trying to estimate a second-order plus time delay (SOPTD) process model using MATLAB’s System Identification Toolbox. However, in some cases, the estimated model does not start from zero, even though my input data is normalized and clamped to ensure non-negative values. and somecases they start above zero
I have tried several steps to fix this:
Adding an offset to the normalized data to avoid negative values.
Setting the process gain (Kp) to 1 and fixing it.
Setting constraints on the time delay (Td) and ensuring it is non-negative
Despite these attempts, 5 out of my 20 plots still exhibit negative values at the start. I need help understanding why this happens and how to ensure the model never goes negative. I am trying to estimate a second-order plus time delay (SOPTD) process model using MATLAB’s System Identification Toolbox. However, in some cases, the estimated model does not start from zero, even though my input data is normalized and clamped to ensure non-negative values. and somecases they start above zero
I have tried several steps to fix this:
Adding an offset to the normalized data to avoid negative values.
Setting the process gain (Kp) to 1 and fixing it.
Setting constraints on the time delay (Td) and ensuring it is non-negative
Despite these attempts, 5 out of my 20 plots still exhibit negative values at the start. I need help understanding why this happens and how to ensure the model never goes negative. idproc, procest, system identification toolbox, ma MATLAB Answers — New Questions

​