DEEP LEARNING NEURAL NETWORK-BASED DESIGN METHOD AND SYSTEM FOR DIFFRACTIVE ELEMENT

WIPO Patent Application WO/2024/113512

A1

A deep learning neural network-based design method and system for a diffractive element. The design method comprises: generating a plurality of groups of diffractive element structures according to a key parameter of a diffractive element, performing electromagnetic vector calculation on the diffractive element structures to obtain corresponding diffraction efficiency data, and forming a data set by means of the plurality of groups of diffractive element structures and the corresponding diffraction efficiency data thereof (S10); selecting a deep learning architecture, constructing a deep learning neural network, and training and testing the deep learning neural network by using the data set to obtain a deep learning neural network model (S20); and designing a target optical field to acquire corresponding target diffraction efficiency data, and inputting the target diffraction efficiency data into the trained deep learning neural network model to acquire structural data for generating a target diffractive element (S30). Vector electromagnetic simulation is combined with a deep learning neural network, such that the design efficiency of diffractive elements is improved.

LI RUIBIN (CN)
YANG BOWEN (CN)
YANG MING (CN)
ZHU PING (CN)
LUO MINGHUI (CN)
QIAO WEN (CN)

PCT/CN2023/077752

June 06, 2024

February 22, 2023

Click for automatic bibliography generation  

SVG TECH GROUP CO LTD (CN)
UNIV SOOCHOW (CN)

G02B27/42; G06N3/04

CN114647081A	2022-06-21
CN113591298A	2021-11-02
CN111458777A	2020-07-28
CN113569469A	2021-10-29
CN114690404A	2022-07-01
US20210142170A1	2021-05-13

PSHIP FIRM, LLC (CN)

View/Download PDF      PDF Help

Previous Patent: MULTI-LANE DISEASE DISTRIBUTION MAP DRAWING METHOD AND SYSTEM, AND MEDIUM

Next Patent: AL-Y-ZR-MG-MN-SC ALUMINUM ALLOY SUITABLE FOR ADDITIVE MANUFACTURING PROCESS AND ADDITIVE MANUFACTURI...