Engine Vibration Spectrogram Matching for Faster Anomaly Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current engine vibration analysis methods are inefficient due to the difficulty in obtaining and comparing high-frequency vibration signatures, reliance on general algorithms not adapted to specific engines, and the complexity of on-board algorithm implementation and certification processes, leading to tedious and experience-dependent anomaly detection.
Innovation Solution
A method and system for analyzing engine vibrations by forming spectrograms, selecting zones of interest, comparing them to a reference database, and projecting onto a representation space to find similar behaviors, allowing for automatic and objective diagnosis and enrichment of the database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-frequency vibration measurements are acquired to obtain accurate vibration signature, then measurement precision is improved, but data transmission and processing complexity increases
Solution Approach 1:
The patent extracts only the essential vibration signature characteristics from high-frequency measurements rather than transmitting and processing all raw high-frequency data. This is achieved by analyzing vibration patterns at lower frequencies that still capture the essential diagnostic information, thereby reducing data transmission and processing complexity while maintaining measurement precision for anomaly detection.
Solution Approach 2:
The patent creates simplified representations (copies) of the vibration signature that capture the essential diagnostic information without requiring full high-frequency data. By working with these simplified copies rather than the complete high-frequency signal, the system reduces computational complexity and data transmission requirements while preserving the ability to detect anomalies accurately.
2Device complexity
If general algorithms calibrated for engine families are used instead of specific engine algorithms, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent adjusts algorithm parameters based on specific engine characteristics rather than requiring completely different algorithms for each engine. By modifying parameters such as frequency ranges, threshold values, and analysis windows to match the specific engine's operational characteristics, the system achieves precision comparable to engine-specific algorithms while using a unified algorithmic framework that reduces overall complexity.
3Productivity
If onboard algorithm implementation is performed to enable real-time analysis, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent extracts only the essential processing functions needed for real-time anomaly detection and implements them onboard, while leaving more complex analysis functions for ground-based processing. This selective extraction enables real-time monitoring capabilities onboard without requiring the full computational power that would significantly increase device complexity and certification requirements.
4Measurement precision
If expert visual analysis of spectrograms is performed to detect anomalies, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent implements automated anomaly detection algorithms that perform the analysis function that previously required expert visual inspection. The system automatically identifies anomalies in vibration spectrograms by comparing them against learned patterns and thresholds, thereby maintaining the high precision of expert analysis while eliminating the time loss associated with manual visual inspection of multiple spectrograms.
Data Source
Figure 1
Figure 2~3A
Figure 3B~3C
AI summary
The invention relates to a method and system for the vibration analysis of an engine, comprising: processing means (3) for acquiring vibration signals related to the engine (15) and for forming at least one spectrogram that is representative of an operational state of said engine; processing means (3) for selecting at least one area of interest in said at least one spectrogram; processing means (3) for comparing each area of interest with a set of corresponding comparison areas belonging to annotated spectrograms stored in a reference database, and for determining a subset of comparison areas having vibration behaviour similar to that of said area of interest; and output means (9) for displaying said subset of comparison areas and annotations linked with said subset of comparison areas.