Perancangan Sistem Prediksi Keandalan Turbin Gas Dengan Pendekatan Algoritma Jaringan Saraf Tiruan

Pramudya, Rafly Andra (2025) Perancangan Sistem Prediksi Keandalan Turbin Gas Dengan Pendekatan Algoritma Jaringan Saraf Tiruan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009221137-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (3MB) | Request a copy

Abstract

PT Petrokimia Gresik sebagai industri petrokimia dan fertilizer membutuhkan energi listrik untuk berjalannya proses produksi. Konsumsi energi listrik PT Petrokimia Gresik pada Agustus 2021 menggunakan konsumsi sebelas persen dari PLN dan sisanya menggunakan energi listrik yang diproduksi dari departemen utility dengan gas turbin generator sebagai pembangkit listrik. Keandalan turbin gas sangat penting untuk diperhatikan karena kegagalan dapat menyebabkan gangguan operasional, biaya tinggi, dan risiko keselamatan. Penelitian ini bertujuan memprediksi keandalan turbin gas di PT Petrokimia Gresik menggunakan jaringan saraf tiruan (JST) dengan algoritma Levenberg-Marquardt, berdasarkan data historis dan pembacaan sensor. Hasil penelitian menunjukkan model JST dengan 10 node pada lapisan tersembunyi memberikan performa terbaik, dengan mean square error (MSE) sebesar 0,0023 untuk pelatihan, 0,0053 untuk validasi, dan 0,0067 untuk pengujian. Deteksi anomali mengidentifikasi 109 kejadian, dengan keandalan sistem awal sebesar 100% yang menurun hingga 91,28%. Efisiensi kompresor sistem selama beberapa periode memiliki rentang 70–96%, sementara efisiensi turbin berada pada rentang 35–46%. Model JST yang dihasilkan mampu memprediksi keandalan secara akurat dan memberikan wawasan penting untuk mendukung pemeliharaan prediktif.
===============================================================================================================================
PT Petrokimia Gresik as a petrochemical and fertilizer industry requires electrical energy to run the production process. The electrical energy consumption of PT Petrokimia Gresik in August 2021 uses eleven percent consumption from PLN and the rest uses electrical energy produced from utility department with gas turbine generator as power plant. The reliability of gas turbines is very important to consider because failures can cause operational disruptions, high costs, and safety risks. This research aims to predict the reliability of gas turbines at PT Petrokimia Gresik using artificial neural network (JST) with Levenberg-Marquardt algorithm, based on historical data and sensor readings. The results showed that the JST model with 10 nodes in the hidden layer provided the best performance, with a mean square error (MSE) of 0.0023 for training, 0.0053 for validation, and 0.0067 for testing. Anomaly detection identified 109 events, with an initial system reliability of 100% that decreased to 91.28%. System compressor efficiency over multiple periods ranged from 70-96%, while turbine efficiency was in the range of 35-46%. The resulting JST model is able to accurately predict reliability and provides important insights to support predictive maintenance.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	turbin gas, keandalan, jaringan saraf tiruan, deteksi anomali, pemeliharaan prediktif. gas turbine, reliability, artificial neural network, anomaly detection, predictive maintenance.
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T59.7 Human-machine systems. T Technology > TA Engineering (General). Civil engineering (General) > TA169 Reliability (Engineering) T Technology > TJ Mechanical engineering and machinery > TJ217.6 Predictive Control T Technology > TJ Mechanical engineering and machinery > TJ778 Gas turbines T Technology > TS Manufactures > TS174 Maintainability (Engineering) . Reliability (Engineering)
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Rafly Andra Pramudya
Date Deposited:	31 Jan 2025 02:09
Last Modified:	31 Jan 2025 02:09
URI:	http://repository.its.ac.id/id/eprint/117305

Actions (login required)

View Item