Vibrational Mode Classification via Hierarchical Subdomain Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for classifying vibrational modes from Finite Element Method (FEM) simulations are inefficient and prone to errors, leading to suboptimal design outcomes in structural dynamics analysis, particularly in identifying and mitigating vibrations in mechanical objects under dynamic excitation.

Innovation Solution

The method involves classifying vibrational modes by analyzing data from geometric subdomains of varying levels of detail, using statistical data to represent displacement and strain energy across multiple spatial extents, and employing machine learning to automate the classification process, reducing the need for manual evaluation and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual evaluation of all local points is performed for classifying vibrational modes, then measurement precision is improved, but loss of time increases

Engineering Contradiction:
Improveclassification accuracyVSAvoidclassification time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The model is divided into multiple geometric subdomains at different levels of detail (coarse and fine levels). Instead of evaluating all local points uniformly, the method segments the analysis into hierarchical subdomains, where statistical data from coarse subdomains provides overview information and fine subdomains provide detailed local information. This segmentation enables efficient classification by processing a reduced set of representative subdomains rather than all individual local points.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The manual evaluation process is replaced by an automated machine learning classification system. The method uses trained classifiers that automatically process simulation results and statistical data from geometric subdomains to classify vibrational modes, eliminating the need for manual inspection while maintaining or improving classification accuracy and significantly reducing time consumption.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If data from all local points is processed for classification, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improveclassification accuracyVSAvoiddata processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The data processing system is segmented into hierarchical levels corresponding to different geometric subdomains. Coarse subdomains aggregate data from multiple fine subdomains, creating a multi-level data structure that reduces overall complexity. The classification process operates on this hierarchical structure, processing statistical summaries at each level rather than raw data from all individual points.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The method extracts only the essential statistical data (mean, standard deviation, maximum, minimum values) from simulation results at each geometric subdomain level. This extraction process filters out redundant information and retains only the most relevant features for classification, simplifying the data processing system while maintaining classification accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

3Manufacturing precision

If detailed analysis of all local points is performed, then manufacturing precision is improved, but productivity decreases

Engineering Contradiction:
Improvedesign qualityVSAvoiddesign efficiency
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The design analysis process is segmented into hierarchical geometric subdomains that can be processed at different levels of detail. This enables parallel processing of multiple subdomains and efficient utilization of computational resources, improving productivity while maintaining the necessary analysis depth for high-quality design decisions through the multi-level statistical evaluation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The method performs preliminary statistical aggregation of simulation data at coarse geometric subdomain levels before detailed classification. This preliminary action pre-processes and summarizes data in a way that accelerates subsequent classification operations, enabling faster design iterations without sacrificing the precision needed for high-quality manufacturing decisions.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4047502A1Method for classifying the type of a vibrational mode and real-world object designed based on a classification comprising such a method
Publication Date: 2022.08.24 VOLKSWAGEN AG
  • EP4047502A1 patent drawingFigure 1a~1b
  • EP4047502A1 patent drawingFigure 1c~2
  • EP4047502A1 patent drawing

AI summary

The present invention relates to a method for classifying the type of at least one first vibrational mode of the vibrational modes which are obtained from a simulation of a model of a real-world object under dynamic excitation. The present invention also relates to a real-world object which has been designed at least in part based on a classification comprising a respective method.