An Approach to Non-Linear Experimental Modal Analysis

It was the aim of the DYNAMITED (DYNamics AssessMent and Improvment of TEst Data) research project conducted by a European industrial and university team associated around EADS ASTRIUM and funded by the European Space Agency (ESA) to assess and improve dynamic test procedures. In the field of post test activities, the detection, identification, quantification and prediction of the non-linear behavior of space structures is of prime importance. Present evaluation methods for spacecraft vibration tests are based on the assumption of linear structural behaviour. The resonance shifts and FRF peak variations observed in the case of non-linear structural behavior are generally not reflected in practice by non-linear evaluation procedures. In order to avoid overloading of the structure during the qualification test on a shaking table the dynamic response is generally controlled at specified levels and locations by input notching. This approach generates an effectively quasi linear structural behavior at the different input levels which enables the utilization of classical linear modal extraction tools to be applied separately at each level. However, the measured dynamic responses (transmissibilities) reveal peak shifts and amplitude changes depending on the input level of the base excitation. In the paper we present an approach using three different input levels where the response levels are controlled to be constant within a narrow band frequency range around the dominant resonances. We describe the premises of a technique with the aim of predicting the responses to other than the measured input levels. This is achieved by applying interpolation and extrapolation techniques to the modal data extracted from the transmissibilities measured at three different input levels. Results are presented from an application of the technique to the vibration test data measured during a typical STM satellite structure (Structural Test Model) test campaign.