Experimental Modal Analysis Using Impact Hammer Testing with Random Forest-Based Prediction of Magnetorheological Elastomer Dynamics

Abstract

This study presents a novel integration of impact hammer-based experimental modal analysis with Random Forest Regression (RFR) to rapidly characterise the frequency-domain dynamic behaviour of Carbonyl Iron Particle (CIP)-based Magnetorheological Elastomers (MREs) under varying magnetic fields. Using only applied current and excitation frequency as input features, the RFR model predicts FRF amplitude, phase, and coherence with R2 values exceeding 0.96 across both low-frequency (0-70 Hz) and high-frequency (> 70 Hz) regimes. This hybrid experimental-computational framework significantly reduces the number of repeated tests required, enabling faster parametric studies and paving the way for real-time, AI-enhanced tuning of smart vibration isolation systems. © Published under licence by IOP Publishing Ltd.