Vibration Spectrum Pattern Matching for Asset Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional vibration monitoring systems for assets like wind turbines rely heavily on rule-based methods that require precise kinematic information and fail to detect novel faults or correctly identify faults with signatures that vary from standard frequencies, leading to potential misses in fault detection.
Innovation Solution
A system and method that use pattern matching techniques to analyze the shape of vibration spectrum data, grouping assets with similar spectral patterns and employing clustering algorithms to identify both known and novel faults, independent of specific fault frequencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If rule-based methods with precise kinematic information are used for fault detection, then detection reliability for known faults is improved, but the system fails to detect novel faults or faults with varying signatures
Solution Approach 1:
The patent replaces the mechanical rule-based detection system with a data-driven machine learning system. Instead of relying on predefined kinematic rules and frequency thresholds, the system uses neural networks and pattern recognition algorithms to automatically learn fault signatures from vibration data, enabling detection of both known and novel faults without requiring precise kinematic information
Solution Approach 2:
The patent transforms the detection approach by changing from fixed frequency-domain parameter analysis to time-domain waveform pattern analysis. The system examines overall vibration waveform characteristics and temporal patterns rather than relying on specific frequency values, allowing adaptation to faults with varying signatures while maintaining detection reliability
2Measurement precision
If spectrum spike analysis at specific frequencies is performed, then detection precision for standard faults is improved, but faults with shifted or varying frequencies are missed
Solution Approach 1:
The patent transitions from analyzing vibration data in the frequency domain to analyzing it in the time domain. By examining waveform patterns, temporal characteristics, and signal morphology rather than frequency spectra, the system achieves robust fault detection that is insensitive to frequency shifts or variations in rotational speed
Solution Approach 2:
The patent creates a digital model or reference pattern of healthy vibration waveforms and compares actual vibration patterns against this model. The machine learning system learns normal vibration patterns and automatically identifies deviations, providing precise fault detection without requiring knowledge of specific fault frequencies or theoretical signatures
3Measurement precision
If manual review of vibration spectrum plots is required, then diagnostic accuracy is improved, but productivity and efficiency are reduced
Solution Approach 1:
The patent implements a self-diagnosing system where machine learning algorithms automatically analyze vibration data, identify faults, and generate diagnostics without human intervention. The system performs both detection and diagnosis autonomously, eliminating the need for manual spectrum analysis while maintaining or improving diagnostic accuracy through consistent automated pattern recognition
Solution Approach 2:
The patent incorporates feedback mechanisms where the system continuously learns from new data and refines its diagnostic capabilities. Automated alerts are generated based on real-time analysis, and the system adapts to new fault patterns through ongoing training, improving both accuracy and productivity through intelligent feedback loops
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A system and method including receiving vibration spectrum data from a plurality of different assets; determining, based on a shape of the vibration spectrum data for each of the plurality of assets, clusters for the plurality of assets, assets being grouped in a same cluster having vibration spectrum data of a similar spectral shape; determining for each of the clusters, based on an application of domain derived pattern recognition rules for the vibration spectrum data, one of a plurality of fault classifications; generating an output including an association of each of the plurality of assets with the fault classification of the cluster in which the particular asset is grouped; and saving a record of the output.