Cost Effective Alternative of Pyranometer: Solar Radiation Prediction Using Artificial Intelligence

It is solar radiation which is the Earth's one of the major sources of energy. It has a monumental role in the Earth surface radiation symmetry, hydrological cycles, flora photosynthesis, and weather and climate events. So, the proper prediction of this solar radiation is therefore considered to be very important in all the fields like solar power industry and climatological research. Without measuring solar irradiance in a particular area and at the specific period of time of the year, we can't assume the proper amount of energy to be achieved from the sun using solar panel. Solar radiation is generally obtained with pyranometer sensor devices. But the fact is that the initializing and maintaining cost of these devices are very high. So, these are not widely used in the field. Furthermore, it can provide only confined coverage as a source of data for solar radiation. So various types of data acquired from meteorological stations of sensor are used to predict solar radiation with no use of pyranometer. Again, pyranometer can't give accurate result in winter due to the domes are covered with fogs. So, manual monitoring is required while using these sensor devices by an expert monitor. Yet, remote accessing of pyranometer data is though possible but highly expensive. In this paper, a low-cost system has been developed using artificial Intelligence (AI) and Internet of Things (IoT), which predicted solar irradiance efficiently. Linear Regression as Machine Learning model has been developed for predicting solar radiation taking temperature and relative humidity as input variables. The accuracy score of LR model has been recorded 0.7429 for daily solar radiation data. Predicted solar radiation has been achieved in a text file in the same folder of python file. The ultimate moto of this study is to develop a cost-effective solar irradiance measuring system which will not need manual monitoring. Remote data receiving problem has also been solved.