Parametric Identification of the Dynamic Photovoltaic Model by a Physics-Informed Neural Network

Parameter identification in photovoltaic modules is essential for control, monitoring and diagnostic purposes. This paper shows the preliminary results obtained in the identification of the parameters of the non-linear dynamic model of a photovoltaic module by using a physics informed neural network. The model includes the series and the shunt resistance and a voltage-dependent non-linear capacitor modeling the semiconductor's dynamic behavior. It is shown that, by integrating simulation data with governing physical laws, a physics informed neural network is able to ensure an accurate and robust parameter identification even the presence of noise. The preliminary results presented in the paper are obtained by using time domain simulation data, which prelude to further tests based on experimental measurements.