-, Muhammad (2015) Perbaikan Peramalan Beban Jangka Pendek Berbasis Wavelet Neural Network Dengan Optimalisasi Prediksi Frekuensi Tinggi. Undergraduate thesis, Institut Technology Sepuluh Nopember.
Preview |
Text
2213105080-Undergraduate Thesis.pdf - Published Version Download (2MB) | Preview |
Abstract
Dalam dunia kelistrikan, peramalan beban jangka pendek
merupakan hal yang penting untuk meningkatkan operasi sistem tenaga
listrik yang lebih optimal dan handal. Cukup banyak factor yang
berpengaruh pada proses pengolahan data untuk peramalan beban jangka
pendek pada umumnya.
Dalam Tugas Akhir ini akan diimplementasikan peramalan
beban jangka pendek berbasis wavelet neural network dan
dikembangkan dengan sebuah metode partial autocorrelation function
untuk pengolahan data input beban, khususnya pada komponen
frekuensi tinggi. Metode ini merupakan tahap di mana setelah data
diproses secara wavelet pada frekuensi tinggi dan sebelum data diolah
dalam neural network, data tersebut diolah sehingga didapatkan korelasi
yang signifikan guna meningkatkan keakuratan hasil peramalan beban
frekuensi tinggi dan hasil akhir peramalan beban.
Hasil akhir peramalan pada penelitian ini lebih baik dari
penelitian sebelumnya, di mana pada penelitian ini menggunakan input
indeks hari kerja, indeks jam, input data beban historic untuk frekuensi
rendah (H-7, H-14, H-21), korelasi input beban untuk frekuensi tinggi
(t-50, t-53,t-82, t-90, t-91,t-92,t-93,t-94), input data beban kemarin dan
input data beban hari ini pukul 24.00 dengan nilai minimal error sebesar
0,01%, maksimal error sebesar 6,55% dan MAPE (Mean Average
Percentage Error) sebesar 2,46%.
=====================================================================================================
In the world of electrical, short-term load forecasting is the
most important thing to increase the power electrical optimation which
make it more great and reliable. There are some factors enough which
affecting on data processing for short term load forecasting generally.
In this final project,will be implemented a short-term load
forecasting based wavelet neural network and innovated by using
partial autocorrelation to processing input data load, especially, for
high frequency component. This method is a step where data load after
proceed by using wavelet on high frequency and before the proceed data
into neural network, the data load proceed so that got the significant
correlation use to increase accuracy of load forecasting result of high
frequency and the final forecasting result.
Final load forecasting result on this final project was better
than before, where this research use data input work-day index, clocktime
index, data historic for low frequency (H-7, H-14, H-21),
correalation input load for high frequency (t-50, t-53,t-82, t-90, t-91,t-
92,t-93,t-94), data load of yesterday, and data input load on 24.00 with
minimum error 0,01% , maximum error around 6,55%, and MAPE
(Mean Average Percentage Error) is 2,46%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSE 621.374 5 Muh p |
| Uncontrolled Keywords: | Neural Network, Wavelet, partial autocorrelation function |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK201 Electric Power Transmission |
| Divisions: | Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis |
| Depositing User: | Mr. Tondo Indra Nyata |
| Date Deposited: | 10 Oct 2019 08:03 |
| Last Modified: | 10 Oct 2019 08:03 |
| URI: | http://repository.its.ac.id/id/eprint/71115 |
Actions (login required)
 |
View Item |