Automatic Fault Tree Generation for Machine Fault Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of fault diagnosis using Fault Tree Analysis (FTA) depends on the skill and experience of the creator, leading to potential omissions or inaccuracies in fault tree construction and probability estimation, resulting in variations in diagnosis accuracy due to personal dependency.
Innovation Solution
A fault diagnosis device that automatically generates a fault tree for a machine by associating component and sensor faults, using input/output data to couple component trees and calculate abnormality degrees, enabling real-time fault analysis and alarm issuance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a fault tree is manually created by an expert, then comprehensive fault analysis is achieved, but personal dependency and omission of phenomena occur
Solution Approach 1:
The system performs automatic fault tree generation by having the machine itself provide operational data that feeds into the fault tree construction process. The operational data automatically collected from sensors and system logs serves as input for generating the fault tree, eliminating the need for manual expert analysis while reducing personal dependency in fault tree creation.
2Loss of information
If manual fault tree creation is performed, then fault modes can be identified, but time-consuming and experience-dependent process occurs
Solution Approach 1:
The system performs preliminary data collection and processing by continuously gathering operational data from the machine before fault analysis is needed. This pre-collected data is then automatically used to generate the fault tree when needed, eliminating the time-consuming manual creation process while ensuring comprehensive fault mode information is captured.
3Measurement precision
If probability of occurrence is defined for each phenomenon, then fault probability estimation is achieved, but accurate definition is difficult due to enormous use conditions
Solution Approach 1:
The system uses feedback from actual machine operational data to dynamically adjust and refine probability estimates for fault phenomena. Instead of manually defining probabilities for numerous use conditions, the system continuously monitors actual occurrences and uses this feedback to update probability values, improving accuracy while reducing the complexity of probability definition.
Data Source
AI summary
Personal dependency related to fault tree construction is reduced, and the reliability of an operating machine is improved by improving the accuracy of a fault diagnosis. The present invention provides a fault diagnosis device for a machine in operation, the device comprising: an abnormality degree analysis unit that calculates the abnormality degree of each component configuring the machine by comparing input/output data of the machine with a threshold value; a fault tree automatic generation unit that holds a fault tree of each component in which the fault of each component and the fault of a sensor in each component are associated with each other and generates the fault tree of the entire machine by coupling the fault trees of the components on the basis of a correlation between the input/output data of each component; a fault analysis unit that analyzes the fault of the machine on the basis of the abnormality degree and information of the fault tree of the entire machine; and a display unit that displays information analyzed by the fault analysis unit and issues an alarm.