Manufacturing Abnormality Analysis Using Overall and Element Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional abnormality analysis devices in manufacturing systems fail to accurately detect unauthorized access due to the presence of multiple sequence candidates, leading to difficulties in determining whether extracted sequence candidates are indeed unauthorized accesses.
Innovation Solution
An abnormality analysis device that calculates an overall abnormal degree for the manufacturing system and individual abnormal degrees for each constituent element, determining if the overall abnormal degree exceeds a threshold value, thereby allowing for accurate detection of abnormalities with reduced calculation overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional analysis devices extract multiple sequence candidates for unauthorized access, then the detection coverage is improved, but the accuracy of detecting actual unauthorized access deteriorates due to false positives from multiple candidates
Solution Approach 1:
The patent segments the analysis process into two distinct phases: first analyzing individual sequence candidates independently to generate initial abnormality indicators, then analyzing the combination of multiple sequence candidates together to determine the final unauthorized access detection result. This segmentation allows each phase to focus on specific aspects, improving overall detection accuracy while managing complexity.
Solution Approach 2:
The patent introduces an intermediary analysis step that evaluates the combined effect of multiple sequence candidates before making the final unauthorized access determination. This intermediary layer acts as a mediator between individual sequence analysis and final detection, resolving conflicts between multiple candidates and improving accuracy by considering their collective behavior rather than treating them in isolation.
2Measurement precision
If comprehensive analysis of all sequence candidates is performed, then detection accuracy is improved, but calculation overhead and processing time increase significantly
Solution Approach 1:
The patent performs preliminary analysis on individual sequence candidates first, generating initial abnormality indicators before combining them for comprehensive analysis. This preliminary action filters out obviously normal sequences early in the process, reducing the computational burden of subsequent comprehensive analysis while maintaining detection precision for suspicious sequences.
Solution Approach 2:
The patent applies partial analysis to individual sequence candidates initially, focusing computational resources on analyzing each candidate's basic characteristics without performing full comprehensive analysis on all candidates simultaneously. This partial action approach maintains detection precision for abnormal sequences while significantly reducing overall calculation overhead by avoiding redundant computations on normal sequences.
Data Source
AI summary
An abnormality analysis device including: an overall information obtainer that obtains overall information indicating an overall feature amount of a manufacturing system; an overall abnormal degree calculator that calculates an overall abnormal degree that is an abnormal degree of a whole of the manufacturing system by statistically processing the overall information; an individual information obtainer that obtains individual information indicating a feature amount of each of the plurality of constituent elements; an individual abnormal degree calculator that calculates an individual abnormal degree that is an abnormal degree of each of the plurality of constituent elements by statistically processing the individual information; and a determiner that determines whether or not the overall abnormal degree exceeds a threshold value, wherein the individual abnormal degree calculator calculates the individual abnormal degree when the determiner determines that the overall abnormal degree exceeds the threshold value.


