Abstract
Keywords: Fault detection and isolation, structural systems, suspensions, modal updates, system identification.
This study aims at forming a general framework for the identification of suspension faults, using vibration measurements and on the fly updates for the modal characteristics of the system. Under the assumption that a suspension fault is reflected in the vertical dynamics performance, a discrete model is identified directly from data, using parameter estimation, in the fault-free stage and the nominal values of the modal quantities are computed. The estimated model serves as a parallel computational system that, under no faults, produces the same vibration outputs (with a standard error) as the vehicle. Thus, any deviation among them becomes a feature of possible fault occurrence. The latter is being identified and isolated via the change of the new (fault) modes of the system.