Perancangan Sistem Kontrol Temperatur pada Heat Exchanger Berbasis Artificial Neural Network (ANN) untuk Self-Tuning PID

Nyhayatuzzain, Rohil (2025) Perancangan Sistem Kontrol Temperatur pada Heat Exchanger Berbasis Artificial Neural Network (ANN) untuk Self-Tuning PID. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009211067-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Heat exchanger adalah alat penukar panas yang dikategorikan sebagai sistem nonlinier dan sensitif terhadap gangguan. Agar dapat beroperasi secara optimal, diperlukan sistem kontrol yang andal. Pengendali Proportional-Integral-Derivative (PID) adalah pengendali konvensional yang umum digunakan dalam proses industri, termasuk pada sistem heat exchanger, namun kinerjanya sering tidak stabil akibat perubahan dinamika proses dan gangguan yang menyebabkan perlunya tuning secara berulang pada kondisi tersebut. Penelitian ini mengembangkan sistem kontrol PID yang dikombinasikan dengan Atificial Neural Network (ANN) untuk self-tuning PID atau disebut ANN-PID pada temperatur outlet shell heat exchanger. Heat exchanger dimodelkan dengan persamaan diferensial dan sistem kontrol yang disimulasikan pada Simulink. Model neural network yang dirancang terdiri dari 1 input layer, 1 hidden layer, dan 1 output layer, dengan jumlah neuron pada hidden layer yaitu 5. Input layer dan output layer terdiri dari 3 neuron yang disesuaikan dengan kebutuhan yaitu parameter gain untuk PID. Metode Backpropagation Levenberg-Marquardt digunakan dalam pelatihan data untuk neural network. Pengendali PID konvensional dan ANN-PID diuji dengan tracking setpoint, uji closed-loop, dan uji terhadap disturbance. Hasil simulasi menunjukkan bahwa pengendali ANN-PID memiliki performansi lebih baik dibandingkan pengendali PID konvensional. Pada uji closed-loop pengendali ANN-PID dapat mengurangi waktu untuk mencapai setpoint 178.509 detik lebih cepat daripada pengendali PID. Pada uji tracking setpoint pengendali ANN-PID mampu merespon perubahan setpoint ≥ 30% secara lebih efektif baik pada uji kenaikan maupun penurunan setpoint.
===================================================================================================================================
Heat exchanger is a heat transfer device with a nonlinear system that is sensitive to disturbances. Conventional PID controllers are widely used in industrial processes, including heat exchanger systems, but their performance often becomes unstable due to dynamic process changes and disturbances. Therefore, frequent repeatedly retuned is required under such conditions. In this study, a PID control system combined with a neural network for self-tuning PID, known as ANN-PID, was designed for a temperature outlet shell heat exchanger. The heat exchanger was modeled using differential equations, and the control system was simulated in Simulink. The designed neural network model consists of 1 input layer, 1 hidden layer, and 1 output layer, with 5 neurons in the hidden layer. The input and output layers each consist of 3 neurons corresponding to the PID gain parameters. The Levenberg-Marquardt Backpropagation method was used for training the neural network using Matlab-Simulink. Both conventional PID and ANN-PID controllers were tested using setpoint tracking, closed-loop, and disturbance scenarios. The results show that the ANN-PID controller performs better than the conventional PID controller. In the closed-loop test, the ANN-PID controller reduced the time to reach the setpoint by 178.509 seconds compared to the PID controller. In the setpoint tracking test, the ANN-PID responded more effectively to setpoint changes of ≥ 30%, both for increases and decreases. Under disturbance conditions, the ANN-PID controller maintained lower error and achieved faster recovery than the conventional PID controller.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Artificial Neural Network (ANN), Hea Exchanger, Kontrol Adaptif, Kontrol Temperatur, Self-tuning PID. ========================================================================================================================== Adaptive Control, Artificial Neural Network (ANN), Heat Exchanger, Self-tuning PID, Temperature Control
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TJ Mechanical engineering and machinery > TJ217 Adaptive control systems T Technology > TJ Mechanical engineering and machinery > TJ263 Heat exchangers T Technology > TJ Mechanical engineering and machinery > TJ223 PID controllers
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Rohil Nyhayatuzzain
Date Deposited:	05 Aug 2025 03:57
Last Modified:	05 Aug 2025 03:57
URI:	http://repository.its.ac.id/id/eprint/127179

Actions (login required)

View Item