Analysis of power generation and artificial intelligence based prediction of different types of photovoltaic systems

Technological developments, which have become indispensable in the modern world, have made human beings completely dependent on energy. Most of this increasing energy need is met by fossil fuels today. The limited use of fossil fuel sources, their negative effects on the environment, and their expensiveness have increased the use and research of clean energy sources as an alternative to fossil fuels in recent years. At this point, generating electricity from the sun, a renewable, environmentally friendly, and clean energy source is an issue where investments are increasing worldwide, especially in our country. In 2013, Düzce University Scientific and Technological Research and Application Center (DÜBIT) has a total installed power of 7.5 kW, each of which has 2 with a power of .5 kW; amorphous silicon (a-Si), polycrystalline silicon (poly-Si) and monocrystalline silicon (mono-Si) photovoltaic panels were installed. In this thesis, efficiency and performance analyses of three different types of photovoltaic panels on the roof of DÜBIT, according to seasons and years, for the years 2014-2019 were made. On the other hand, a data set was created using the data of three different types of photovoltaic panels installed on the roof of DÜBIT between the years 2014-2018, as well as meteorological data such as relative humidity, cloudiness, and sunshine duration from meteorology. Classical artificial neural network (ANN) and Long-Short-Term Memory (LSTM) deep learning models based on artificial intelligence have been developed to predict the amount of solar radiation for Düzce. Keywords: Photovoltaic panels, Efficiency, Performance analysis, ANN, LSTM.