An accurate real time neural network based irradiance and temperature sensor for photovoltaic applications

Abstract

The functioning of photovoltaic (PV) systems mainly depends on the climatic conditions, including incident solar irradiance intensity and module temperature. Thus, irradiance and operating temperature are key factors for assessing and monitoring the performance of PV systems. This work develops a PV solar cells-based sensor for accurate measurement of both incident irradiance and the PV operating temperature. The proposed sensor relies on a Neural Network (NN) soft sensor, trained to estimate PV modules incident solar irradiance and operating temperature without direct measurements. The trained NN model uses the short circuit current (ISC) and open-circuit voltage (VOC) of a reference PV module as inputs of the estimation. The accuracy of the proposed sensor is validated through indoor and outdoor experiments using a PV emulator and a real panel, respectively. Indoor results affirmed the accurate tracking capability of the proposed estimator for both irradiance and module temperature, in constant and rapidly changing conditions, including sudden irradiance changes. The estimator achieved an exceptional level of performance with a Mean Absolute Percentage Error (MAPE) of 1.45% for irradiance and 1.64% for temperature. Similarly, the outdoor experiments using real-time data further validate the obtained results with a Normalized Root Mean Square Error (NRMSE) of 1.22% for irradiance and 3.50% for temperature, outperforming other methods published in the literature.