Generative Adversarial Network for Shear Wall Structural Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing structural design methods for high-rise shear wall buildings are time-consuming, inefficient, and prone to variability due to reliance on expert experience, and existing computer-aided methods consume significant resources and struggle to meet rapid design requirements.
Innovation Solution
A method utilizing a generative adversarial network (GAN) to generate structural designs by extracting key elements from architectural drawings, coding them, and inputting these features into a pre-trained structural-design-oriented GAN model to produce structural drawings quickly and efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing structural design methods relying on expert experience are used, then design quality can be maintained, but design time and efficiency are significantly reduced
Solution Approach 1:
The patent uses a generative adversarial network to learn and copy expert design patterns from historical structural drawings. The GAN model is trained on datasets containing architectural drawings and corresponding structural drawings, enabling it to generate new structural designs that replicate the quality and characteristics of expert-designed structures without requiring actual expert intervention for each design task.
Solution Approach 2:
The patent replaces the mechanical system of expert human judgment and manual design with an automated computational system based on generative adversarial networks. This substitution transforms the design process from a manual, experience-dependent activity into an automated, data-driven process that can rapidly generate multiple design schemes while maintaining quality standards.
2Productivity
If computer-aided design methods are used to improve efficiency, then design speed increases, but computational resources and time consumption increase significantly
Solution Approach 1:
The patent performs preliminary action by pre-training the generative adversarial network model on comprehensive datasets of architectural and structural drawings before actual design tasks. This pre-training phase allows the model to learn design patterns and relationships in advance, so that during actual design execution, the model can rapidly generate structural schemes without requiring intensive real-time computation, thus reducing operational time consumption.
3Reliability
If manual structural design by experts is performed, then design accuracy and safety can be ensured, but design variability and inconsistency occur
Solution Approach 1:
The patent applies homogeneity by training the generative adversarial network on diverse but consistently formatted datasets of expert-designed structural drawings. The model learns the underlying patterns and rules that ensure design safety and consistency across different projects. Once trained, the model generates structural designs with uniform quality standards and safety requirements, eliminating the variability and inconsistency that occur when different experts perform manual design.
Data Source
AI summary
A method for the scheme design of shear wall structure based on a generative adversarial network includes: obtaining an architectural drawing; extracting key elements from the architectural drawing, coding the key elements by colors, and generating image features; inputting the image features into a pre-trained structural-design-oriented generative adversarial network model for processing to generate a structural drawing.