Failure Causal Model Design Support for Faster Root Cause Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing design support systems lack the ability to efficiently collect and analyze maintenance records to determine the cause of product failures, relying heavily on human input for failure inspection information which may not include necessary data for root cause analysis.
Innovation Solution
A design support system that utilizes a failure causal model to associate failure causes with symptoms, allowing maintenance personnel to input symptoms and receive updated probability information for determining the most likely cause, thereby reducing the number of steps required to investigate failures and informing design changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If maintenance personnel input detailed failure inspection information with high degree of freedom, then the system can record situations not captured by sensors, but the input process becomes time-consuming and may not include necessary root cause analysis data
Solution Approach 1:
The patent introduces a natural language processing intermediary that translates maintenance personnel's free-text symptom descriptions into structured failure cause analysis data. This mediator automatically extracts key information, identifies potential failure causes, and populates analysis fields without requiring manual structured input, thus reducing input time while preserving comprehensive failure information.
Solution Approach 2:
The system enables self-service by allowing maintenance personnel to simply describe failure symptoms in natural language without needing to understand complex failure analysis frameworks. The system automatically performs the complex tasks of information extraction, cause identification, and data structuring, making the process as simple as writing a free-text description.
2Measurement precision
If designers manually investigate failure causes using maintenance records, then comprehensive analysis is possible, but the number of investigation steps increases significantly
Solution Approach 1:
The patent replaces the mechanical manual investigation process with an automated information processing system. Natural language processing algorithms automatically analyze maintenance records, extract relevant failure information, identify potential causes, and generate analysis results, substituting the manual mechanical steps of reading, analyzing, and synthesizing information.
Solution Approach 2:
The system implements feedback by automatically presenting analyzed failure causes and evidence to designers, who can then verify and refine the analysis. This feedback loop allows the system to learn from designer corrections and improve future automated analyses, maintaining high accuracy while reducing manual investigation steps.
3Reliability
If the system requires comprehensive authenticity input for all failure causes, then analysis completeness is improved, but maintenance personnel time and effort increase
Solution Approach 1:
The patent applies partial action by requiring authenticity input only for the most likely failure causes identified by the system, rather than requiring comprehensive input for all possible causes. The natural language processing system prioritizes causes based on probability and relevance, allowing maintenance personnel to focus on validating the top candidates while accepting automated assessments for lower-priority causes.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
There is provided a design support system including a designer terminal, a maintenance personnel terminal, and a design support device, in which the design support device includes a failure causal model database, a failure causal model updating unit that updates the failure causal model based on an assumed cause of a failure of the input target equipment, a maintenance support unit that estimates a type of a failure having a high occurrence probability and the cause thereof based on a symptom of the input target equipment and the failure causal model, and outputs the estimated result to the maintenance personnel terminal, a maintenance record database that stores authenticity information input from the maintenance personnel terminal for the estimated type of the failure and the cause, and a failure cause analysis unit that updates the failure causal model based on the authenticity information stored in the maintenance record database, estimates a failure cause of the target equipment based on the updated failure causal model, and outputs the estimated result to the designer terminal.