Ana C. Meinicke, Carlos E. S. Cesnik
DOI Number: N/A
Conference number: IFASD-2024-105
The ability to estimate aerodynamic loads and flight parameters in flight using only internal sensors is desirable for guidance, navigation, and control. This is particularly true for
harsh aerothermal environments experienced in supersonic and hypersonic flight, but it is also applicable to any other flight regime. The vehicle-as-a-sensor concept is a novel nonintrusive sensing strategy for airframes in flight. It leverages the internal measurement of the deformed state of the vehicle, its temperatures, and its accelerations to infer its aerodynamic state. An inverse model consisting of a strain-to-load, or strain/temperature-to-load, neural network trained with static elastic solutions of a detailed finite element model of a slender hypersonic vehicle is considered here. The deformed state of the vehicle is assumed to be measured with high-density strain and temperature measurements on its internal surface, enabled by the application of continuous fiber optic sensors. Constraints pertaining to the use of fiber optic sensors are considered, where limitations such as applying the optical cables only on the internal surface of the vehicle, and the necessity of choosing adequate strain component directions to recover the aerodynamic state with sufficient accuracy are accounted for. Inverse models considering selected sensors based on the physics involved in the problem are shown to successfully recover pressure distribution on the outer surface of the vehicle.