Optimasi Desain Magnetorheological Valve Tipe Meandering untuk Rancangan Peredam Truk dengan Metode Metaheuristic
Muhammad Nasrullah, Ir. Irfan Bahiuddin, S.T., Ph.D., IPM, GRCE
2025 | Tugas Akhir | D4 TEKNIK PENGELOLAAN DAN PERAWATAN ALAT BERAT
Penelitian ini mengoptimalkan desain Magnetorheological (MR) valve
tipe meandering untuk aplikasi peredam truk melalui pendekatan model
prediktif yang menggabungkan Artificial Neural Network (ANN) dan Chaotic
Particle Swarm Optimization (CPSO). Model ANN dikembangkan untuk
memprediksi kerapatan fluks magnetik berdasarkan data simulasi Finite
Element Method (FEM), dengan tingkat akurasi tinggi (MSE 0,00002; R²
0,9997). Zona radial luar menunjukkan performa prediktif terbaik dibandingkan
zona annular luar dan radial dalam, menunjukkan ANN dalam memodelkan hubungan
nonlinier antara parameter geometri, arus listrik, dan fluks magnetik. Model
ini digunakan sebagai fitness function dalam optimasi multi-objective
function melalui CPSO, dengan tujuan meminimalkan pressure drop viscous
dan memaksimalkan pressure drop total. Bobot objektif optimal sebesar
0,6 dan 0,1 masing-masing ditetapkan untuk tujuan tersebut, menghasilkan
konfigurasi optimal berupa annular gap 1,20 mm dan radial gap
0,38 mm yang dinilai mudah untuk difabrikasi. Desain ini menghasilkan pressure drop viscous 0,9674 MPa, yield 6,6397 MPa, dan total 7,6071 Mpa
dengan gaya redaman maksimum 7.1506 N, serta rentang kendali 6,863, yang
menunjukkan adaptabilitas tinggi untuk aplikasi peredam truk. CPSO menunjukkan
kestabilan dan efisiensi komputasi dengan nilai fungsi objektif -0,1802 ±
0,0000 dalam sepuluah run dan waktu konvergensi 13–16 menit. Pendekatan
prediktif ANN-CPSO terbukti efektif, efisien, dan robust dalam
menghasilkan desain MR valve dengan kinerja redaman optimal untuk sistem
peredam truk.
This
study optimizes the design of a meandering-type Magnetorheological (MR) valve
for truck damper applications through a predictive modeling approach that
integrates an Artificial Neural Network (ANN) with Chaotic Particle Swarm
Optimization (CPSO). The ANN model was developed to predict magnetic flux
density based on Finite Element Method (FEM) simulation data, achieving high
accuracy (MSE = 0.00002; R² = 0.9997). The outer radial zone demonstrated the
best predictive performance compared to the outer annular and inner radial
zones, highlighting the ANN’s capability in modeling the nonlinear relationship
between geometric parameters, electric current, and magnetic flux. This model
was employed as the fitness function in a multi-objective optimization using
CPSO, aiming to minimize viscous pressure drop while maximizing total pressure
drop. Optimal objective weights of 0.6 and 0.1 were assigned to these goals,
resulting in an optimal configuration with an annular gap of 1.20 mm and a
radial gap of 0.38 mm, which are considered feasible for fabrication. The
design produced a viscous pressure drop of 0.9674 MPa, a yield pressure drop of
6.6397 MPa, and a total pressure drop of 7.6071 MPa, corresponding to a maximum
damping force of 7,150.6 N and a control range of 6.863, indicating high
adaptability for truck damper applications. CPSO exhibited stability and
computational efficiency with an objective function value of –0.1802 ± 0.0000
across ten runs and a convergence time of 13–16 minutes. The proposed ANN–CPSO
predictive approach proved to be effective, efficient, and robust in producing
MR valve designs with optimal damping performance for truck suspension systems.
Kata Kunci : Magnetorheological valve, Artificial Neural Network (ANN), Chaotic Particle Swarm Optimization (CPSO), optimasi desain, peredam truk.
