The Prediction of Solar Radiation Using Fuzzy Logic: a Case Study

Abstract

Solar energy is used in many applications such as producing agricultural food, renewable energy, and heating and lighting systems… etc. Nowadays, countries all over the world, especially the developing countries are facing a great challenge which is providing sustainable energy for consumers. Electricity is the most common type of energy that is used by consumers in which oil or nuclear power is used to produce sufficient amount of electricity for the constant increase of the population in the present. However, both oil and nuclear energy negatively affect the global warming; therefore, solar energy is aspired by many countries to decrease the effects of the global warming and produce renewable sources of energy. The aim of this study is to predict the use of solar radiation for solar energy to produce electricity in Duhok city due to the fact that "national electricity" is not enough for the great number of consumers; as a result, people depend on "local or private generators" which mainly depend on oil to produce electricity. Fuzzy logic approach is used to estimate the solar radiation. The four fuzzy systems are created using the available data in Duhok City in 2016. Daily observations for temperature, humidity and wind speed for four seasons are analyzed to estimate the solar radiation. The predicted outputs of fuzzy logic system are compared with the actual solar radiation. In addition, the fuzzy system approach is evaluated using Mean Absolute Percentage Error (MAPE) and Absolute Percentage Error (APE). The outcomes of MAPE and APE are 5. 86%, 1.54%, 2.76% and 1.52 for four seasons (winter, summer, spring and fall), respectively. According to the results, the performance of fuzzy system is reasonably effective in predicting the solar radiation.

Development and Assessment of Feed Forward Back Propagation Neural Network Models to Predict Sunshine Duration

“…For every combination of input parameters, the training algorithms of Levenberg-Marquardt and tangent transfer functions with one hidden layer and varying numbers of neurons were examined. The goal is to find the best architecture of MLP-FF with the highest coefficient of determination, the lowest proportion error of mean absolute error and the lowest proportion error of root mean square [27,28].…”