Prediksi Radiasi Sinar Matahari Menggunakan Metode Deep Autoregressive

Abstract

Solar energy is one of the renewable energy sources that is growing very rapidly in its utilization. Through solar power plants (PLTS), solar radiation can be converted into electrical energy using solar cells that utilize photovoltaic (PV) technology. However, the performance of PV systems can be affected by various environmental factors, such as temperature, humidity, wind speed, and light intensity. In addition, the operation of solar power systems with high penetration can lead to voltage fluctuations and unstable solar electricity production, causing an imbalance between energy demand and supply. In this case, accurate prediction of solar radiation is essential to make good planning in managing and operating solar power systems so that it can produce energy optimally and maintain the balance of energy supply and demand. In this study, the Deep Autoregressive (DeepAR) method is applied to train a model to predict solar radiation. The results of this study show that the model trained with the DeepAR method has good performance in predicting solar radiation with the best evaluation metric value achieved in the prediction time span of 7 days with an R² value of 0.956675, MAE of 7.835733, MSE of 97.33104, and RMSE of 9.86565.