Adapted linear decimation procedures for TFR analysis of non-stationary vibration signals of vehicle suspensions

The paper addresses the adaptation of linear decimation procedures (LDP) for time-frequency representation (TFR) analysis of non-stationary vibration signals of vehicle suspension. It is extremely important for designing suspensions elements to achieve proper characteristics in term of propagation of vibration. One of the major factors for the identification of defined frequency components in terms of resonance phenomena. Given that a car is a multi-DOF dynamic system, the meaning of suspension becomes even more important due to its functions such as: isolating and damping vibrations and assuring constant contact of wheel to road. Many of the important information of the vibration phenomena and dynamic system properties refer to the frequency domain. Given the volatile nature of vibrations resulting from the nonlinear characteristics of suspension elements of the vehicle and the random nature of dynamic interactions, determining the frequency characteristics is a very complex issue. For non-stationary signals the time-frequency transformation methods are recommended. But applying transformation algorithms during signal processing is not time-effective. Thus the new method, developed in AGH University of Science and Technology, for transferring non-stationary signals into pseudo-stationary signals has been tested on vibrations recorded on a real car suspension system. The paper presents the research aimed at examining the possibility of adapting the linear decimation procedure for identification of characteristic components of non-stationary vibration signals in a car. Also presented is the effect of applying of the LDP for the analysis of TFR of the vibration signal on examples of vehicle suspension and gear-meshing. There is also short description of hardware implementation of the LDP.