Modal Analysis Clustering for Physical Mode Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current modal analysis techniques face challenges in distinguishing physical modes from mathematical modes, particularly due to the presence of spurious modes and harmonic content, which complicates the identification of resonance frequencies and damping values in structures like rotating machinery, leading to inefficient operational modal analysis when assumptions about white noise input are violated.

Innovation Solution

A method using non-hierarchical clustering and a spuriousness metric, such as the silhouette coefficient, to automatically detect physical modes by optimizing the number of clusters and calculating distances between mode estimates, allowing for the identification and removal of harmonic components from signals, thereby improving the accuracy of modal parameter estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual mode selection is used on the stabilization diagram, then the user can identify physical modes, but the process is time-consuming and highly variable depending on user ability

Engineering Contradiction:
Improvemode selection accuracyVSAvoidtime for mode selection
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs automatic mode selection using algorithms that independently identify physical modes from mathematical modes without requiring user intervention. The stabilization diagram processing is automated through computational methods including clustering algorithms and model selection criteria, enabling the system to serve itself in the mode selection task.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The manual mechanical process of visually inspecting and selecting modes from the stabilization diagram is replaced by computational algorithms. These algorithms automatically process the modal estimates and stabilize the diagram to identify physical modes, substituting human visual inspection and decision-making with automated computational methods.

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

2Reliability

If the modal order is over-specified to capture all physical modes, then complete mode coverage is achieved, but the number of spurious mathematical modes increases making selection difficult

Engineering Contradiction:
Improvecompleteness of mode identificationVSAvoidcomplexity of mode selection process
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The algorithm segments the identified modes into distinct categories: physical modes and mathematical modes. By applying clustering techniques to the modal estimates from the over-specified model, the system automatically groups and separates valid physical modes from spurious mathematical modes, making the selection process systematic rather than complex and manual.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system uses feedback from the stabilization diagram and model selection criteria to automatically adjust and refine mode identification. The algorithm iteratively processes the modal estimates, using feedback from residual analysis and model comparison to distinguish physical from mathematical modes, reducing the complexity of selection while maintaining reliability.

Inventive Principle:
Principle #23Feedback

3Object-affected harmful factors

If cepstrum analysis is used to filter harmonic peaks, then dominant harmonics are removed, but the technique is not error-proof and may remove valid physical modes

Engineering Contradiction:
Improveinfluence of harmonic peaksVSAvoidaccuracy of mode detection
Core Design Contradiction:
Object-affected harmful factorsVSReliability

Solution Approach 1:

The system introduces an intermediary step between raw signal processing and final mode identification. Rather than directly applying cepstrum analysis that may erroneously remove valid modes, the patent uses automated modal analysis with model selection criteria as an intermediary that carefully distinguishes between harmonic content and genuine physical modes, reducing false removals while still filtering harmful harmonics.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3649568B1Method for automatic detection of physical modes in a modal analysis model
Publication Date: 2024.05.01 VRIJE UNIV BRUSSEL
  • EP3649568B1 patent drawingFigure 1
  • EP3649568B1 patent drawingFigure 2a
  • EP3649568B1 patent drawingFigure 2b

AI summary

The first aspect of the invention is related to a new method for automatically detecting physical modes within the data resulting from a modal analysis estimation algorithm (e.g. LSCE, PolyMax or other). The automatic detection method of the invention is based on a non-hierarchical clustering method wherein the number of clusters is automatically optimized, further making use of a metric for spuriousness within each cluster. According to the second aspect of the invention, the method for automatically detecting modes is used in a method removing harmonics from a signal.