PVfit: Latest News

It's Happening!

20 November 2020

22 August 2020

  • Today we have released the first in a series of updates to this PVfit website frontend. Mostly a parity update, you will nonetheless find improved model documentation and usability improvements. Additional features are on the way, including v3 of the backend REST API.
  • Model naming conventions have been streamlined. The "local" single-diode model (SDM-L) is now the single-diode equaiton (SDE). The "global" single-diode model (SDM-G) is now the single-diode model (SDM). Endpoints for the backend REST API have been updated accordingly, but the old endpoints should remain active until the v2 API is retired.
  • We have paused development of the double-diode equation/model due to other priorities and inconsistent performance of the fitting algorithm. Inquire with us if you miss this capability and are interested in beta testing.
  • Special thanks to Tor Gilbertson (aka Mother) for rocking the frontend refactor! Please direct your feedback/suggestions here.

13 July 2020

  • The pv-fit.com domain name was retired in favor of pvfit.app. This includes the REST API access points at api.pv-fit.com, which is now api.pvfit.app. Please update any/all REST API usages.
  • The v1 REST API has also been retired.

1 October 2019

  • In May at the 2019 12th PV Performance Modeling and Monitoring Workshop, we presented this poster on using PVfit for a PV-based effective irradiance and temperature sensor, which provides a more direct measurement approach than a meteorological (MET) station.
  • We seek instrumentation partners to help validate the aformentioned non-intrusive and non-obstructive PV-based effective irradiance and temperature measurement technique. The fit implementation should eventually be available here and as part of the REST API.
  • The Python 3 implementation of the "forward" single-diode equation and model used by PVfit has been open-sourced on GitHub! New code for computing temperature-dependent spectral mismatch correction factors has also been added. NOTE: Some of the API is not yet finalized/stable, but it is designed to seemlessly integrate with the parameter fit results from the REST API.

20 January 2019

  • We fixed some edge cases where less than perfect I-V curves caused failures in the estimation of initial condition.

15 January 2019

  • We fixed some ommissions/typos in the new v2 API documentation for SDM-L and SDM-G.

7 January 2019

  • The new year has brought the v2 release of the SDM with some API changes, which are intended to make using PVfit more explicit and intuitive, e.g., parameterization changes and better communicaiton of fitting exceptions/success/failure.
  • We have moved away from using ODR for our fits, because the over-parametrized fits were often suboptimal in the presence of significant model discrepancy, which is not uncommon for the SDM.
  • In SDM-G, we removed the temperature coefficient for the material band gap, which was not well identified over the temperature range of typical I-V curve datasets and apparently made convergence less robust.
  • The new least-squares-based fitting algorithm has also shown good results fitting SDM-G to "sparse" IEC 61853-1 matrix datasets for x-Si and HIT materials! (Many thanks to Bill Marion and NREL for making available the mPERT dataset, which was used for tesing these new fits.)
  • Crystalline GaAs was added to the list of "stock" material systems for the SDM-G fit.
  • We squashed a bug in the computation of the resistances at Isc and Voc.
  • The v1 API endpoints for the SDM-L and SDM-G are deprecated, but should remain operational for at least 3 months. This website now exclusively uses v2 for the SDM.
  • The much anticipated v1 release of DDM-L and DDM-G with analagous functionality should be coming soon!

25 July 2018

  • Double diode model fitting has arrived! Our local double diode model (DDM-L) fitting service is now in beta. We likely have a lot of hardening to do and appreciate any feedback.
  • We're hoping that the PV community can now more easily compare SDM-L fits to DDM-L fits. Keep an eye on the photocurrent-normalized root-mean-square deviations (RMSD) for each fit, and compare plots of the current-sum residuals using your favorite graphing application.

17 July 2018

  • We moved the News to this page, and PV Material Type selection is now easier in SDM-G.
  • We fixed the SDM-G fitting issue from this website and https access has been restored. If you are still experiencing issues, then please force reload the site and/or clear your browser cache. Thank you for your patience!

16 July 2018

  • We are actively investigating an apparent issue with the SDM-G fittings from this website. Direct access via the REST API appears to be unaffected.

15 July 2018

  • Our single-diode model (SDM) API is out of beta! The new version 1 (v1) URLs to POST data to are now https://api.pv-fit.com/v1/sdm/local (for SDM-L) and https://api.pv-fit.com/v1/sdm/global (for SDM-G). IMPORTANT: There were some final (breaking) changes to the variable names passed in and out via JSON. See the updated API documentation. Our intent is that any future changes to this v1 API should be backwards compatible to existing codes.
  • We fixed a website user interface bug in SDM-G that reported the wrong values for 'bgmodeV0' and 'alphabgmod0'.
  • While our 'api.pv-fit.com' has always used encryption, this website now also uses https.
  • We are still working on providing local and global double-diode model fits.
  • This website is due for an update to use Polymer 3, which will hopefully fix some of the math formula typesetting issues seen on some browsers.

2 June 2018