PERAMALAN RADIASI GLOBAL MATAHARI JANGKA PENDEK MENGGUNAKAN MODELTRIPLE EXPONENTIAL SMOOTHING-FEED FORWARD NEURAL NETWORK

Rani Fajriyah Islamiyati Asfah, Unit Three Kartini

Abstract


Abstrak

Energi matahari merupakan salah satu dari sumber tenaga listrik yang tidak terbatas dan tersedia dalam jumlah besar. Matahari menghasilkan energi berupa radiasi yang mempunyai rentang panjang gelombang yang sangat besar (Tjasyono, 2004).Energi matahari memiliki pancaran radiasi matahari yang dapat digunakan sebagai energi alternatif. Pengaplikasiannya dapat berbentuk Photovoltaic. Penelitian ini membahas intensitas radiasi matahari di wilayah Unesa tepatnya pada daerah Fakultas Teknik. Penelitian ini diharapkan dapat mengetahui potensi energi matahari yang tersedia untuk digunakan dalam pemasangan Photovoltaic. Model untuk peramalan radiasi global matahari pada penelitian ini menggabungkan dari beberapa model. Pada penelitian yang telah dilakukan sebelumnya oleh Soumyabrata Dev., dkk (2018) hanya menggunakan satu model yaitu model Triple Exponential Smoothing (TES) tetapi kebaharuan pada penelitian ini ialah peramalan radiasi global matahari menggunakan dua model Triple Exponential Smoothing dan Feed Forward Neural Network serta menggunakan data meteorologi. Hasil penelitian peramalan radiasi global matahari jangka pendek denganmodelTriple Exponential Smoothing-Feed Forward Neural Network (TES-FFNN) menunjukkan bahwa tingakat keakurasian dari peramalan radiasi menggunakan modelMean Absolute Percentage Error (MAPE) sebesar 0,2012% pada model TES-FFNN dan 0,2703% pada model TES. Dapat disimpulkan bahwa nilai peramalan radiasi global matahari dengan model TES-FFNN lebih baik daripada penelitian menggunakan model Triple Exponential Smoothing (TES) dalam meramalkan radiasi global matahri selama 1 hari.

Kata Kunci : Peramalan radiasi,Triple Exponential Smoothing, Feed Forward Neural Network, Mean SquaredError, Mean Absolute Percent Error

 

Abstract

Solar energy is one of the unlimited sources of electricity and is available in large quantities. The sun produces energy in the form of radiation which has a very large wavelength range (Tjasyono, 2004). Solar energy has radiant solar radiation that can be used as alternative energy. Its application can take the form of photovoltaics. This study discusses the intensity of solar radiation in the Unesa region precisely in the area of the Faculty of Engineering. This research is expected to find out the potential of solar energy available for use in photovoltaic installations. The models for forecasting global solar radiation in this study combine from several models. In a previous study conducted by Soumyabrata Dev., Et al (2018) only used one model, the Triple Exponential Smoothing (TES) model, but the novelty in this study was the forecasting of global solar radiation using two Triple Exponential Smoothing and Feed Forward Neural Network models and meteorological data. The results of research on global short-term solar radiation forecasting with the Triple Exponential Smoothing-Feed Forward Neural Network (TES-FFNN) model show that the accuracy level of radiation forecasting uses the Mean Absolute Percentage Error (MAPE) model of 0.2012% on the TES-FFNN model and 0.2703% on the TES model. It can be concluded that the forecast value of global solar radiation with the TES-FFNN model is better than research using the Triple Exponential Smoothing (TES) model in predicting global sun radiation for 1 day.Keywords: Forecasting Radiation,Triple Exponential Smoothing, Feed Forward Neural Network, Mean SquaredError, Mean Absolute Percent Error

Full Text:

PDF

Refbacks

  • There are currently no refbacks.