PEMODELAN PRODUKSI ENERGI  PEMBANGKIT LISTRIK TENAGA SURYA  1000 WP DENGAN ALGORITMA NAIVE BAYES

Arif Fahrizal; Dani Rusirawan; Lita Lidyawati

doi:10.36787/jti.v16i2.864

Arif Fahrizal Institut Teknologi Nasional Bandung
Dani Rusirawan Institut Teknologi Nasional Bandung
Lita Lidyawati Institut Teknologi Nasional Bandung

DOI: https://doi.org/10.36787/jti.v16i2.864

Kata Kunci: modul fotovoltaik, machine learning, MAE, RMSE, MAPE

Abstrak

Dalam penelitian ini, akan diuraikan pemodelan karakteristik produksi energi Pembangkit Listrik Tenaga Surya (PLTS) 1000 Watt peak (Wp) dalam periode Juni 2020-Desember 2021, dengan menggunakan salah satu algoritma machine learning, yaitu Naive Bayes. Fitur yang digunakan dalam pemodelan adalah radiasi matahari, temperatur & pancaran sinar, yang dibagi menjadi lima label kelas yaitu Very low, Low, Medium, High, dan Very High. Perbandingan data latih dan uji dibagi menjadi 3 skenario, yaitu 90%:10%, 80%:20%, 75%:25%. Hasil pemodelan memperlihatkan bahwa skenario perbandingan 75%:25% memiliki nilai mean absolute error (MAE), root mean squared error (RMSE) dan mean absolute percentage error (MAPE) terkecil (dibanding skenario lain), masing-masing sebesar 0,14; 0,38; dan 6,52, yang artinya skenario ini memiliki akurasi relative lebih tinggi (dibanding skenario lain), jika dibandingkan terhadap karakteristik produksi energi referensi.

Abstract

In this research, the modeling of energy production characteristics of 1000 Watt peak of the Solar Power Plant (SPP) will be elaborated for the period of June 2020-December 2021, using one of the machine learning algorithms, i.e., Naïve Bayes. The features used are solar radiation, temperature & light beam. Each of feature is divided into five class labels, namely Very low, Low, Medium, High, and Very High. The comparison of training and test data is divided into 3 scenarios i.e., 90%: 10%, 80%: 20%, 75%: 25%. Based on modeling, it is found that a 75%;25% scenario shows the smallest of MAE, RMSE, and MAPE, i.e., 0.14, 0.38, and 6.52, respectively. It means that this scenario has the highest accuracy in this modeling (if compared to the actual's energy production characteristics).