Rotating Machine Diagnosis Using Reductive-Redundant Vibration Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for diagnosing the technical condition of rotating machines require the construction of kinetostatic models, a priori configured frequency bands, and large historical datasets, which complicates the configuration process and limits the detection of faults to predefined spectral ranges.
Innovation Solution
A method for automatic technical diagnosis of rotating machines using reductive-redundant data processing of vibration signals, which eliminates the need for kinetostatic models and a priori configured frequency bands, allowing for unsupervised assessment of technical condition and detection of faults across the entire frequency band.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If kinetostatic models and a priori configured frequency bands are used for fault detection, then the detection accuracy within predefined spectral ranges is improved, but the device complexity and configuration time increase significantly
Solution Approach 1:
The patent extracts and eliminates the complex kinetostatic model construction and a priori frequency band configuration steps from the fault detection process. By using reductive-redundant data processing, the system directly analyzes vibration signals without requiring these preliminary complex configurations, thereby reducing device complexity while maintaining detection capability.
Solution Approach 2:
The system performs self-configuration through reductive-redundant processing, where the data processing unit automatically identifies fault conditions and spectral characteristics without external intervention or pre-configured models. The system serves itself by adapting to the specific vibration signals of each machine, eliminating the need for manual kinetostatic model construction.
2Speed
If a priori configured frequency bands are used for vibration analysis, then the processing speed for predefined spectral ranges is improved, but the adaptability to detect faults beyond predefined ranges is reduced
Solution Approach 1:
The patent implements dynamic frequency band adjustment through reductive-redundant processing. The system automatically adapts the spectral analysis ranges based on the actual vibration signals detected, allowing the frequency bands to expand or contract dynamically. This enables the system to maintain processing efficiency while detecting faults across the entire frequency spectrum, not limited to predefined bands.
Solution Approach 2:
The reductive-redundant processing mechanism provides universal fault detection capability across different machine types and operating conditions. The same processing algorithm can analyze various vibration signatures without requiring machine-specific pre-configuration, making the system versatile for detecting faults in diverse rotating machines while maintaining consistent processing performance.
3Measurement precision
If large historical datasets are used for training diagnostic models, then the accuracy of supervised learning is improved, but the loss of time in data collection and processing increases
Solution Approach 1:
The system eliminates the need for external historical datasets by performing self-configuration through reductive-redundant processing of real-time vibration signals. The data processing unit automatically learns fault patterns from the actual operational data of the machine being monitored, eliminating time-consuming data collection and processing of historical datasets while achieving high diagnostic accuracy.
Solution Approach 2:
The patent performs preliminary data processing and feature extraction directly during machine operation through reductive-redundant processing. By preparing and analyzing data in real-time rather than requiring extensive historical datasets to be collected and processed beforehand, the system reduces time loss while maintaining the ability to accurately diagnose faults.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly simplifies the configuration process, enables the detection of faults beyond predefined spectral ranges, and allows for effective unsupervised technical diagnosis of rotating machines, thereby improving the reliability and efficiency of condition monitoring systems.
Implementation Method 1
measurement of mechanical vibrations, in particular the measurement of vibrations transverse to the direction of their propagation in solids
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The subject matter of the invention relates to a method for the automatic technical diagnosis of rotating machines, including detection of faults in rotating machines, in particular vibration measurements of rotating machines by the automatic detection of changes in the signals of machine vibration. The method consists in that monitoring and diagnosis are performed in terms of identifying both the assemblies or parts of an object of the rotating machine subjected to the diagnosis, in such a way that vibration data are recorded in the form of a vibration signal from at least one vibration sensor (2) mounted on the housing of the mechanical elements of the rotating machine (1), wherein each subsequent vibration signal is recorded at the same length and at the same sampling frequency. Then, collected vibration signals are processed in a vibration data acquisition unit (3) equipped with a reductive-redundant mechanism (4) by means of which the vibration signal is modified by calculating the values of spectral components in the domain of frequency or orders and the values are segmented relative to subsequent tested spectral resolutions on the basis of current scale of class intervals. Then, an averaged spectrum is calculated and added to the state matrix and a statistical measure of the comparison of spectral amplitudes is calculated for individual class intervals at the standard score measure and the procedure is repeated for each tested spectral resolution in the range from the value equal to the inverse of the vibration signal length to half the sampling frequency. Finally, the results of the vibration growth index are saved, then are sorted from the highest to the lowest value thus allowing the potential occurrence of a fault to be determined in terms of identifying both a specific assembly or part of the rotating machine (1).