scholarly journals Phase-to-pattern inverse design paradigm for fast realization of functional metasurfaces via transfer learning

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Ruichao Zhu ◽  
Tianshuo Qiu ◽  
Jiafu Wang ◽  
Sai Sui ◽  
Chenglong Hao ◽  
...  

AbstractMetasurfaces have provided unprecedented freedom for manipulating electromagnetic waves. In metasurface design, massive meta-atoms have to be optimized to produce the desired phase profiles, which is time-consuming and sometimes prohibitive. In this paper, we propose a fast accurate inverse method of designing functional metasurfaces based on transfer learning, which can generate metasurface patterns monolithically from input phase profiles for specific functions. A transfer learning network based on GoogLeNet-Inception-V3 can predict the phases of 28×8 meta-atoms with an accuracy of around 90%. This method is validated via functional metasurface design using the trained network. Metasurface patterns are generated monolithically for achieving two typical functionals, 2D focusing and abnormal reflection. Both simulation and experiment verify the high design accuracy. This method provides an inverse design paradigm for fast functional metasurface design, and can be readily used to establish a meta-atom library with full phase span.

2021 ◽  
Vol 35 (11) ◽  
pp. 1334-1335
Author(s):  
Sawyer Campbell ◽  
Eric Whiting ◽  
Ronald Jenkins ◽  
Pingjuan Werner ◽  
Douglas Werner

Phase-gradient metasurfaces enable designers to tailor the behavior of electromagnetic waves at surfaces by exploiting the generalized form of Snell’s law. This ability has led to the investigation of metalenses which have the potential to significantly reduce the size, weight, and power (SWaP) of conventional optical systems. While traditional lenses are made from individual glasses, metalenses are comprised of patterned meta-atom unit Cells which are arranged in such a way so as to give the metalens its desired behavior. Therefore, any metalens’s performance is ultimately determined by that of its underlying unit cell components. However, designing meta-atoms that simultaneously achieve high performance over wide frequency bandwidths and fields-of-view is an extremely challenging problem that is best addressed with powerful optimization and inverse-design techniques.


Author(s):  
Mehrdad Zangeneh ◽  
Damian Vogt ◽  
Christian Roduner

In this paper the application of 3D inverse design code TURBOdesign−1 to the design of the vane geometry of a centrifugal compressor vaned diffuser is presented. For this study the new diffuser is designed to match the flow leaving the conventional impeller, which is highly non-uniform. The inverse method designs the blade geometry for a given specification of thickness and blade loading distribution. The paper describes the choice of loading distribution used in the design as well as the influence of the diffuser inlet flow distribution on the vane geometry and flow field. The flow field in the new diffuser is analysed by a 3D viscous flow code and the result is compared to that of the conventional diffuser. Finally the results of testing the stage performance of the new diffuser is compared with that of the conventional stage.


Author(s):  
Benedikt Roidl ◽  
Wahid Ghaly

A new dual-point inverse blade design method was developed and applied to the redesign of a highly loaded transonic vane, the VKI-LS89, and the first 2.5 stages of a low speed subsonic turbine, the E/TU-4 4-stage turbine that is built and tested at the university of Hannover, Germany. In this inverse method, the blade walls move with a virtual velocity distribution derived from the difference between the current and the target pressure distributions on the blade surfaces at both operating points. This new inverse method is fully consistent with the viscous flow assumption and is implemented into the time accurate solution of the Reynolds-Averaged Navier-Stokes equations. An algebraic Baldwin-Lomax turbulence model is used for turbulence closure. The mixing plane approach is used to couple the stator and rotor regions. The dual-point inverse design method is then used to explore the effect of different choices of the pressure distributions on the suction surface of one or more rotor/stator on the blade/stage performance. The results show that single point inverse design resulted in a local performance improvement whereas the dual point design method allowed for improving the performance of both VKI-LS89 vane and E/TU-4 2.5 stage turbines over a wide range of operation.


2015 ◽  
Vol 112 (24) ◽  
pp. 7449-7453 ◽  
Author(s):  
Daniel M. Sussman ◽  
Yigil Cho ◽  
Toen Castle ◽  
Xingting Gong ◽  
Euiyeon Jung ◽  
...  

We use a regular arrangement of kirigami elements to demonstrate an inverse design paradigm for folding a flat surface into complex target configurations. We first present a scheme using arrays of disclination defect pairs on the dual to the honeycomb lattice; by arranging these defect pairs properly with respect to each other and choosing an appropriate fold pattern a target stepped surface can be designed. We then present a more general method that specifies a fixed lattice of kirigami cuts to be performed on a flat sheet. This single pluripotent lattice of cuts permits a wide variety of target surfaces to be programmed into the sheet by varying the folding directions.


Author(s):  
H. Watanabe ◽  
M. Zangeneh

The application of sweep in the design of transonic fans has been shown to be an effective method of controlling the strength and position of the shock wave at the tip of transonic fan rotors, and the control of corner separations in stators. In rotors sweep can extend the range significantly. However, using sweep in conventional design practice can also result in a change in specific work and therefore pressure ratio. As a result, laborious iterations are required in order to recover the correct specific work and pressure ratio. In this paper, the blade geometry of a transonic fan is designed with sweep using a 3D inverse design method in which the blade geometry is computed for a specified distribution of blade loading. By comparing the resulting flow field in the conventionally and inversely designed swept rotors, it is shown that it is possible to apply sweep without the need to iterate to maintain pressure ratio and specific work when using the inverse method.


Sign in / Sign up

Export Citation Format

Share Document