Multi-variable Model Analysis System for Design Principle Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In complex structures like automobile suspensions, the large number of design variables makes it difficult to determine desired characteristic values such as operational stability and ride quality, leading to increased design time and development costs, as existing analysis systems struggle to extract the design principle inherent in the model.
Innovation Solution
A multi-variable model analysis system that creates models using orthogonal tables, calculates characteristic values, clusters similar models, calculates correlation coefficients, and extracts design variables with high correlation coefficients to grasp the design principle, reducing the number of variables needed for simulation and optimizing design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of design variables is increased to accurately model complex structures, then the simulation precision is improved, but the theoretical formula becomes complicated and the calculation time increases
Solution Approach 1:
The patent extracts and identifies the essential design variables that have the most significant impact on characteristic values through sensitivity analysis. By selecting only the critical variables (e.g., suspension stiffness, damping coefficients, geometric parameters) rather than using all possible variables, the system maintains simulation precision while simplifying the theoretical formula and reducing calculation complexity.
Solution Approach 2:
The patent segments the complex simulation problem into multiple independent sensitivity analysis steps. Each design variable is evaluated separately for its impact on characteristic values, allowing the system to identify and isolate the most important variables. This segmentation approach enables systematic reduction of variables while maintaining accuracy.
2Measurement precision
If the number of design variables is increased to capture all interactions, then the simulation accuracy is improved, but the design time and development cost increase
Solution Approach 1:
The patent performs sensitivity analysis and identifies critical design variables before the actual simulation and design optimization process. By preliminarily determining which variables have the most significant impact on characteristic values, the system can focus computational resources only on those variables during simulation, significantly reducing design time while maintaining accuracy.
Solution Approach 2:
The system extracts and prioritizes the most influential design variables through sensitivity analysis. By removing less significant variables from the simulation model, the system reduces the computational burden and design time while preserving the accuracy needed for effective design optimization.
3Measurement precision
If all design variables are used in the simulation, then the characteristic values can be determined accurately, but the relation between variables and characteristic values becomes difficult to grasp
Solution Approach 1:
The patent extracts and highlights the most significant design variables and their relationships with characteristic values through sensitivity analysis. By focusing on the critical variables (such as suspension stiffness, damping, and geometric parameters) that have the most impact on characteristic values, the system makes the design principles more interpretable and easier to understand while maintaining determination accuracy.
Solution Approach 2:
The patent applies different levels of analysis to different variables based on their importance. Critical variables are analyzed in detail with explicit relationships to characteristic values, while less significant variables are treated more broadly. This local quality approach maintains accuracy for the most important relationships while simplifying the overall model interpretability.
Data Source
AI summary
A multi-variable model analysis system comprises a model creation unit for creating a plurality of models individually having a plurality of variables, a characteristic value calculation unit for calculating the characteristic values of the models on the basis of the variables of the models given and for writing the variables and characteristic values of the models, a clustering unit for classifying the plural models having the characteristic values of a high similarity, into an identical cluster; a correlation coefficient calculation unit for calculating the correlation coefficients of the variables of the models in individual clusters and for writing the correlation coefficients in a memory map; and an extraction unit for extracting the variable having a correlation coefficient exceeding a predetermined value in the individual clusters, from the memory map.