Industrial Anomaly Classification for Known and Unknown Faults
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging to classify all types of anomalies in industrial machines effectively due to the difficulty in collecting data during abnormal operations, which hinders the preparation of models for known anomalies and the identification of unknown anomalies.
Innovation Solution
An anomaly classification device that acquires, stores, and analyzes data from past anomalies to create models for determining whether an anomaly is known or unknown, enabling classification of anomaly causes without prior knowledge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If unsupervised learning is performed using data from normal operation to detect anomalies, then anomaly detection capability is improved, but the ability to classify anomaly causes accurately deteriorates
Solution Approach 1:
The patent segments the anomaly analysis process into two distinct stages: first, unsupervised learning for anomaly detection using normal operation data; second, supervised learning for anomaly cause classification using collected anomaly data. This segmentation allows each stage to optimize for its specific purpose without compromise.
Solution Approach 2:
The patent performs preliminary unsupervised learning to detect anomalies before attempting classification. By first identifying that an anomaly exists, the system can then focus resources on classifying the detected anomaly using the supervised learning model trained on anomaly data.
2Measurement precision
If models are prepared to classify all types of anomalies, then anomaly cause identification capability is improved, but data collection time and cost increase
Solution Approach 1:
The patent implements partial action by preparing classification models only for anomaly types that have been actually detected and collected data for. Rather than preparing models for all possible anomaly types in advance, the system dynamically creates models based on the specific anomalies encountered, reducing unnecessary data collection efforts.
Solution Approach 2:
The patent changes the parameter of model preparation from static (preparing all models in advance) to dynamic (preparing models based on detected anomaly types). This allows the system to adapt the level of classification detail to the actual needs revealed during operation.
3Measurement precision
If comprehensive anomaly classification models are created, then classification accuracy is improved, but device complexity increases
Solution Approach 1:
The patent segments the classification system into multiple specialized models, each trained on specific anomaly data. Rather than creating one complex model attempting to handle all anomaly types, the system uses multiple simpler models that can be selected and applied based on the detected anomaly type.
Solution Approach 2:
The patent implements a dynamic model selection mechanism where the system chooses which classification model to apply based on the characteristics of the detected anomaly. This allows the system to use only the necessary model complexity for each specific case rather than maintaining high complexity for all cases.
Data Source
AI summary
An anomaly classification device determines whether a detected anomaly is a known anomaly or an unknown anomaly and presents to the user how to manage the detected anomaly not only for a case of a known anomaly but also for a case of an unknown anomaly. This anomaly classification device acquires, as anomaly data, data related to a physical quantity detected when an anomaly occurred in an industrial machine, creates a model used for determining whether or not the anomaly data is anomaly data that is based on a known anomaly cause and a model used for classifying which anomaly cause the anomaly data belongs to, and uses the created models to determine whether or not the anomaly data is based on a known anomaly cause and classify which anomaly cause the anomaly data is based on.