#TITLE_ALTERNATIVE#

• Main Author: NUGROHO (NIM 13303067), FIRMAN
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/10453
• Institution: Institut Teknologi Bandung

PT. XYZ is one of chemical process industry that controller system design become a part from the process itself. One part of PT. XYZ is synthesis reactor plant having nonlinear characteristic. According with process nonlinear characteristic itself, urea synthesis reactor plant needs nonlinear controller to rely on. The superiority of neural network structure that has nonlinear function in activation node could play a role as controller to handle complexity of chemical process. This research is concerned with a development of feedback control method designed using neural network. The adaptive neuro-fuzzy approach is implemented to model the dynamic responds of urea plant reactor where a neural network controller controls the process dynamics while model validation criteria is reached. A hybrid learning rule using genetic algorithm is also used to minimize the difference between the actual and a given desired trajectory and maximize another process variable where two of that variable is interacted. Simulation results show that neural network control performs well to track the given desired set-points besides to optimize another process variable. Performance comparison was also made between the designed and PID cascade controller.