port MATLAB PVLIB's Boyle/Coello soiling model to python

[wholmgren]

sandialabs/MATLAB_PV_LIB#26

help wanted...

[alexandermorgan]

I would like to participate in this issue. Is there a preference as to which model to implement first, i.e. Kimber Dynamic or Boyle/Coello? I could start by doing a pandas rewrite of the code @mikofski shared in issue #669. Mark's code was very clear and in Python, so maybe it's easiest to start with the Kimber Dynamic method.

[nappaillav]

Hello
As per the suggestion of @cwhanse, I have started working on, porting of the code from Matlab to python. I have taken reference from pvl_soiling_hsu.m code.
I have few doubts like should I reproduce the whole Matlab code (both pvl_soiling_hsu and depo_velocity or just the depo_velocity).

[DanRiley]

@nappaillav If you have questions when porting any part of the Humboldt State University (HSU) code, feel free to contact me. I worked on developing that code from the code that was sent to me by the paper authors.

I recommend reproducing the entirety of the MATLAB code. The internal function depo_velocity is used by the variable deposition form of the model, but requires everything in pvl_soiling_hsu.m in order to correctly model the soil impact.

[cwhanse]

I think the expected behavior should be soiling_ratio = f(inputs), as pvl_soiling_hsu.m does. The MATLAB input ModelType selects the soiling deposition model from 3 options (option 1 uses delo_velocity helper function), I'd default to ModelType=2.

FYI: the paper authors requested us to name the function with HSU rather than a last name. It would be nice that this function become known as the Humboldt State University soiling model to honor their request.

[nappaillav]

Thanks for the reply @cwhanse @DanRiley
I'm trying to handle the argparse in python similar to matlab.
Else I have a complete code ported in python. Apart from that I'm yet to test the functions.
Hence It would be great if you can provide me with some test data which I can use to see if the python code reproduces the same values as the Matlab HSU code.

[cwhanse]

I wouldn't work too hard to imitate the MATLAB argument parsing. You can take defaults from the MATLAB function's Optional inputs. I wouldn't port the validation rules; rather, describe the value restrictions in the docstring.

@DanRiley do you have a few sets of test data?