Vibrational Mode Classification via Hierarchical Subdomain Segmentation
Find Innovative SolutionsGenerate 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
Engineering 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
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.
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.
2Measurement precision
If data from all local points is processed for classification, then measurement precision is improved, but device complexity increases
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.
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.
3Manufacturing precision
If detailed analysis of all local points is performed, then manufacturing precision is improved, but productivity decreases
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.
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.
Data Source
Figure 1a~1b
Figure 1c~2
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.