Analisa Perbandingan Estimasi Parameter Recurcive Least Square (RLS) dan Pendekatan Jaringan Syaraf Tiruan pada Self-Tunning Regulator (STR) untuk Pengaturan Tekanan Process RIG 38-714

Gunawan, Febry Angga (2018) Analisa Perbandingan Estimasi Parameter Recurcive Least Square (RLS) dan Pendekatan Jaringan Syaraf Tiruan pada Self-Tunning Regulator (STR) untuk Pengaturan Tekanan Process RIG 38-714. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 07111340000104_Undergraduate_Theses.pdf]
Preview
Text
07111340000104_Undergraduate_Theses.pdf

Download (4MB) | Preview

Abstract

Pada sistem pengaturan proses sering terjadi proses perubahan beban. Perubahan tersebut dapat menimbulkan perubahan dinamika dari sistem. Di industri kontroler yang umum digunakan adalah PI atau PID konvensional. Namun, karena proses perubahan beban kontroler itu kurang bisa memenuhi spesifikasi. Pengaturan adaptif adalah salah satu metode pengaturan yang mana kontroler dapat memberikan respon memodifikasi perilakunya karena perubahan dinamika dari proses dan karakteristik dari gangguan. Self-Tunning Regulator (STR) adalah salah satu skema pengaturan adapatif. Estimasi parameter adalah salah satu bagian dari STR. Pada penelitian ini implementasi STR dengan estimasi parameter Recursive Least Square (STR RLS) dan pendekatan jaringan syaraf tiruan (STR NN) dilakukan pada sistem pengaturan tekanan di Process Rig 38-741. Hasil pengujian menunjukan pada kondisi beban nominal STR NN dengan learning rate = 25 memiliki performa yang paling mendekati dengan hasil desain dengan nilai overshoot kecil dan settling time paling cepat dan sesuai. Pada pengujian dengan kondisi adanya perubahan beban STR NN dengan learning rate = 20 menunjukan performa terbaik terhadap semua kriteria yang dipakai. Sementara untuk pengujian beban nominal terhadap variasi set-point STR NN dengan learning rate = 10 menunjukan performa terbaik pada semua kriteria yang dipakai.
============================================================================================================
In the process control system often occurs the process of changing the load. Such changes may lead to changes in the dynamics of the system. In most industries the controller used is a conventional PI or PID. However, because the process of changing the controller is less able to meet the specifications. Adaptive control is one of the control methods that the controller can respond modifies its behavior due to changes in the dynamics of the process and the characteristics of the disturbance. Self-Tunning Regulator (STR) is one of the adapatif control schemes. The parameter estimation is one part of STR. In this research, the implementation of STR using parameter estimation of Recursive Least Square (STR RLS) and artificial neural network (STR NN) are applied to pressure control system in Process Rig 38-741. The test results show that for nominal load condition STR NN with learning rate = 25 has the best performance close to the design result with small overshoot value and the fastest and most appropriate settling time. Implementation for condition of the load change was existing, STR NN with learning rate = 20 shows the best performance against all the criteria used. While for implementation for the nominal load condition against set-point variations STR NN with learning rate = 10 shows the best performance on all criteria used.

Item Type: Thesis (Other)
Additional Information: RSE 629.836 Gun a 3100018076243
Uncontrolled Keywords: Self-Tunning Regulator, Pressure Process Rig 38-714, Recursive Least Square, Artificial Neural Network
Subjects: Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
T Technology > TJ Mechanical engineering and machinery > TJ223 PID controllers
T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL521.3 Automatic Control
Divisions: Faculty of Electrical Technology > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User: Gunawan Febry Angga
Date Deposited: 02 Jan 2019 02:18
Last Modified: 19 Jan 2024 15:13
URI: http://repository.its.ac.id/id/eprint/53100

Actions (login required)

View Item View Item