SINDy prediction for ME sensor output

[SM-CAU]

sindy_sine.txt
Hello,
I am using PySINDy to predict ME sensors output for a sinusoidal, gaussian, and chirp signal to be modelled with a duffing system. For sinusoidal signal, I hope the model is stable and works fine (code for the model is attached)as it can simulate with good accuracy for any part of the signal. What I did there is: I used a custom library with cosines having the frequencies found from the FFT(w) of our signal which takes 'time (t)' as a variable. This makes my driving force term (F cos wt) of a duffing equation. Another custom library for building the candidate library terms for the rest of the terms of a duffing equation.
For gaussian signal, the amplitude of the frquency changes. So the I defined a custom library with gaussian function which can then be the amplitude of the driving force. But, the built SINDy model for gaussian (attached
gaussian_sindy.txt
) can't predict the signal.
Can anyone give me hint concerning what to do, or what might be wrong the idea or implementation. thanks

[Jacob-Stevens-Haas]

If your code is too large to post here, you need to think of a simpler example that demonstrates your problem.