Keywords: parameter identification, mechanical properties, railway track model, experimental modal analysis, transfer and coherence function, S-form weighted least squares.

This paper reports on a parametric study performed on a railway track to estimate the mechanical properties of rail and substructure. This study proposes a practical method for identification of an efficient track model. The track parameters are characterized on an ordinary Euler-Bernoulli beam with discrete supports. The track is modelled using an assumed mode method. The modal characteristics such as damping factors and mode shapes of structures are generally estimated by experimental modal analysis. This method could proceed from the frequency response function to the structure's physical (mass, stiffness and damping) values via a transfer function. However the expression of the transfer function for the proposed model is rather difficult and varies depending on number of spans. Therefore the extraction of the parameters without a need to determine the transfer function is presented, which was never reported in previous investigations and uses some numerical techniques of optimization. Least squares techniques have been applied to identify the modal parameters of complex mechanical structures. This method is based on searching the best estimation of parameters set which minimizes the difference between the measured and the simulated frequency response function. Since the number of physical parameters is high, the proposed estimation contains a multi-step process where an S-form weighted least squares technique is used at the last stage of optimization. The experimental transfer and coherence functions are calculated by averaging several signals of hammer and track excitation on the desired frequency range. Two configuration types for accelerometer placement and impact position on the track are presented; collocated and non-collocated measurements. The relevance of each parameter in the general model is determined by demonstrating its variation on the final results. Lastly a comparison between an experimental excitation and identified model response for the same impact is given.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £65 +P&P)