Anomaly Diagnosis via Frequency Spectrum Width
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional online diagnosis methods for rotating machines fail to detect anomalies when they do not manifest as a peak in a specific frequency component, leading to reduced availability of production facilities due to unplanned shutdowns.
Innovation Solution
The system analyzes the width of the frequency spectrum of AC current values inputted or outputted from the device to be diagnosed, judging an anomaly if the spectrum width is wider than normal, utilizing current intensity increases and proportion of time with increased current intensity in the vicinity of the drive frequency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline diagnosis is performed by stopping the rotating machine, then anomaly detection accuracy is improved, but production facility availability deteriorates
Solution Approach 1:
The patent transitions from static offline diagnosis to dynamic online diagnosis by continuously monitoring frequency spectrum width while the rotating machine operates. This allows anomaly detection without stopping the machine, maintaining production availability while detecting degradation through real-time spectral analysis
Solution Approach 2:
The system performs preliminary detection of anomalies by monitoring frequency spectrum width changes before they develop into critical failures. This early warning capability allows for scheduled maintenance planning, preventing sudden failures while keeping the machine operational during the monitoring phase
2Device complexity
If peak-based frequency analysis is used, then simple anomaly detection is achieved, but detection capability deteriorates when anomalies do not manifest as peaks
Solution Approach 1:
The patent changes the detection parameter from peak amplitude to frequency spectrum width. This parameter transformation allows detection of anomalies that do not produce distinct peaks, such as those masked by other frequency components or lacking clear spectral signatures, thereby improving detection reliability while maintaining method simplicity
Data Source
AI summary
An anomaly diagnosis system is provided that includes: a current-value obtaining part 111 that obtains a value of an AC current inputted to or outputted from a device to be diagnosed; a frequency analyzing part 112 that converts the current values obtained at intervals into a frequency spectrum which correlates a current intensity with a frequency; an anomaly-judging-value calculating part 113 that calculates at intervals a frequency width of the frequency spectrum at a current intensity “As” satisfying As=m×At (0<m<1), where “At” is a peak current intensity associated with an analysis frequency which is an integer multiple of a frequency of the AC current; and a judging part 114 that judges whether the frequency width is wider than that in a normal condition and, if it holds true, judges that the device is in an anomalous condition.