Assessment of nonlinear distortions in modal testing and analysis of vibrating automotive structures

In this paper, new developments for the nonparametric processing of modal test data are presented. Classically, random noise signals are applied to deal with possible nonlinear distortions during frequency response function measurements of linear dynamic systems. However, the use of multisine excitation signals allows the engineer to control much more his experiments. First of all, the nonparametric estimation of multivariable frequency response functions can be more easily based on an "errors-in-variables" stochastic framework. In addition, the application of a well-chosen multisine excitation permits improvement of the data quality, as well as the detection, qualification and quantification of nonlinear distortions during FRF measurements. To make the presented techniques available for multi-input modal testing, attention is paid to the design of optimal multi-input excitations by maximizing the Fisher information matrix as well as minimizing the crest factor of the applied excitation. (c) 2006 Elsevier Ltd. All rights reserved.