Optical aerodynamic measurements of hypersonic free-flight using Bayesian state estimation

Optical aerodynamic measurements of hypersonic free-flight using Bayesian state estimation

Free-flight aerodynamic measurements via optical tracking have recently gained popularity due to model simplicity, unconstrained body motion, and lack of flow interference from supporting stings. However, the signal to noise ratio of high-speed optical aerodynamic measurements is unfavourable, requiring significant data smoothing for meaningful results. This work presents a new method of optical six-degree-of-freedom aerodynamic measurements founded on Bayesian state estimation. A sigma-point Kalman filter and Rauch-Tung-Striebel smoother computes the state estimate at each time using images from two high-speed camera images and a kinematic process model. The pixel location of markers detected in camera images are each treated as individual measurements, and are related to the body position and pose in local coordinates using calibrated camera models. This new method allows aerodynamic measurements using two un-synchronised cameras at arbitrary positions. As a demonstration, this work measures aerodynamics of a cube in Mach 6 flow in the University of Southern Queensland’s TUSQ wind tunnel. Images are taken with an existing z-type schlieren camera assembly and one additional standard high-speed camera over approximately 30 ms. During three tests the cube has low initial rotation; in a fourth run we rotate the cube to approximately 600 RPM prior to release in the flow. Measured force and moment coefficients are compared to simulations using compressible flow computational fluid dynamics conduced as part of this work. The two result sets show strong agreement, and high tracking accuracy is demonstrated by superimposing the body motion over camera images. This new method may allow wind tunnels to conduct optical aerodynamic measurements with less modification and equipment than traditional stereoscopic methods. Moreover, the use of Bayesian state estimation algorithms is shown to be effective at reducing otherwise-large measurement noise.

Files associated with this item cannot be displayed due to copyright restrictions.