Vibration analysis of a suspension system subject to high level of measurement noise

Using only vertical acceleration measurements for the sprung and unsprung masses of a suspension system of a commercial city bus, the goal of the paper is to develop an analysis method to find the vibration modes of the mechanical system from data measured during real life operation. The identified vibration modes can be used to (in)validate first principle physical models of the system, while the identified ARMA models can be used to develop uncertainty models. The challenge in the problem is that the measurements are subject to very high level of noise due to maneuvering of the vehicle, nonlinear effects of the suspension system, vibration of the engine and the gear system, and sensor noise. Nonparametric and parametric modeling methods are applied to evaluate the quality of the measurements and find the invariant properties of the suspension system. It is shown based on multiple experiments that independently of the actual road properties and operating conditions, eigen-frequencies of some vibration modes can be determined with relatively small uncertainty, while the corresponding damping factors have varying amount of uncertainty. Comparing the results with the modes of a full car vehicle model developed based on physical considerations, it can be concluded that an identification algorithm for obtaining the parameters of the physical model must be complemented with a suitable uncertainty modeling and classification.