Wiranata, Addien Wahyu (2025) Sistem Prediksi Parameter Kontrol Operasional Pembangkit Cofiring Combine Cycle Menggunakan Metode Deep Learning untuk Peningkatan Pembangkitan Daya. Masters thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 6022231018-Master_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
6022231018-Master_Thesis.pdf Restricted to Repository staff only Download (11MB) | Request a copy |
Abstract
Bauran Energi terbarukan dengan menggunakan teknologi cofiring memberikan dampak yang sangat signifikan terhadap penggunaan Biomassa. Penggunaan Biomassa dengan kualitas berbeda sangat berpengaruh terhadap kinerja bagi suatu pembangkit. Penggunaan Deep Learning Time Series Forecasting akan mengevaluasi kinerja suatu pembangkit dengan mengalisa fungsi kerja dari governor dan keluaran generator. Penggunaan Long Short-Term Memory (LSTM) dengan dikombinasikan pada algoritma Multilayer Perceptron, Convolutinal, dan adam optimizer akan mampu memgoptimalkan proses keluaran daya pembangkit. Proses dianalisis sebagai hubungan antara parameter kontrol (control target) dan parameter prediksi (prediction target) dalam sistem pembangkit listrik. Fokus utama pada dua parameter kontrol, yaitu Governor (mm) dan Output Generator (kW), serta pengaruhnya terhadap parameter-parameter prediksi seperti Flow Steam (m^3/h), Temperature Steam (°C), Pressure Steam (Psig), dan Coal Flow (ton/h). Analisis dilakukan berdasarkan nilai koefisien korelasi pada empat titik waktu historis (T-12, T-6, T-3, dan T-1). Hasil analisis menunjukkan bahwa Governor (mm) memiliki korelasi sangat tinggi terhadap Temperature Steam (°C) ( (0.914), Pressure Steam (Psig) (-0.841), Output Generator (kW) (0.817), dan Coal Flow (ton/h) (0.937) pada waktu T-1. Hanya Flow Steam 〖(m〗^3/h) yang menunjukkan korelasi tinggi namun tidak sangat tinggi (0.645).Sementara itu, Output Generator (kW) sebagai parameter kontrol juga menunjukkan korelasi sangat tinggi terhadap Flow Steam 〖(m〗^3/h) (0.920), Temperature Steam (°C) (0.901), dan Coal Flow (ton/h) (0.830). Hubungan negatif yang kuat juga tercatat antara Output Generator (kW) dan Pressure Steam (Psig) (-0.719). Hal ini menegaskan bahwa waktu T-1 merupakan titik yang paling relevan dalam hal prediktabilitas dan pengaruh parameter kontrol. Serta sebagai unjuk kerja kombinasi Deep Learning Forecasting memiliki nilai error paling rendah dalam ketepatan control yaitu kurang dari 5.33%.
===================================================================================================================================
The renewable energy with cofiring technology has a very significant impact on the use of Biomass. The use of biomass with different qualities greatly affects the performance of a plant. The use of Deep Learning Time Series Forecasting focus is on two control parameters, namely Governor (mm) and Output Generator (kW), as well as their influence on prediction parameters such as Steam Flow (m^3/h), Steam Temperature (°C), Steam Pressure (Psig), and Coal Flow (ton/h). The use of Long Short-Term Memory (LSTM) combined with Multilayer Perceptron, Convolutional, and Adam optimizer algorithms will be able to optimize the process of generating power output. The analysis was conducted based on the value of the correlation coefficient at four historical time points (T-12, T-6, T-3, and T-1). The results of the analysis showed that the Governor had a very high correlation with Temperature Steam (°C) (0.914), Pressure Steam (Psig) (-0.841), Output Generator (kW) (0.817), and Coal Flow (m^3/h) (0.937) at T-1. Only Steam Flow 〖(m〗^3/h) showed a high but not very high correlation (0.645). Meanwhile, the Output Generator (kW) as a control parameter also showed a very high correlation with Steam Flow (m^3/h) (0.920), Temperature Steam (°C) (0.901), and Coal Flow (ton/h) (0.830). A strong negative relationship was also noted between the Output Generator (kW) and the Steam Pressure (Psig) (-0.719). This confirms that the T-1 time is the most relevant point in terms of predictability and influence of control parameters. And the combination of Deep Learning Forecasting has the lowest root mean square error (RMSE) value in control accuracy, which is less than 5.33%.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Cofiring Technology, Deep Learning Time Series Forecasting, Teknologi Cofiring, Deep Learning Time Series Forecasting, Koefisien Korelasi, Long Short Term Memory, Root Mean Square Error (RMSE), Coefficient Correlation, Long Short Term Memory, Root Mean Square Error (RMSE) |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1007 Electric power systems control |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis |
| Depositing User: | Addien Wahyu Wiranata |
| Date Deposited: | 25 Jul 2025 04:21 |
| Last Modified: | 25 Jul 2025 04:21 |
| URI: | http://repository.its.ac.id/id/eprint/121293 |
Actions (login required)
 |
View Item |