Local Deep CAD Neural Network for B-Rep Topological Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CAD systems lack efficient solutions for computing local topological signatures of B-rep graphs, which are essential for improved CAD object retrieval and topological similarity evaluation.
Innovation Solution
A computer-implemented method of machine-learning that learns a local Deep CAD neural network to compute local topological signatures for each graph node of a B-rep graph, enabling efficient local CAD object retrieval and similarity evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If global topological signatures are computed to represent the global topology of the B-rep, then CAD object retrieval and topological similarity evaluation are improved, but memory and CPU requirements increase
Solution Approach 1:
The patent divides the global topological signature computation into local topological signatures for each graph node in the B-rep graph. Instead of computing one signature for the entire B-rep, the system computes multiple smaller signatures locally at each node, reducing the memory footprint while maintaining topological analysis capability.
Solution Approach 2:
The patent extracts local topological information from each graph node independently, taking out the essential topological features needed for similarity evaluation without retaining the complete global structure in memory. This extraction approach maintains measurement precision while reducing resource consumption.
2Measurement precision
If global topological signatures are computed to represent the global topology of the B-rep, then topological similarity evaluation is improved, but processing time increases
Solution Approach 1:
The patent segments the topological analysis into independent local computations at each graph node. This segmentation allows for parallel processing of multiple nodes simultaneously, reducing overall processing time while maintaining the precision of topological similarity evaluation through aggregation of local results.
Solution Approach 2:
The patent computes only the necessary local topological signatures needed for similarity evaluation rather than performing complete global topological analysis. This partial action approach achieves sufficient measurement precision with reduced computational overhead and faster processing.
3Quantity of substance
If local topological signatures are computed for each graph node, then memory efficiency is improved, but system complexity increases
Solution Approach 1:
The patent uses a Deep CAD neural network that learns to generate local topological signatures through training on B-rep graphs. The network architecture copies successful patterns from training data to efficiently compute local signatures, managing system complexity through learned representations rather than explicit complex algorithms.
Data Source
AI summary
A machine-learning method including obtaining a training dataset of B-rep graphs. Each B-rep graph represents a respective B-rep. Each B-rep graph includes 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 further 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 includes learning, based on the training dataset, a local Deep CAD neural network. The local Deep CAD neural network takes as input a B-rep graph and to output, for each graph node of the input B-rep graph, a local topological signature of the B-rep element represented by the graph node.