Representative Design Selection Using Local Mutual Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current engineering design optimization methods fail to effectively identify and represent design concepts across multiple description spaces, leading to inefficient design concept identification and optimization, as they do not consider the similarity or correlation between different description spaces.
Innovation Solution
A computer-implemented method using local mutual information (LMI) to quantify the similarity between description spaces, identifying design data samples as representative designs that are highly informative across multiple description spaces, ensuring high similarity and correlation, thereby selecting archetype designs that preserve performance and specification changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional optimization methods are used to find independent solutions in two description spaces, then high quality solutions can be found for each space, but the relation between design data samples in the two different description spaces is not provided
Solution Approach 1:
The patent merges multiple description spaces into a unified framework by introducing a similarity measure that operates across all description spaces simultaneously. This allows the system to capture relationships between design data samples in different description spaces while maintaining the quality of solutions in each individual space.
Solution Approach 2:
The patent creates a universal approach that can handle multiple description spaces (specification space, performance space, operational space) within a single optimization framework. The similarity measure and representative design selection process work universally across all description spaces, providing both high-quality solutions and inter-space relationships.
2Reliability
If large datasets of design data are analyzed to identify meaningful design concepts across multiple description spaces, then comprehensive design insights can be obtained, but large computational resources are required
Solution Approach 1:
The patent extracts the essential information from large design datasets by identifying and selecting representative designs that capture the key characteristics of design concepts. Instead of processing all data points uniformly, the system extracts and focuses on the most informative samples, significantly reducing computational requirements while maintaining identification accuracy.
Solution Approach 2:
The patent applies partial action by selecting only the necessary subset of design data samples (representative designs) rather than processing the entire large dataset. This selective approach processes only the essential information needed to identify design concepts, reducing computational energy consumption while maintaining reliable results.
3Adaptability or versatility
If design concepts are identified by clustering design data samples, then groups of designs with similar characteristics can be formed, but ensuring cluster assignment persists across all considered spaces is computationally complex
Solution Approach 1:
The patent segments the complex multi-space clustering problem into a more manageable form by using a similarity measure that inherently handles multiple description spaces. The representative design selection process then simplifies the clustering output by identifying key representative samples from each concept, reducing the computational complexity while maintaining the ability to form accurate cross-space clusters.
Data Source
AI summary
The present disclosure relates to a computer-implemented method for performing a design process by processing design data of a physical object. The method includes acquiring a dataset including design data samples of design data, determining at least one design concept including design data samples from the acquired dataset based on at least a similarity of feature values of design features, calculating a local similarity measure between at least two description spaces for each design data sample included in the determined design concept, selecting a number of design data samples based on the calculated local similarity measure as most representative designs and least representative designs, outputting the selected most representative design to an engineering process for further processing or deleting the least representative design from the design data samples of the respective design concept, and processing at least one of the most representative design data samples in the engineering process.