Modeling the energy efficiency of an air heated solar collector with machine learning algorithms

Increases in the cost-demand of fossil fuels used in energy production and environmental concerns have led people to alternative energy sources. The use of solar energy as a clean and renewable energy source has gradually increased. It is widespread to use solar collectors working as heat exchangers to benefit from solar energy. In addition, studies on the use of data in solar energy in the analysis and evaluation of energy systems, their accurate estimation, and the creation of utility models have increased in recent years. Machine learning is the most common method used in different parts of the world in data analysis in solar collectors. Machine learning algorithms are flexible and non-parametric modeling tools. It is quite effective in complex problems. It can find solutions to many problems such as estimation, classification, and clustering. This thesis presents a useful model with a machine learning algorithm for the energy efficiency value obtained by the 1st Law of Thermodynamics of the solar air collector (HGK). The pace regression, artificial neural network, decision tree, and support vector machine algorithms were used for this model. Thanks to Pace regression, the thermal performance of HGK is expressed with a mathematical equation. The energy efficiency values of an HGK, which have been experimentally studied before, are modeled by the decision tree algorithm with 3% accuracy. The accuracy results of the models obtained show that the energy equations produced by machine learning algorithms may have the potential to be used in different types of solar collectors.