Optical System Online Design via Intelligent Light Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing optical system design methods rely heavily on offline algorithms and numerical modeling, which are computationally intensive and limited by the complexity of optical systems and light wave propagation, making it challenging to achieve high accuracy and efficiency in online self-designing.
Innovation Solution
The method proposes using intelligent optical computing to establish a connection between the characteristics of a universal optical system and the parameters of a differentiable neural network, allowing for online designing by mapping the optical design process to neural network learning in a physical system, leveraging spatial symmetry and Lorentz reciprocity for efficient computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline algorithms and numerical modeling are used for optical system design, then design accuracy can be improved, but computational resources and time consumption increase exponentially
Solution Approach 1:
The patent replaces traditional electronic computing systems with an optical computing system that uses light propagation through optical components (lenses, mirrors, spatial light modulators) to perform design calculations. The optical system physically implements the design algorithm through light field transformations, eliminating the need for electronic computation and achieving parallel processing of design parameters.
Solution Approach 2:
The patent creates a physical optical copy of the design problem by mapping the optical system design onto a corresponding optical computation system. The optical fields, components, and propagation paths are configured to replicate the mathematical operations required for design optimization, allowing direct physical measurement of design parameters without iterative electronic computation.
2Manufacturing precision
If brute force and target-directed optimization are used, then optimization performance can be improved, but device complexity and computational resources increase
Solution Approach 1:
The optical computing system performs self-design by automatically optimizing its own configuration through light propagation physics. The system uses the natural wave nature of light and optical interference patterns to self-adjust and converge on optimal design parameters without requiring external electronic control or complex optimization algorithms, thereby reducing system complexity while maintaining high optimization performance.
3Measurement precision
If accurate numerical modeling of optical systems is attempted, then model precision can be improved, but computational burden increases due to light wave propagation complexity
Solution Approach 1:
The patent substitutes electronic numerical modeling with physical optical modeling. Instead of computationally simulating light wave propagation through complex algorithms, the system uses actual light waves propagating through optical components to naturally embody the physical laws of optics, achieving accurate modeling without computational burden.
Data Source
Figure 1
Figure 2A~2E
Figure 3A~4E
AI summary
A method and an apparatus for optical system online designing based on intelligent optical computing are disclosed. The method includes: constructing an optical system based on a differentiable online neural network and mapping light propagation input data to the differentiable online neural network to obtain a network mapping result; training the online neural network based on the network mapping result and a fully-forward-mode to compute first light propagation output data of the optical system; computing an error field according to the first light propagation output data and application target data, and inputting the error field to the optical system in the fully-forward-mode to output second light propagation output data; and computing gradient information of an online neural network parameter, and updating the online neural network parameter to obtain a trained optical system parameter.