Failure Mode Detection From Device Notifications Using Topic Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current maintenance policies in systems like manufacturing and production face challenges in efficiently identifying failure modes, leading to increased maintenance costs and reduced machine availability, reliability, and safety.
Innovation Solution
A system utilizing a two-stage data training approach with Latent Dirichlet Allocation (LDA) model for text analysis to determine failure modes from notifications, messages, and logs, enabling automated failure mode identification and root cause analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional failure identification procedures (FMEA, FMECA, FTA) are used, then comprehensive failure mode analysis is achieved, but maintenance costs increase and processing time is extended
Solution Approach 1:
The patent replaces traditional mechanical failure identification procedures (FMEA, FMECA, FTA) with an automated text analysis system using machine learning models. The system processes notifications, messages, and logs through trained classifiers to automatically determine failure modes, eliminating the need for manual analysis and significantly reducing processing time while maintaining identification accuracy.
Solution Approach 2:
The patent creates a trained classification model that copies the expertise of traditional failure analysis methods. By training the model on historical data labeled using conventional procedures, the system replicates their analytical capabilities in an automated digital form, enabling fast processing without sacrificing the comprehensive analysis quality of traditional methods.
2Reliability
If traditional failure identification procedures are implemented, then failure modes are identified, but maintenance costs constitute a major part of total operating costs
Solution Approach 1:
The patent implements a self-service failure identification system where the automated text analysis model processes notifications and determines failure modes without requiring external expert intervention. The system serves itself by automatically analyzing incoming data, classifying failures, and providing results that can directly trigger maintenance actions, eliminating the need for costly manual analysis while improving machine availability through faster response.
3Measurement precision
If automated text analysis with LDA model is used, then failure mode determination accuracy is improved, but system complexity increases
Solution Approach 1:
The patent segments the text analysis system into distinct functional modules: an LDA-based topic modeling component that identifies latent topics in notifications, and a classification component that maps topics to failure modes. This segmentation allows each module to specialize in one aspect of the analysis, improving overall accuracy while making the system more manageable and interpretable despite its complexity.
Data Source
AI summary
Some embodiments provide a program that retrieves a set of notifications describing failures that occurred on a set of monitored devices. The program further determines a set of topics based on the set of notifications. The program also determines failure modes associated with the set of topic from a plurality of failure modes defined for the set of monitored devices. The program further determines failure modes associated with the set of notifications based on the set of topics and the failure modes associated with the set of topics. The program also receives a particular notification that includes a particular set of words describing a failure that occurred on a particular monitored device in the set of monitored devices. The program further determines a failure mode associated with the particular notification based on the particular set of words and the determined failure modes associated with the set of notifications.