Skip to main content

PVPRO: Methods for Determining Photovoltaic Degradation from Maximum Power Production Data

DuraMAT is developing PVPRO​, an open-source software package for determining the time evolution of I-V parameters for a photovoltaic (PV) string using typical maximum power data. This method will allow users to identify a diverse set of PV degradation modes without time-consuming field surveys.

Power production data is a rich resource for understanding photovoltaic (PV) degradation, but the full power of the information hidden in this data has not yet been leveraged. Time-series data of deployed systems list the system operating current and voltage at a variety of temperatures and irradiance, from which new advanced data analytics techniques can extract detailed information on the electrical performance of the array. Recent work has suggested the exciting possibility to use maximum power point (MPP) production data to infer current voltage (I-V) curves, identify faults or estimate temperature coefficients. However, these promising initial results have not been rigorously validated, tested over diverse data sets or provided to the community in a standard broadly applicable form. In many cases, further development is needed to standardize procedures such as data cleaning and analysis.

To learn more about PVPRO, see GitHub.

Core Objective

Central Data Resource


Lawrence Berkeley National Laboratory, SLAC National Accelerator Laboratory, Sandia National Laboratories, and National Renewable Energy Laboratory


PVPRO is best applied to time-series data sets that contain DC voltage, current, module temperature, and plane-of-array irradiance.




To learn more about this project, contact Todd Karin.

A U.S. map with lines to a box labeled Production Data with data lines for string curr. and string volt.; points to another box labeled PVPRO that lists methods: filter power data, meteorological data, circuit model, parameter estimation, and uncertainty analysis; and lists output: I-V Parameters and Degradation Mode Estimates; and then another box labeled Application, representing BIG data analysis and actionable analytics