Abnormality Determination Using Scatter Degree of Prediction Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing abnormality determination devices struggle to accurately detect unstable states in inspection targets, which can lead to undetected faults, as they do not necessarily determine such states as abnormalities.
Innovation Solution
An abnormality determination device that calculates the scatter degree of differences between prediction information and observation information, using scenarios to represent state changes, and determines the abnormality based on this calculated degree.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing abnormality determination devices are used, then the inspection process is simple, but the unstable state of inspection targets cannot be accurately detected
Solution Approach 1:
The patent segments the abnormality determination process into three distinct modules: a calculation unit that computes scatter degrees of differences between prediction and observation information, a determination unit that compares scatter degrees against thresholds, and a scenario management unit that handles multiple state change scenarios. This segmentation enables accurate detection of unstable states while maintaining systematic organization and manageable complexity.
Solution Approach 2:
The patent introduces scatter degree as an intermediary metric that bridges prediction information and observation information. Instead of directly comparing raw data, the system calculates scatter degrees that quantify the uncertainty and differences, then uses these intermediary values for abnormality determination. This intermediary approach enhances detection accuracy for unstable states.
2Productivity
If site visits are conducted for inspection, then direct testing can be performed, but labor and time requirements increase with more inspection targets
Solution Approach 1:
The patent enables the inspection system to serve itself by automatically collecting observation information from inspection targets, calculating scatter degrees, and determining abnormalities without requiring human inspectors to visit each site. The system processes multiple targets simultaneously through automated data acquisition and analysis, dramatically improving productivity while reducing inspection time.
Solution Approach 2:
The patent replaces the mechanical inspection process (physical site visits, manual testing, and human judgment) with an automated information processing system. Observation devices automatically collect data, calculation units compute scatter degrees algorithmically, and determination units make abnormality assessments based on predefined thresholds, eliminating the need for manual inspection activities.
Data Source
AI summary
Provided are an abnormality determination device and the like that are able to accurately determine abnormalities regarding an inspection subject. The abnormality determination device is configured to calculate a scatter degree of differences between prediction information on an observation target and observation information on the observation target, the prediction information being an information generated in accordance with a scenario that represents an aspect of state change of the observation target, the observation information generated by an inspection target; and determine whether or not the inspection target is abnormal based on the calculated degree.