Evolutionary System Identification of Flight Motion Simulators

Abstract

This study focuses on a system identification technique for a hydraulic based flight motion simulator (FMS). Because nonlinearities such as stiction and nonlinear orifice flow rate are often inherent in hydraulic systems, the system identification of the FMS can be very challenging. Simple non-parametric methods such as frequency response and step response methods have previously been used to estimate the parameters of the FMS model. These methods can sometimes provide inadequate results since the parameters are timevariant. Sophisticated parametric methods such as Extended Kalman filter, Unscented Kalman filter and Particle filter have also been used to estimate dynamic responses of a nonlinear system. However, the performance of those filters, in terms of convergence stability and error, are subjected to the initial conditions of the filters. In practice, tuning of those initial conditions can be time consuming. This paper highlights a novel evolutionary system identification approach which takes advantage of the knowledge of the physical system, the power of a non-parametric method such as frequency response, and the simplicity of the linear Kalman filter. Experimental results have shown that the estimated response of the FMS matches very well with the measured response despite changes in the operating conditions.