Operation Management Device Failure Detection via Correlation Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional operation management devices lack the ability to clearly present failure generation points and causes in an easily understandable manner, particularly for administrators without extensive knowledge of the target system's structure and behavior, and struggle to accurately identify failures from time-series performance information.
Innovation Solution
An operation management device that collects performance information from managed devices, generates a correlation model based on time-series changes, and extracts failure periods by analyzing deviations from the model, providing clear visualization and analysis tools to administrators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional operation management devices are used to detect failures, then failure detection capability is provided, but the complexity of understanding system structure and behavior increases for administrators
Solution Approach 1:
The patent introduces correlation information as an intermediary between performance parameters and failure detection. Instead of requiring administrators to directly understand complex system structures, the system calculates correlation coefficients between performance parameters and uses these correlations as mediators to identify failures, thereby simplifying the understanding process while maintaining detection accuracy
Solution Approach 2:
The patent replaces the mechanical approach of directly analyzing system structure and behavior with an information-based approach using correlation calculations. By substituting complex structural analysis with correlation coefficient computations, the system makes failure detection more accessible to administrators without extensive system knowledge
2Quantity of substance
If performance information is displayed continuously in time series, then complete operational history is available, but it becomes difficult to identify failure points and causes
Solution Approach 1:
The patent extracts correlation information from the continuous time series performance data. By calculating correlation coefficients between different performance parameters and extracting only the relevant correlation patterns, the system separates useful failure-indicating information from the overwhelming continuous data, making failure points easily identifiable
Solution Approach 2:
The patent segments the continuous time series data into meaningful correlation relationships between different performance parameters. By organizing data according to correlation coefficients rather than displaying all data continuously, the system creates discrete, manageable segments that highlight failure patterns without overwhelming the administrator
3Measurement precision
If detailed analysis of performance information changes is performed, then failure causes can be specified, but the workload for administrators increases
Solution Approach 1:
The patent implements self-service by having the system automatically calculate correlation coefficients and identify failures without requiring administrator intervention in the analysis process. The system performs detailed correlation analysis autonomously and presents only the essential failure information, thereby maintaining high measurement precision while significantly reducing administrator workload
Solution Approach 2:
The patent introduces feedback mechanisms where the system continuously monitors performance parameters, calculates correlations, and provides real-time feedback about potential failures. This automated feedback loop eliminates the need for administrators to manually analyze performance changes, reducing workload while maintaining precise failure detection
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An operation management device includes: an information collection module which collects, from a managed device, first and second performance information showing a time series change in the performance information; a correlation model generation module which derives a correlation function between the first and second performance information and creates a correlation model based on the correlation function; a correlation change analysis module which judges whether or not the current first and second performance information acquired by the information collection module satisfy the relation shown by the conversion function between the first and second performance information of the correlation model within a specific error range; and a failure period extraction module which, when the first and second performance information does not satisfy the relation shown by the conversion function of the correlation model , extracts a period of that state as a failure period.