Aplikasi Metode Neuro-Fuzzy Pada Sistem Pengendalian Antisurge Kompresor

Imam Abadi, Aulia Siti Aisjah, Riftyanto N.S.




Abstract


Operating condition that unstabil to compressor is called surge, which defined as self oscillations between pressure and flow, that can cause serious damage to compressor. Because of this problem, it is necessary to design of antisurge compressor control. Using conventional control like PI (Proportional and Integral) controller, effective only for certain condition but uneffective for condition with non linier systems. To handle this problems, it is proposed using Neuro-Fuzzy controller. Application of these system will improve performance indicator of PI controller applied in combination Neural Network and Fuzzy Logic Control to get robust performance system. For respon transient experiment, with differential pressure ( P) 37.91 kg/cm2, Neuro-Fuzzy controller result better performance corcerning to PI controller with settling time 7.3 seconds, maximum overshoot 11.6223 %, error percent 0.0563013 % and error steady state 0.0213438 kg/cm2. Whereas, PI controller result settling time 22.3 seconds, maximum overshoot 17.8996 %, error percent 0.185294 % and error steady state 0.0702449 kg/cm2. For respon transient system simulation with noise and respon transient simulation with noise and load, Neuro-Fuzzy controller also shows better performance with settling time, maximum overshoot and error steady state smaller than PI controller.



Abstract in Bahasa Indonesia :

Salah satu bentuk kondisi operasi yang tidak stabil pada kompresor adalah kondisi yang disebut surge, dimana terjadi ketidakstabilan antara aliran dan tekanan yang dapat menyebabkan kerusakan yang serius pada kompresor. Oleh karena itu akan dirancang suatu pengendalian anti surge pada kompresor. Perancangan dilakukan dengan pengendali neuro-fuzzy, yang menggabungkan kemampuan belajar Neural Network dengan kemampuan pengambilan keputusan pada fuzzy. Dengan harapan akan diperoleh performansi yang lebih baik daripada pengendali PI yang nantinya dijadikan sebagai pembanding. Pada simulasi uji respon sistem, dengan setpoint beda tekanan ( P) sebesar 37,91 kg/cm2, pengendali Neuro-Fuzzy menghasilkan performansi yang lebih baik dari pengendali PI dengan settling time 7,3 detik, maksimum overshoot 11,6223 %, persen error 0,0563013 % dan error steady state 0,0213438 kg/cm2. Sedangkan pengendali PI menghasilkan settling time 22,3 detik, maksimum overshoot 17,8996 %, persen error 0,185294 % dan error steady state 0,0702449 kg/cm2.

Kata kunci: antisurge, simulasi, neuro-fuzzy, settling time, maksimum overshoot, persen error, error steady state.


Keywords


antisurge, simulation, neuro-fuzzy, settling time, maximum overshoot, error percent, steady state error.

Full Text: PDF

The Journal is published by The Institute of Research & Community Outreach - Petra Christian University. It available online supported by Directorate General of Higher Education - Ministry of National Education - Republic of Indonesia.

©All right reserved 2016.Jurnal Teknik Elektro, ISSN: 1411-870X

 

shopify traffic stats
View My Stats




Copyright © Research Center Web-Dev Team