Analisis Performa Sistem Kendali BPNN Terhadap ANFIS Pada Sistem Suspensi Aktif Model Seperempat Kendaraan (Quarter-Car)

Rosari, Agnes Cahyaning (2023) Analisis Performa Sistem Kendali BPNN Terhadap ANFIS Pada Sistem Suspensi Aktif Model Seperempat Kendaraan (Quarter-Car). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02111940000208-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (4MB) | Request a copy

Abstract

Sistem suspensi merupakan komponen penting dalam menjaga kenyamanan dan keamanan berkendara khususnya pada permukaan jalan yang tidak rata. Sistem suspensi terus dikembangkan, salah satunya suspensi aktif yang dianggap mampu memberikan kualitas berkendara paling baik dibandingkan dengan suspensi lainnya. Suspensi aktif menambahkan aktuator hidrolik, elektromagnetik, atau lainnya yang dapat meminimalkan getaran akibat permukaan jalan yang tidak rata. Aktuator dapat menghasilkan gaya untuk mengendalikan suspensi dengan bantuan sistem kendali. Berbagai sistem kendali telah dikembangkan untuk meminimalkan getaran akibat permukaan jalan yang tidak rata. Salah satunya berbasis Neural Network seperti Backpropagation Neural Network (BPNN). Dalam penelitian ini, performa kendali BPNN akan dianalisis dan dikomparasi dengan kendali ANFIS berdasarkan RMS respon head acceleration pengemudi. Performa tersebut dapat menentukan tingkat kenyamanan berkendara yang sesuai dengan ISO 2631. Model quarter-car digunakan sebagai dinamika sistem suspensi kendaraan dengan diberikan gangguan profil jalan polisi tidur dan jalan acak. Parameter input kendali BPNN berupa error dan derivative error. Sementara parameter output berupa actuator force. Parameter tersebut diperoleh dari simulasi kendali PID. Penelitian dilakukan menggunakan software MATLAB R2022b serta Simulink. Hasil dari penelitian ini diketahui bahwa BPNN dapat meminimalkan respon head acceleration pada sistem suspensi aktif sesuai standar ISO 2631 dengan net terbaik BPNN pada profil jalan polisi tidur yaitu 2 hidden layer, 5 neuron, dan fungsi aktivasi tansig dengan MSE 0.00056189. Net terbaik BPNN pada profil jalan acak memiliki 4 hidden layer, 5 neuron, dan fungsi aktivasi tansig dengan MSE 0.000028761. Kendali BPNN menghasilkan RMS sebesar 0.4012 dengan persentase reduksi 84.80% pada jalan polisi tidur, dan RMS sebesar 0.3764 dengan persentase reduksi 77.08% pada jalan acak. Sementara kendali ANFIS menghasilkan RMS sebesar 0.2991 dengan persentase reduksi 88.67% pada jalan polisi tidur, dan RMS sebesar 0.3574 dengan persentase reduksi 78.24% pada jalan acak. Berdasarkan nilai RMS yang diperoleh dan sesuai ISO 2631, kendali BPNN memiliki tingkat kenyamanan sedikit tidak nyaman baik pada jalan polisi tidur maupun jalan acak, dengan batas waktu paparan getaran di bawah 16 jam. Sementara kendali ANFIS memiliki tingkat kenyamanan sangat nyaman pada jalan polisi tidur dan sedikit tidak nyaman pada jalan acak, dengan batas waktu paparan getaran masing-masing di bawah 24 jam dan 16 jam.
===============================================================================================================================
The suspension system plays an essential role in maintaining driving comfort and safety, especially on uneven road surfaces. The suspension system is continuously being developed with active suspension being considered to provide the best ride quality compared to other suspensions. Active suspension adds hydraulic, electromagnetic, or other actuators that can reduce vibrations on uneven road surfaces. The actuator will generate forces to control the suspension with the help of a control system. Various control systems have been developed to reduce vibrations including Neural Network-based systems such as Backpropagation Neural Network (BPNN). In this research, the performance of BPNN controller will be analyzed and compared with ANFIS controller based on the RMS of driver’s head acceleration response. The performance will determine the level of driving comfort according to ISO 2631. A quarter-car model is used to represent the dynamics of the vehicle suspension system, considering the disturbances of single-bump and random road profiles. The input parameters for BPNN controller are the error and derivative error, while the output parameter is the actuator force. These parameters are obtained from PID control simulation. The research is performed using MATLAB R2022b and Simulink. The results indicate that BPNN can reduce head acceleration responses in active suspension systems according to ISO 2631 standards, with the best BPNN network on single bump road profile having 2 hidden layers, 5 neurons, and tansig activation function with MSE of 0.00056189. While the best BPNN network for random road profile has 4 hidden layers, 5 neurons, and tansig activation function with MSE of 0.000028761. The BPNN control has a RMS of 0.4012 with a reduction percentage of 84.80% on the single bump road, and RMS of 0.3764 with a reduction percentage of 77.08% on the random road. Meanwhile, the ANFIS control has RMS of 0.2991 with a reduction percentage of 88.67% on the single bump road and RMS of 0.3574 with a reduction percentage of 78.24% on the random road. Based on the RMS values, the BPNN control exhibits a little uncomfortable comfort level for both road profiles with the vibration exposure times below 16 hours. On the other hand, ANFIS control demonstrates a very comfortable comfort level on the single bump road and a little uncomfortable comfort level on the random road, with vibration exposure times below 24 hours and 16 hours respectively.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Backpropagation Neural Network, ISO 2631, Quarter-Car, Sistem Suspensi Aktif, Active Suspension System, Backpropagation Neural Network, ISO 2631, Quarter-car
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TJ Mechanical engineering and machinery > TJ213 Automatic control. T Technology > TJ Mechanical engineering and machinery > TJ217 Adaptive control systems T Technology > TL Motor vehicles. Aeronautics. Astronautics > TL257 Springs and suspension
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Mechanical Engineering > 21201-(S1) Undergraduate Thesis
Depositing User:	Agnes Cahyaning Rosari
Date Deposited:	23 Aug 2023 06:34
Last Modified:	23 Aug 2023 06:34
URI:	http://repository.its.ac.id/id/eprint/103067

Actions (login required)

View Item