Rotating Equipment Diagnostics Using Sensor and Text Case Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current diagnostic systems for gas turbines face challenges in automatically identifying relevant historic cases that incorporate both sensor and textual information, requiring extensive manual parameterization and lacking integration of multi-disciplinary expertise, leading to inefficient troubleshooting.
Innovation Solution
A method combining Natural Language Processing (NLP) and Deep Learning to integrate textual information with sensor data, using case-based reasoning to create unified representations for optimized diagnostics, reducing manual effort and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual documentation of diagnostic knowledge in ticketing systems is used, then ease of operation for technicians is improved, but information sharing and automated case discovery capability deteriorates
Solution Approach 1:
The system creates structured copies of unstructured diagnostic knowledge by extracting entities, symptoms, and solutions from free-text ticket annotations. This transforms individual technician notes into reusable knowledge units that can be automatically shared and queried across the system, resolving the contradiction between documentation ease and knowledge sharing.
Solution Approach 2:
A knowledge extraction intermediary layer is introduced between the ticketing system and diagnostic users. This intermediary automatically processes unstructured text into structured formats using NLP techniques, enabling both technicians to easily document and the system to automatically share knowledge without requiring manual restructuring of the original convenient documentation format.
2Ease of manufacture
If standard text similarity measures like TF/IDF are used, then implementation simplicity is improved, but diagnostic accuracy by ignoring sensor data deteriorates
Solution Approach 1:
The system merges multiple information sources including sensor data, text annotations, and event logs into a unified case representation. This combination allows the system to leverage both the simplicity of standard text measures and the precision of multi-modal data analysis, achieving accurate case matching without sacrificing implementation feasibility.
Solution Approach 2:
The case representation uses a composite structure combining different data types (sensor readings, text features, event sequences) with appropriate weighting. This composite approach enables the system to maintain implementation simplicity while achieving high diagnostic accuracy through multi-source information integration.
3Adaptability or versatility
If manual parameter tuning is required for diagnostic systems, then adaptability to specific cases is improved, but system complexity and manual effort deteriorates
Solution Approach 1:
The system performs self-parameterization by automatically learning optimal weights and parameters from historical diagnostic data. Instead of requiring manual tuning, the system adapts to specific diagnostic scenarios by training on past cases, thereby achieving adaptability without increasing manual complexity.
Solution Approach 2:
The system dynamically adjusts parameters based on learned patterns from data rather than fixed manual settings. This allows the diagnostic system to adapt to different gas turbine models and failure modes automatically, maintaining versatility while eliminating the need for complex manual parameterization.
4Measurement precision
If extensive manual parameterization is used to improve diagnostic accuracy, then measurement precision is improved, but productivity and troubleshooting time deteriorates
Solution Approach 1:
The system performs preliminary parameter optimization by pre-processing and structuring diagnostic data during system initialization or offline periods. This preliminary action prepares the data in advance, enabling fast accurate diagnostics during actual troubleshooting without requiring time-consuming manual parameterization at the point of need.
Solution Approach 2:
Manual parameter tuning is replaced with automated machine learning algorithms that learn optimal parameters from data. This substitution maintains high diagnostic accuracy while eliminating the productivity loss associated with manual parameterization, as the automated system rapidly adapts without human intervention.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates generally to a method and an apparatus for optimizing diagnostics of rotating equipment, in particular a gas turbine. The apparatus comprises: a) Means (11) for providing status information about status of the rotating equipment over a series of time windows whereby status can be derived from sensor features (10) of at least one available sensor taking measurements during a predefinable time period, b) means (20) for using deep learning which combines provided historic sensor information with sequence of events data indicating warnings and/or alerts of the rotating equipment, c) whereby status information is supplemented with via deep learning predicted probabilities whether a warning and/or an alert has occurred within a time window, d) means (24) for providing an amount of textual diagnostic knowledge cases, e) means (17) for extracting semantic information on text features from the textual diagnostic knowledge cases, and f) means (19) for combining status information and semantic information into a unified representation enabling optimization of the diagnostics.