A neural network-informed self-aware deployable structure with application to phased array antennas

Abstract

State of the art radio frequency (RF) arrays are growing larger in pursuit of increased signal-to-noise ratio. In support of this goal, elaborate forms of metrology are being developed to support the increased footprints. This work provides a unique solution to fulfill the metrology requirements of large-scale deployable RF antennas through the implementation of neural network demodulation of fiber optic strain sensors. The fiber optics are patterned with Fiber Bragg Gratings (FBGs) to encode strain on to back-reflected shifts in the wavelength of incident light. Experiments show the neural network can predict the deformation of a test structure within single millimeters for small amplitude motions. Therefore, the current technique meets the required lambda/20 precision needed for large scale deployable RF arrays operating at S-band or longer wavelengths.

Dynamic deployment sensing of thin-shell composite structures with fiber Bragg gratings

“…14 However, FBG sensing in ultra-thin composite shells (≤ 200µm) is still in early stages of investigation 15 while real time sensing of complex deployment motions remains limited in scope. 16 FBGs in optical fibers have the potential to entirely alter the state of the art in remote deployment monitoring. They are more mass, power, and volume efficient when compared to traditional externally mounted camera arrays while not being sensitive to external lighting conditions.…”

“…Some work has even been done that leverages neural networks to extrapolate 3D displacements of a radio-frequency antenna panel from arrays of 1D FBG strains. 20 Other flexible embedded sensors (e.g. metal strain gauges, soft electronics 21 ) are not as scalable due to each sensor requiring its own set of conductors that impede the composite structure's stowability and deployment behavior.…”