Abstract:
Solar intensity is a necessary parameter in testing and installing a photovoltaic power system including research on the use of solar cells. Its intensity is generally measured by using a solar intensity sensor or pyranometer which has limitation on its price. However, if light sensors such as small solar panels could be used to measure solar intensity, this limitation would be reduced.
This thesis proposes an algorithm design for small solar panels to measure solar intensity which had almost the same results as those measured by a pyranometer. Data from the small panels were firstly used to make curve fitting with the data obtained from the standard pyranometer. Then they were used to train a backpropagation artificial neural network in order to adjust the data from the solar panels. After that, the light values drawn from the panels were directly fed as the trained neural network input. Finally, the values obtained from the output of neural network were compared with those obtained from the standard pyranometer.
Based on the proposed method applied to 3-volt and 5-volt photovoltaic panels, compared with the results of using the SP-110 pyranometer for a 10-day period, similar results were found with an average error of 1.16% and 1.07% for 3-volt and 5-volt photovoltaic panels, respectively. Hence, it can be concluded that the proposed low-cost small solar panels could be used to measure solar intensity.