Deep CAD Neural Network for Topological Signature Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current CAD systems face inefficiencies in CAD object retrieval and similarity evaluation due to the complexity of 3D models, which require effective and efficient methods to capture both geometric and topological features without losing information through approximation.
Innovation Solution
A computer-implemented method using a Deep CAD neural network that processes B-rep graphs to generate topological signatures, allowing for efficient retrieval and similarity evaluation of CAD objects by encoding both geometric and topological information into compact vector representations, enabling CPU and memory-efficient manipulation and similarity searches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional CAD systems use complex 3D models for retrieval and similarity evaluation, then measurement precision and reliability are improved, but device complexity and loss of time increase
Solution Approach 1:
The patent extracts topological features from complex 3D CAD models and represents them as simplified graphs. The B-rep graph captures essential topological relationships (faces, edges, vertices, and their connections) while discarding unnecessary geometric details, enabling efficient retrieval and similarity evaluation without sacrificing measurement precision
Solution Approach 2:
The patent creates a topological graph copy of the CAD model that preserves topological structure while simplifying the representation. This graph copy serves as a lightweight surrogate for the original complex model, allowing fast similarity comparisons and retrieval operations
2Measurement precision
If traditional CAD systems use complex 3D models for retrieval, then measurement precision is improved, but productivity and loss of time worsen
Solution Approach 1:
The patent segments the CAD model into discrete topological elements (faces, edges, vertices) and represents their relationships as a graph structure. This segmentation enables efficient processing and comparison by breaking down complex models into manageable, comparable units
Solution Approach 2:
The patent creates a simplified topological graph copy that preserves essential structural information while enabling fast computation. This copy allows for efficient retrieval operations and similarity evaluation without requiring processing of the full complex 3D model
3Measurement precision
If traditional CAD systems use complex 3D models, then measurement precision is improved, but loss of information and device complexity increase
Solution Approach 1:
The patent extracts and preserves critical topological information (connectivity relationships between faces, edges, and vertices) while discarding redundant geometric data. The B-rep graph maintains topological signatures that are sufficient for similarity evaluation and retrieval purposes
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
The disclosure notably relates to a computer-implemented method of machine-learning. The method comprises obtaining a training dataset of B-rep graphs. Each B-rep graph represents a respective B-rep. Each B-rep graph comprises graph nodes each representing an edge, a face or a co-edge of the respective B-rep and being associated with one or more geometrical and/or topological features. Each B-rep graph comprises graph edges each between a respective first graph node representing a respective co-edge and a respective second graph node representing a face, an edge, an adjacent co-edge, or a mating co-edge associated with the respective co-edge. The method further comprises learning, based on the training dataset, a Deep CAD neural network. The Deep CAD neural network is configured to take as input a B-rep graph and to output a topological signature of the B-rep represented by the input B-rep graph.