Though my input and output data is inversely proportional, system identification toolbox is giving me a positive gain

Hello all. I have a set of input and output data collected from a process. As my input increases, my output decreases and so technically my gain should be negative. But when I load the data in system identification app, and estimate using process models, Im getting a huge positive gain. Where could be the potential mistake?

Answer 1

[Dhruv]

Hi Saraswathi,

There might be some areas where you can check for potential issues:

1. Data Preprocessing: Ensure your input and output data are correctly aligned and preprocessed. Any inconsistencies or noise in the data might affect the model estimation.
2. Model Selection: Verify that the chosen model in the System Identification Toolbox is appropriate for capturing the dynamics of your system, including any inverse relationships.
3. Parameter Initialization: The initial guess for the model parameters can significantly influence the estimation results. Consider providing a manual initial guess that reflects the expected negative gain.

I hope these checks will help you identify and correct the issue.