Implementation of an intelligent sensor for measurement and prediction of solar radiation and atmospheric temperature

Abstract

The aim of this study was to develop an intelligent sensor for acquiring temperature, solar radiation data and estimate cloudiness indexes, and use these measured values to predict temperature and solar radiation in a close future. The prototype produced can ultimately be used in systems related to thermal comfort in buildings and to the efficient and intelligent use of solar energy. To incorporate these functionalities, a small and portable prototype was developed, which consisted in: a CCTV camera with a fish-eye lens, for sky images acquisition; a computer of format mini-itx with a Linux operative system, for data acquisition and processing; a GPS, to enable automatic use, independent of the system's geographical position; a pyranometer, for regular measurements of solar radiation; a temperature probe, for regular measurements of outdoor temperature; a shadow band, to eliminate the sun's flare effect on sky images; Arduino, an open source electronics prototyping platform that acquires data from the temperature and solar radiation sensors, as well as processing the data provided by the GPS and controlling the shadow band; neural networks of the type NARX, which use the acquired data to forecast the cloudiness index, solar radiation and temperature, in the next four hours period. The system was programmed to acquire data, both from the sensors and the camera, every five minutes.