STUDI PERAMALAN BEBAN LISTRIK JANGKA PENDEK
SISTEM JAWA-BALI MENGGUNAKAN METODE JARINGAN
SARAF TIRUAN DAN KOEFISIEN ENERGI
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
Rhoma Utama, Pangestu Yola
Sudibyo, Uno Bintang
Darmana, Tasdik
Sudibyo, Uno Bintang
Darmana, Tasdik
Subject
Teknik Elektro
Datestamp
2021-06-10 10:07:34
Abstract :
PT.PLN (Persero) UIP2B requires load forecasting to plan weekly and daily system operations and still uses the Energy Coefficient method to obtain daily and weekly load forecasting data where this method takes a long time and the level of accuracy is not good for short-term load forecasting. In this study, it discusses comparing short-term electrical load forecasting using the artificial neural network method with the Energy Coefficient method, where using this artificial neural network method only requires historical data of electrical loads to forecast and ignores the state of each area in the Java-Bali system. and the Energy Coefficient method is a method that has been used by PT PLN (Persero) UIP2B. To compare the two methods by looking for the smallest error percentage value and the best accuracy using the electric load that has been realized as a reference. In this study using historical data on the electric load of PT PLN (Persero) Load Control Center Main Unit in the Java-bali system from 27 October 2019 to 30 November 2019. The results show that the comparison error from Sunday is -0.48% with the value 101.81% accuracy for the neural network method and -1.81% with an accuracy value of 96.60% for the energy coefficient method, for Saturday the error is 0.64% with an accuracy of 99.36 for the artificial neural network method and -1, 81% with an accuracy value of 101.81% for the energy energy coefficient method, and the average error on weekdays is 0.28% with an accuracy of 99.72 for the artificial neural network method and 4.32% with an accuracy of 95.68% for the coefficient method energy.
PT.PLN (Persero) UIP2B requires load forecasting to plan weekly and daily system operations and still uses the Energy Coefficient method to obtain daily and weekly load forecasting data where this method takes a long time and the level of accuracy is not good for short-term load forecasting. In this study, it discusses comparing short-term electrical load forecasting using the artificial neural network method with the Energy Coefficient method, where using this artificial neural network method only requires historical data of electrical loads to forecast and ignores the state of each area in the Java-Bali system. and the Energy Coefficient method is a method that has been used by PT PLN (Persero) UIP2B. To compare the two methods by looking for the smallest error percentage value and the best accuracy using the electric load that has been realized as a reference. In this study using historical data on the electric load of PT PLN (Persero) Load Control Center Main Unit in the Java-bali system from 27 October 2019 to 30 November 2019. The results show that the comparison error from Sunday is -0.48% with the value 101.81% accuracy for the neural network method and -1.81% with an accuracy value of 96.60% for the energy coefficient method, for Saturday the error is 0.64% with an accuracy of 99.36 for the artificial neural network method and -1, 81% with an accuracy value of 101.81% for the energy energy coefficient method, and the average error on weekdays is 0.28% with an accuracy of 99.72 for the artificial neural network method and 4.32% with an accuracy of 95.68% for the coefficient method energy.
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
