Abnormality Detection Model for Multi-Mode Production Facilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for determining abnormalities in production facilities operating in multiple modes fail to account for time-series relationships between operation modes, leading to inaccurate abnormality detection when the switching order differs from the predetermined order.
Innovation Solution
A model generation device specifies partial moving images for each operation mode and generates an abnormality determination model based on the time-series relationship between these images, enabling accurate abnormality detection in production facilities with multiple operation modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the learning model is constructed by clustering and learning collected moving images with similar scenes, then the model can be built using existing techniques, but the time-series relationship between operation modes cannot be reflected in the model
Solution Approach 1:
The patent segments the moving images by operation modes and extracts time-series relationships between different modes. The learning model is constructed to include both the clustered scene information and the temporal sequence information of mode transitions, thereby preserving the time-series relationship that was lost in conventional clustering approaches.
2Reliability
If conventional abnormality determination techniques are used, then the system can determine abnormalities based on scene similarity, but it cannot detect abnormalities when the switching order of operation modes differs from the predetermined order
Solution Approach 1:
The patent introduces a feedback mechanism where the learning model is trained with predetermined correct switching orders of operation modes. During abnormality determination, the system compares actual mode switching sequences against the learned correct sequences, providing feedback-based validation that enables detection of switching order abnormalities while maintaining reliable abnormality determination.
3Adaptability or versatility
If the production facility operates in multiple operation modes with time-series relationships, then the facility can produce multiple product types, but existing learning models cannot appropriately determine abnormalities in such multi-mode operations
Solution Approach 1:
The patent makes the learning model dynamic by incorporating time-series relationships between operation modes. The model adapts to different operating conditions by learning the temporal patterns of mode transitions, allowing it to maintain high abnormality detection accuracy across multiple operation modes while capturing the dynamic nature of production facility operations.
Data Source
AI summary
A model generation device (10) includes a specifier (103) that specifies partial moving images and a model generator (104) that generates an abnormality determination model. The partial moving images are included in a moving image acquired by imaging a production facility operable in multiple operation modes and are images for the respective operation modes. The model generator (104) generates, based on a time-series relationship between the partial moving images specified by the specifier (103) for the respective operation modes, the abnormality determination model for determination of whether an abnormality is present at the production facility based on the moving image acquired by imaging the production facility.