SPLTRAK Abstract Submission
Estimation and Degradation Analysis of Physics-based Circuit Parameters for PV Systems Using Only DC Operation and Weather Data
Baojie LI1, Xin CHEN1, Todd KARIN1,2, Anubhav JAIN1
1Energy Technologies Area, Lawrence Berkeley National Laboratory, Berkeley, CA, United States
/2PV Evolution Labs (PVEL), Napa, CA, United States

Physics-based circuit parameters like series and shunt resistance are also essential to provide insights into the degradation modes of PV arrays. However, the calculation of these parameters typically relies on I-V curves, which require specific measurement devices and may impede the regular operation of PV systems. Although a reference module equipped with I-V tracers may be installed next to the array, the recorded I-V curves cannot fully represent the array condition. Therefore, I-V curves of the entire array are not always available to obtain the circuit parameters, especially for large-scale PV power plants. 
Therefore, this paper proposes a methodology (PVPRO) to estimate these parameters for degradation analysis without the need for I-V tracers. Rather, it estimates these parameters using only operation (DC voltage and current) and weather data (irradiance and module temperature). PVPRO first performs a multi-stage data cleaning and filtering. Next, the time-series DC data are clipped into windows to fit the equivalent single-diode model to estimate the circuit parameters on minimizing the loss between real and estimated values. Finally, the estimated parameters are plotted as function of time to quantitatively analyze the degradation trends. PVPRO is evaluated on synthetic datasets in the presence of noise and is applied to a well-qualified field system (2015-2019). The degradation trend by PVPRO is in reasonable agreement with the ground truth, especially for the seasonal variation (correlation coefficient of 0.71).
Future work will apply PVPRO to more large-scale PV systems and deconvolve the degradation pathways. The current version of PVPRO is available on Github: