IMECE2005-80582 ACTIVE AND EVENT-DRIVEN PASSIVE MECHANICAL FAULT IDENTIFICATION IN GROUND VEHICLE SUSPENSION SYSTEMS

ABSTRACT Data interrogation methodologies are needed for identifying loads and faults in suspensions, tires, and other vehicle components to help design more durable systems and reduce the total cost of ownership. The application of passive and active data interrogation methodologies to passenger vehicle suspension systems is discussed here. For passive diagnostics, operating acceleration response data in conjunction with fundamental mechanics models are utilized. Mechanical faults in suspension components, e.g. degradation to shock, are identified using force state maps and transmissibility functions. First, it is shown that damage causes changes in the frequency characteristics of restoring forces, provided by the force state maps, which help to detect damage. Second, autoregressive nonlinear transmissibility models are used to locate faults and also characterize the degree to which faults alter nonlinear correlations in the response data. Force state maps are suited to narrow band inputs (e.g., sinusoidal) and transmissibility models are suited to broad-band inputs (e.g., random). This difference in preferential bandwidth for the two different data analysis methods motivates the selection of the diagnostic algorithm in an event-driven manner. For active diagnostics, experimental sensitivity functions, which are algebraic combinations of measured frequency response data, estimate the change in the forced response of the system with perturbation in stiffness or damping. By comparing the sensitivity functions to finite difference functions, faults can be detected, located, and quantified. The passive and active techniques are applied to experimental vehicle data and various issues (e.g., quantifying faults) are discussed