Abstract

This thesis concerns with the development, design optimization and control of an MRE-based semi-active seat suspension system to mitigate the whole-body vibration (WBV) transmitted to the human body. The dynamic model of a seated human-body using 4-DOF linear lumped parameter model is firstly derived. Simulation results for seat to head transmissibility (STH) using the 4-DOF model show that the WBV of a human in sitting posture is most sensitive within the frequency range of 4-12 Hz, which is confirmed by the current standard on WBV exposure assessment, ISO 2631/1:1997. The 4-DOF seated human model was then combined with the single DOF model of the seat suspension to derive the coupled 5-DOF human and seat suspension system. The 5-DOF model was utilized to obtain the optimal passive parameters (stiffness and damping) of the seat suspension to minimize the weighted average (rms) vibration acceleration input transmitted to the human operator. The frequency weighting functions, provided by both current (ISO 2631/1:1997) and old (ISO 2631/1:1985, ISO 2631/3:1985) standards, are utilized to realize the effect of weighting factors on the optimal seat suspension designs. Vibration transmissibility from the floor to the operator head in the frequency range up to 100 Hz is also evaluated using the optimal seat suspension parameters obtained using new and old standards on WBV. An adaptive magneto-rheological elastomer (MRE) vibration isolator was optimally designed to be integrated into the passive seat suspension to semi-actively provide vibration mitigation over a relatively wide range of frequency. The MRE samples were characterized using a rotational rheometer under relatively wide ranges of magnetic and mechanical loadings. Assuming that MREs operate in a linear viscoelastic region, a field-dependent phenomenological model is developed to predict the MREs’ mechanical properties (storage and loss moduli) as functions of the excitation frequency and applied magnetic field. The MRE-based vibration isolator is combined with three passive springs (two in parallel, and one in series with the isolator). The proposed configuration in this study, allows relatively large stroke of the seat suspension, while maintaining MREs samples to be operated under low deformation (<30% strain), where the nonlinear strain-softening behavior of MREs is minimum. Results showed that by increasing the applied current from zero to maximum 2 A, the equivalent stiffness of the adaptive seat suspension increased from 8.94 kN/m to 17.5 kN/m under the frequency range where human shows maximum sensitivity (4-10 Hz).
Finally, a particle swarm optimization based fuzzy neural network (PSO-FNN) controller has been designed to evaluate the close-loop performance of the proposed MRE based adaptive seat suspension under different input excitation conditions such as harmonic, random noise, harmonic superimposed with random white noise and bump shock. The simulation results show that PSO-FNN controlled seat suspension compared with the passive suspension can decrease the rms of the vibration acceleration transmitted to the operator’s head by nearly 63% and 73% under harmonic superimposed with random white noise, and bump shock excitations, respectively.