Manufacturing Equipment Diagnosis via Grouped Data Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional manufacturing equipment diagnosis systems face challenges in predicting abnormalities and malfunctions, particularly when dealing with new issues and noise in data, and require extensive calculation and past knowledge, making them inefficient and unstable.
Innovation Solution
A manufacturing equipment diagnosis support system that collects and analyzes data from multiple similar apparatuses, grouping and comparing features to identify unusual phenomena without relying on extensive calculation or past knowledge, using data analysis range setting, grouping, feature extraction, and testing to detect abnormalities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data mining and big data analysis technologies are used to extract significant information from large amounts of data, then the ability to find regularity and predict abnormalities is improved, but the calculation processing time increases significantly
Solution Approach 1:
The patent segments the large dataset into multiple divided datasets based on time periods or conditions, then performs parallel calculation processing on each segment. This division allows the system to analyze data more efficiently without sacrificing detection accuracy, resolving the contradiction between thorough analysis and processing time.
Solution Approach 2:
Instead of analyzing all data equally, the patent applies partial action by focusing computational resources on specific divided datasets that are most likely to contain abnormality patterns. This selective approach reduces overall calculation time while maintaining the ability to detect significant abnormalities.
2Reliability
If conventional diagnosis systems rely on past knowledge and stored failure data to determine abnormalities, then detection accuracy for known issues is improved, but the system cannot cope with new abnormalities that have never occurred before
Solution Approach 1:
The patent inverts the conventional approach by not starting with past failure data, but rather by directly analyzing current operation data to detect deviations from normal patterns. This inversion allows the system to identify new abnormalities without relying on pre-stored knowledge of specific failure modes, thereby improving adaptability while maintaining reliability through statistical analysis.
Solution Approach 2:
The system performs self-service by automatically learning normal operation patterns from current data and independently identifying abnormalities without requiring external knowledge bases or pre-programmed failure scenarios. This enables the system to adapt to new equipment and abnormality types autonomously.
3Quantity of substance
If data analysis includes noise and unknown disturbance, then comprehensive data collection is improved, but the extraction of regularity becomes unstable
Solution Approach 1:
The patent extracts and separates noise and unknown disturbances from the operation data through filtering and preprocessing steps before performing regularity analysis. By taking out these destabilizing elements, the system can maintain stability in pattern recognition while still utilizing comprehensive data for detection.
Solution Approach 2:
The system changes parameters such as analysis time periods, data sampling rates, and statistical thresholds to optimize the balance between utilizing comprehensive data and maintaining stable regularity extraction. By adjusting these parameters, the system can accommodate noise while preserving detection accuracy.
Data Source
AI summary
A manufacturing equipment diagnosis support system includes: a data collector which collects and records respective data in plural apparatuses to be monitored provided in manufacturing equipment; a data analysis device; and a display. The data analysis device includes: a data analysis range setting unit which sets an analysis range of data, by an item of data and time period; a data grouping unit which classifies the data into a category based on specification and use condition of the apparatus to be monitored, and a category based on a physical quantity which the data shows, to group the data; a feature extracting unit which extracts a feature in each of the data items; an unusual phenomenon specifying unit which specifies an unusual phenomenon candidate item; and a testing unit which tests whether there is a significant difference or not between the unusual phenomenon candidate item and an other data item.