A combined neural network and model predictive control approach for ball transfer unit–magnetorheological elastomer–based vibration isolation of lightweight structures

Abstract

This study addresses the possibility of adopting semi-active magnetorheological elastomers–based isolators for protecting lightweight structures from ground vibration. The exploitation of these smart devices has the main advantage of controlling their stiffness and damping features by acting on the magnetic field generated by a coil on the basis of the actual conditions of both the lightweight structure and the surrounding environment. This allows for combining the reliability of passive devices with the benefits of active control methods. Both mechanical and control system designs could play a crucial role in the challenging problem of improving isolation performances. To solve this issue, we (i) suggest a novel ball transfer unit–magnetorheological elastomer–based isolation system prototype to obtain an improved isolation response of the lightweight structure with respect to the exclusive use of an magnetorheological elastomer and (ii) propose a novel robust combined neural network and model-predictive control approach, allowing proper functioning of the ball transfer unit–magnetorheological elastomer–based isolation system. The effectiveness of the proposed semi-active isolator in guaranteeing vibrational isolation of lightweight structures is evaluated by considering a rack cabinet composed of three storeys and subject to an El Centro earthquake. Numerical simulations confirm and disclose the efficacy of the proposed approach.