Component Fault Diagnosis Using Correlated Event Data Residuals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing diagnostic methods for technical systems often lead to operational disruptions and increased costs due to misdetected component faults, as they struggle to differentiate between real faults and anomalies caused by environmental conditions.
Innovation Solution
A method utilizing regression analysis of event data from components of the same type, under similar environmental conditions, to identify faults by calculating residual values and determining if they exceed predetermined thresholds, thereby correlating data to accurately detect and verify component faults without additional measuring equipment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensors are used to monitor component conditions in real-time, then system availability is improved, but false alarms increase due to environmental conditions causing misdetection
Solution Approach 1:
The patent combines data from multiple sensors monitoring different components of the same type into a unified analysis system. By merging the operational data streams and performing joint evaluation, the system distinguishes between environmental effects (affecting all components) and actual faults (affecting individual components), thereby reducing false alarms while maintaining high system availability
Solution Approach 2:
The system implements feedback by continuously comparing actual sensor readings against predicted values derived from correlated component data. When deviations exceed thresholds, the system generates alerts while using feedback from other components to verify whether the anomaly represents a true fault or environmental interference, improving detection accuracy
2Measurement precision
If correlation analysis of multiple components is performed, then fault detection accuracy is improved, but diagnostic complexity increases
Solution Approach 1:
The patent segments the diagnostic process into distinct modules: data collection from multiple components, correlation analysis engine, deviation calculation module, and threshold comparison system. This segmentation allows complex multi-component analysis to be performed through manageable, independent functional blocks, reducing overall system complexity while maintaining high detection accuracy
Solution Approach 2:
The diagnostic system employs universal algorithms that can analyze any component type through the same correlation and deviation processes. The same mathematical framework and analysis methodology apply across different component groups, eliminating the need for component-specific complex analysis routines and simplifying the overall diagnostic architecture
Data Source
Figure 1
Figure 2
Figure 3a~3b
AI summary
The described subject matter relates to a method and a diagnostic device 20 for identifying and verifying technical faults of components 2a, 2b, 4a - 4h of a technical system such as a train 1, a wind turbine or an elevator. In order to determine whether an error/event detected by a sensor unit 3a, 3b, 5a - 5h of a component 2a, 2b, 4a - 4h constitutes a technical fault of said component 2a, 2b, 4a - 4h, a regression analysis of the component's event data is carried out and a technical fault is determined based on the residual values of the event data.