Rotating Power Transmission Fault Localization Without Band-Pass Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vibration monitoring techniques for rotating mechanical power transmission devices, such as aircraft gears, face challenges in accurately localizing faults due to the reliance on band-pass filtering, which ignores significant information when multiple harmonics are present in the vibration signal, leading to imprecision and reduced reliability in fault detection.
Innovation Solution
A method that involves obtaining and optimizing signals modeled as a product of high and low frequency components using resampling and discrete Fourier transforms, allowing for precise estimation and separation of contributions from individual elements without band-pass filtering, enabling accurate fault detection and localization using a single sensor.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If band-pass filtering is used to extract low frequency component, then fault detection can be performed, but information loss occurs when multiple harmonics are present
Solution Approach 1:
The vibration signal is segmented into multiple frequency components corresponding to different harmonics. Instead of filtering out high frequency components, the method extracts and analyzes each harmonic component separately to identify faults affecting different elements of the rotating device.
Solution Approach 2:
The method transitions from frequency domain filtering to time-frequency domain analysis by calculating instantaneous frequency and using short-time Fourier transforms. This dimensional change allows simultaneous preservation and analysis of multiple harmonic components that would be lost in traditional band-pass filtering.
2Ease of operation
If band-pass filtering around gear meshing frequency is applied, then contribution separation is achieved, but measurement precision decreases
Solution Approach 1:
The method employs dynamic time-windowing with varying window sizes adapted to the instantaneous frequency of each harmonic component. This dynamic approach maintains precise separation of contributions from different gear elements while preserving the full spectral information, unlike static band-pass filtering.
Solution Approach 2:
The analysis parameters including window size, frequency resolution, and transform type are dynamically adjusted based on the detected instantaneous frequency and harmonic content. This allows optimal separation precision for each harmonic component while maintaining overall measurement accuracy.
3Reliability
If visual and endoscopic inspection is performed, then fault verification is possible, but system downtime increases
Solution Approach 1:
The rotating mechanical device performs self-diagnosis through continuous vibration monitoring during normal operation. The system automatically detects, localizes, and diagnoses faults without requiring external inspection, enabling condition-based maintenance that eliminates scheduled shutdowns and reduces system downtime.
Solution Approach 2:
The vibration analysis system operates continuously during device operation, providing uninterrupted fault detection and monitoring. This continuous surveillance replaces periodic visual inspections, maintaining reliability verification without interrupting the useful action of the rotating device.
4Measurement precision
If multiple sensors are used for signal acquisition, then measurement accuracy improves, but device complexity increases
Solution Approach 1:
The method replaces physical signal separation mechanisms (multiple sensors positioned at different locations) with mathematical signal processing techniques. By using instantaneous frequency analysis and time-frequency transforms on a single sensor signal, the system achieves the same fault localization precision without the complexity of multiple sensors.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
The method of the invention comprises: - a step (E10) of obtaining a signal s that can be modeled by a product of a high frequency signal s1 and a low frequency signal s2, generated by the rotating mechanical power transmission device and acquired by a sensor; - a step (E20) of determining estimates of the signals s1 and s2, minimising a difference between all or part of the signal s and a product of these estimates; - a step (E30) of analysing the estimates of the signals s1 and s2 in order to detect the presence of a defect affecting the rotating power transmission mechanical device; and, if a defect is detected following the analysis step (E40), a step (E50) of locating said defect from at least one of the estimates of the signals s1 and s2.