Rolling Mill Vibration Detection Using PCA Outlier Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting abnormal vibration in rolling mills struggle with erroneous detection due to noise from equipment and difficulty in distinguishing vibrations from various sources, especially in continuous cold rolling mills where multiple frequencies are superimposed, leading to challenges in accurately identifying chatter marks.
Innovation Solution
A method involving vibration data collection, frequency analysis, principal component analysis using reference data from normal states, and outlier component extraction to detect abnormal vibrations, which converts vibration intensities into standard intervals for accurate detection across varying rolling speeds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vibration detection is performed using conventional frequency analysis methods, then vibration sources can be identified, but erroneous detection occurs due to noise from equipment and vibration from multiple sources in continuous cold rolling mills
Solution Approach 1:
The patent segments the vibration detection process into multiple independent analysis stages: collecting vibration data from multiple detectors, performing frequency analysis to obtain spectral data, conducting principal component analysis to extract dominant vibration patterns, and comparing against reference data. This segmentation allows each stage to process specific aspects of vibration separately, improving the ability to distinguish chatter marks from other vibration sources and reducing erroneous detection.
Solution Approach 2:
The patent introduces principal component analysis as an intermediary processing step between raw vibration data and final detection results. This intermediary transforms complex multi-source vibration data into simplified principal components that represent dominant vibration patterns, making it easier to identify chatter marks while filtering out noise from equipment and other sources.
2Quantity of substance
If multiple vibration detectors are deployed to capture all vibration sources, then comprehensive vibration data is obtained, but it becomes more difficult to distinguish abnormal vibration from normal operation vibrations
Solution Approach 1:
The patent extracts the essential vibration characteristics by performing frequency analysis to obtain spectral data, then applying principal component analysis to extract dominant vibration patterns from the complex multi-source data. This extraction process isolates the key features related to chatter marks while eliminating redundant information from normal operation vibrations, making abnormal vibration identification easier despite comprehensive data collection.
Solution Approach 2:
The patent transforms vibration data from the time domain to the frequency domain through frequency analysis, changing the parameter representation from time-based vibration signals to frequency-based spectral data. This parameter change enables better differentiation between chatter marks and other vibration sources by analyzing their distinct frequency characteristics.
3Productivity
If vibration detection is performed in continuous cold rolling mills with varying rolling speeds, then production continuity is maintained, but vibrations of plural frequencies are superimposed making chattering detection more difficult
Solution Approach 1:
The patent implements a dynamic detection approach that continuously collects vibration data during rolling operations and adapts the analysis to varying rolling speeds. By performing frequency analysis and principal component analysis on continuously acquired data, the system maintains detection capability despite changes in rolling speed and the resulting superposition of multiple vibration frequencies.
Solution Approach 2:
The patent performs preliminary frequency analysis and principal component analysis to establish reference data representing normal vibration patterns before attempting to detect chatter marks. This preliminary action creates a baseline for comparison, enabling the system to identify abnormal vibrations even in the complex environment of continuous cold rolling with varying speeds.
Data Source
AI summary
A method for detecting abnormal vibration of a rolling mill including a collecting step of collecting vibration data of the rolling mill, a frequency analysis step of generating first analysis data by performing frequency analysis of the vibration data, a principal component analysis step of performing principal component analysis on the first analysis data by using reference data specified in advance on the basis of a normal state as a principal component and thereby generating evaluation data, which is a projection of the first analysis data onto the reference data, and an abnormal vibration detection step of extracting an outlier component from the evaluation data and the first analysis data and detecting an abnormality of the rolling mill from the extracted outlier component.