Bearing Vibration Monitoring Using Statistical Failure Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for monitoring bearing failures in machines are inadequate as they fail to detect certain defects in bearings through vibration measurements.
Innovation Solution
A method involving the determination of multiple statistical parameters from vibration measurements, modeling these parameters with normal distributions, and setting a detection threshold to predict bearing failures by comparing scores derived from these parameters to the threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spectral analysis is performed on vibration signals to identify bearing defects, then certain harmonics representative of defects can be identified, but some defects of the bearing are not identified from the spectrum
Solution Approach 1:
The patent transforms the vibration signal from time domain to frequency domain through spectral analysis, changing the parameter representation from temporal amplitude to frequency spectrum. This allows identification of harmonics that are not visible in time-domain signals, thereby detecting certain defect types while transforming the measurement parameters to reveal different defect characteristics
Solution Approach 2:
The patent introduces a new dimension of analysis by examining the spectral content of vibration signals. Instead of only analyzing time-domain waveforms, the signal is transformed into the frequency domain, adding a spectral dimension that reveals harmonic patterns associated with bearing defects, thus compensating for the limitation of time-domain analysis alone
2Reliability
If multiple statistical parameters are used to predict bearing failure, then different kinds of defects may be detected, but the complexity of the monitoring system increases
Solution Approach 1:
The patent segments the vibration signal analysis into multiple statistical parameter calculations, where each parameter (RMS, skewness, kurtosis, etc.) targets specific defect characteristics. By dividing the analysis into distinct parameter components, the system can detect different defect types through specialized parameters while maintaining a structured and manageable monitoring framework
Solution Approach 2:
The patent creates a multi-functional monitoring system where a single vibration sensor and processing unit calculate multiple statistical parameters simultaneously. This universal approach allows one system to detect various defect types (inner ring, outer ring, rolling element defects) through different parameter combinations, avoiding the need for separate specialized sensors for each defect type
3Measurement precision
If a detection threshold is determined from vibration measurements during a training period, then accurate monitoring specific to the machine application can be achieved, but additional time and measurements are required during installation
Solution Approach 1:
The patent performs preliminary action by conducting a training period measurement during installation to establish machine-specific detection thresholds before normal operation begins. This preliminary data collection and threshold calibration is done upfront, allowing the monitoring system to operate with high accuracy from the start without requiring continuous threshold adjustments during operation
Solution Approach 2:
The system performs self-service by automatically collecting vibration data during the training period and generating detection thresholds without requiring external intervention or expert analysis. The monitoring system configures itself by processing the training data and establishing baseline parameters, reducing the need for manual setup and enabling rapid deployment
Data Source
AI summary
A bearing device includes a bearing provided with an inner ring and an outer ring that rotate concentrically relative to one another. A vibration sensor measures vibrations of the bearing. The bearing device includes a first determining means (10) for determining values of at least a first statistical parameter and a second statistical parameter (P1, P2, P3, P4, P5, P6), a modelling means (11) for modelling the values of the first and statistical parameters with normal distributions, a second determining means (12) for determining a detection threshold (Sd) from the normal distributions during the training period, a third determining means (13) for determining a score (Sc) during normal operation of the machine (1) from the values of the two statistical parameters and the normal distributions, and a comparing means (14) for determining the failure of the bearing (4) from the detection threshold (Sd) and the score (Sc).