Rotating Machine Vibration Correlation for Subsystem Fault Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for monitoring rotating machines with multiple subsystems, such as axial piston pumps and rotary pressure exchangers, face challenges in accurately evaluating complex vibration signals due to overlapping frequencies from different sources, leading to difficulty in identifying potential faults and requiring complex analysis that is sensitive to sensor positioning and environmental factors.
Innovation Solution
A method involving sampling vibration measurement signals into subsignals corresponding to one full revolution of the machine, averaging to create a reference subsignal, and performing cross-correlation analysis of signal snippets assigned to individual subsystems to identify potential faults, which is independent of sensor positioning and environmental conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vibration measurement signals are analyzed using conventional methods, then vibration data can be collected, but the analysis becomes complex and inaccurate due to overlapping frequencies from different sources
Solution Approach 1:
The patent applies segmentation by dividing the vibration measurement signal into multiple revolution-based segments (first through fifth revolution segments). Each segment corresponds to one complete revolution of the rotating machine, allowing frequencies from different sources to be separated and analyzed individually rather than as overlapping composite signals. This segmentation enables precise identification of subsystem-specific vibration patterns.
Solution Approach 2:
The patent introduces an intermediary reference revolution segment that is generated by averaging multiple revolution segments. This reference segment serves as a mediator for comparison against individual revolution segments, enabling the detection of deviations that indicate potential faults. The intermediary reference provides a stable baseline that simplifies the analysis of complex overlapping vibration signals.
2Reliability
If conventional vibration monitoring methods are used, then general vibration levels can be detected, but reliable fault identification becomes difficult due to sensitivity to sensor positioning and environmental factors
Solution Approach 1:
The patent applies periodic action by analyzing vibration signals in discrete revolution-based segments rather than continuous monitoring. Each revolution segment captures a complete periodic cycle of the rotating machine, allowing consistent comparison across multiple revolutions. This periodic approach eliminates sensitivity to sensor positioning because each segment captures the same mechanical events in the same rotational context, regardless of where in the cycle measurement begins.
Solution Approach 2:
The patent applies preliminary action by pre-processing the vibration signal into standardized revolution segments before analysis. By organizing the raw vibration data into consistent one-revolution segments and creating a reference segment in advance, the system establishes a stable framework for fault detection that is independent of environmental factors and sensor positioning variations.
3Productivity
If continuous vibration monitoring is performed, then machine operation can be tracked, but calibration requirements increase due to environmental conditions and sensor placement variations
Solution Approach 1:
The patent applies parameter changes by transforming the vibration signal from continuous time-domain data into discrete revolution-based segments with standardized parameters. Each segment is normalized to represent one complete revolution, changing the parameter framework from continuous amplitude-time to discrete revolution-count. This parameter transformation eliminates the need for calibration against environmental conditions because the revolution-based normalization inherently compensates for variations in sensor placement and operating conditions.
Data Source
Figure 1a~1b
Figure 2~5
Figure 6
AI summary
The invention relates to a method of monitoring a rotating machine configured for energy transfer having a plurality of subsystems, each subsystem comprising one cylinder. In the method a measurement signal of a vibration of the rotating machine is obtained and the vibration measurement signal (2) is sampled into a plurality of vibration subsignals (21, 22, 23, 24, 25), each vibration subsignal (21, 22, 23, 24, 25) corresponding to one full revolution of the rotating machine. A reference vibration subsignal (20) is determined based on an average of the plurality of vibration subsignals (21, 22, 23, 24, 25). The reference vibration subsignal (20) is sampled into a plurality of signal snippets (201, 202, 203, 204, 205), each signal snippet (201, 202, 203, 204, 205) assigned to one subsystem of the rotating machine. A cross-correlation analysis of the plurality of signal snippets (201, 202, 203, 204, 205) is performed for identifying a potential fault state of the rotating machine. The invention further relates to a rotating machine configured for energy transfer having a plurality of subsystems, each subsystem comprising one cylinder, is provided, wherein the rotating machine further comprises a control unit adapted to perform a method according the invention.