Multi-material multi-scale structure design method and system based on deep learning

By combining deep learning with the level set method and neural network, a multi-material and multi-scale structural design model is constructed, which solves the problems of manufacturing difficulty and low computational efficiency in multi-material and multi-scale structural design, and realizes efficient and feasible multifunctional structural design and manufacturing.

CN120633129APending Publication Date: 2025-09-12NAT INNOVATION INST OF DEFENSE TECH PLA ACAD OF MILITARY SCI
0 Cites 0 Cited by

Patent Information

Application Number
CN202510477964.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-16
Publication Date
2025-09-12

AI Technical Summary

Technical Problem

Existing multi-material and multi-scale structure design methods have problems such as high manufacturing difficulty, low design variables and computational efficiency. Especially when multiple materials are locally mixed and the complexity increases, it is difficult to achieve efficient and feasible optimization and manufacturing.

Method used

A deep learning-based method is used, combining the level set method with a neural network, to construct a parametric model of multi-material heterogeneous microstructures, generate a data set and train a proxy model. The multi-material and multi-scale structures are optimized through neural network reparameterization, simplifying the sensitivity analysis and manufacturing process.

Benefits of technology

It achieves efficient, feasible and manufacturable multi-material and multi-scale structural design, improves computational efficiency and simplifies sensitivity analysis, and promotes the design and manufacture of multifunctional structures.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a multi-material multi-scale structure design method and system based on deep learning. The method comprises the following steps: establishing a multi-material heterogeneous microstructure parameterized model; a data set is generated based on the constructed model, equidistant and uniform valuing is carried out on each cutting height at a set interval, each height variable can obtain a corresponding value, and then random sampling is carried out on all heights to obtain a data set of a corresponding sample scale; constructing an agent model, wherein the agent model is used for constructing a mapping relation between the height parameters of the microstructure and the corresponding performance parameters; based on neural network re-parameterization as an optimizer, performing optimization design by constructing constraints in a loss function and combining an agent model; and reconstructing the multi-material multi-scale configuration based on the optimized parameters.
Need to check novelty before this filing date? Find Prior Art