Abstracts Engineering

Development of a nonlinear method for the identification of system parameters from observed time series

by Fabien. Seyvet

The essence of this work consists of developing a robust algorithm for the estimation of Lyapunov exponents and complex exponents from an observed time series. These characteristic exponents are representative of a nonlinear system's behaviour, and enable flight flutter data analysis to be performed, the methodology developed was found to be inadequate for short and noisy data sets. Attention is therefore focused on two methods to alleviate this deficiency, one to reduce the noise that contaminates the time series, and another to concatenate several short experimental data sets. The properties of the different methodologies are assessed on simulated data, and accurate system parameter identification is obtained. An analysis on real flight test data is then performed with the aim of obtaining modal damping and frequency for the CF-18 aircraft in flight conditions. All the numerical results are discussed, and the effects of noise and the length of data are investigated.