Electrical Machine Fault Detection Using Vibration Mode Shapes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting electrical faults in electrical machines, such as vibration monitoring, are inadequate in distinguishing between different fault conditions and fail to reliably identify electrical faults, as they do not consider mode shapes of vibrations.
Innovation Solution
A method that involves measuring vibrations in multiple radial directions of the stator, determining vibration frequencies and mode shapes, and using a combination of these as indicators to identify fault conditions, including broken rotor bars, dynamic and static eccentricity, and inter-turn or inter-coil short circuits, by applying a two-dimensional Fourier transform and analyzing amplitude changes under different loads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If vibration monitoring is used to detect faults in electrical machines, then mechanical faults can be detected, but electrical faults cannot be reliably distinguished or identified
Solution Approach 1:
The patent transitions from conventional single-direction vibration monitoring to multi-directional vibration measurement by placing sensors at multiple radial positions around the stator. This dimensional expansion enables the determination of mode shapes, which provide additional information dimensions for distinguishing between different fault conditions that produce identical vibration frequencies.
Solution Approach 2:
The patent changes the measurement parameters from simple vibration amplitude and frequency to include mode shape characteristics. By analyzing how vibration patterns distribute across multiple radial directions, the system extracts mode shape parameters that serve as unique identifiers for different fault types, thereby improving both reliability and precision in fault detection.
2Ease of operation
If conventional vibration monitoring methods are used, then simple fault indicators can be obtained, but the complexity of interpretation increases and satisfactory fault distinction cannot be achieved
Solution Approach 1:
The patent implements a systematic feedback mechanism where multi-directional vibration measurements are continuously processed to extract mode shape characteristics. These characteristics feed back into the fault diagnosis system, providing clear, interpretable indicators that reduce the complexity of fault identification while preserving comprehensive fault condition information.
Solution Approach 2:
The patent replaces the need for expert empirical interpretation with an automated analysis system that mathematically determines mode shapes from multi-directional measurements. This substitution transforms complex mechanical vibration analysis into a more systematic process that reduces information loss while maintaining ease of operation through automated pattern recognition.
Data Source
Figure 1~2d
Figure 3
AI summary
For identifying a fault in an electrical machine vibration is measured in a plurality of radial directions of the stator 30. On the basis of the vibration measurements a vibration frequency and a mode shape of the vibration at this frequency is determined. Characteristics of the vibration in terms of both the vibration frequency and the mode shape are used to identify a fault condition of the electrical machine.