Estimation of solar radiation per month on horizontal surface using adaptive neuro-fuzzy inference system (case study in Surabaya)

• Main Authors: Maknunah, Jauharotul, Abadi, Imam, Abdurrahman, Isnan, Imron, Chairul
• Format: Conference Proceeding
• Language: English
• Subjects: Adaptive systems; Algorithms; Alternative energy sources; Artificial neural networks; Fuzzy logic; Fuzzy systems; Inference; Machine learning; Measuring instruments; Neural networks; Photovoltaic cells; Root-mean-square errors; Solar energy; Solar radiation
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/1.5095323

One source of natural energy that is always available and has a great energy that is solar energy. Solar energy can emit radiation known as solar radiation. Solar radiation can be utilized into alternative energy that is in the form of electrical energy with the help of Photovoltaic which is currently almost all components require a source of electrical energy. Solar radiation can be measured by a measuring device called pyranometer. Pyranometer including measuring instruments that have a relatively expensive price compared to other measuring instruments. So as to raise research to get an estimation of solar radiation per hour on photovoltaic. This estimate uses Adaptive Neuro-Fuzzy Inference System (ANFIS) to model solar radiation estimators. ANFIS is a fuzzy inference system based on an adaptive neural network that adopts a learning system from an artificial neural network. This research uses an ANFIS which in its learning can use reversal algorithm or hybrid algorithm. The results of this study indicate that the ANFIS method has a smaller Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values than the ELM method of artificial neural networks. Of the two methods used are ELM. ANFIS for ELM method can deliver the RMSE value of 15.42 and using ANFIS method got the RMSE value of 0.87. In addition to the value of MAE on the ELM method obtained by 11.72 and for ANFIS method obtained by 0.22.