An investigation of the component of response spectrum uncorrelated with input

Abstract

Abstract For a linear system, by definition, the response spectrum is related to the input spectrum through a linear transformation. An investigation of the component of the response spectrum disobeying such a correlation gives way to a non-parametric algorithm which can help detect nonlinearities in the system being tested. The uncorrelated component thus obtained contains information pertinent to both structural nonlinearities and any lack of linearity resulting from the digital signal processing (DSP) techniques employed. The current paper discusses an approach that combines cyclic averaging with the aforementioned detection scheme to help segregate the uncorrelated response components based on leakage from those due to the dynamic characteristics of the system. This scheme is then shown to be successful on numerical and experimental data sets with representative examples for both linear and nonlinear systems.