Intelligent Learning Device for Gas Turbine Part State Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current physics-based models for predicting the life of gas turbine engine components are inadequate as they fail to account for wear's impact on the microstructure and do not consider non-engine parameters like operational environment and storage conditions, leading to conservative maintenance schedules.
Innovation Solution
A trained generative flow neural network is used to analyze initial micrographs and operating conditions, generating simulated micrographs and determining expected parameters, including predicted life and maintenance schedules, by integrating physics-based models and neural network-based artificial sample generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If physics-based models are used to determine part life, then maintenance can be scheduled based on expected operation times, but the models fail to account for microstructural changes and non-engine parameters leading to conservative estimates
Solution Approach 1:
The patent merges physics-based models with machine learning models to create a hybrid approach. The physics-based model provides fundamental degradation mechanisms while the machine learning model incorporates microstructural data and non-engine parameters, combining their strengths to achieve more accurate part life predictions without losing critical information from either source.
Solution Approach 2:
The patent introduces microstructural data as an intermediary between operating conditions and part life prediction. By analyzing microstructural changes through image processing and incorporating them into the prediction model, the system bridges the gap between operational parameters and actual degradation, enabling more accurate predictions that account for both engineering and non-engine factors.
2Reliability
If conservative worst case scenario estimates are used, then maintenance schedules can be established, but unnecessary replacements occur reducing productivity
Solution Approach 1:
The patent implements feedback loops where actual part condition data, microstructural analysis results, and operational data are continuously fed back into the prediction model. This allows the system to learn from real-world performance and adjust predictions dynamically, replacing parts only when actually needed rather than following conservative fixed schedules, thereby maintaining reliability while improving productivity.
Solution Approach 2:
The patent transitions from static conservative estimates to dynamic predictions that adapt to actual part condition. By continuously monitoring microstructural changes and operational parameters, the system adjusts part life predictions in real-time, allowing maintenance schedules to be optimized based on actual degradation rates rather than worst-case scenarios, thus improving engine operational efficiency.
3Device complexity
If traditional models are used, then simple calculation is possible, but they cannot predict wear impact on microstructure or factor in operational environment
Solution Approach 1:
The patent segments the prediction model into distinct modules: physics-based degradation models, machine learning models for microstructural analysis, image processing components, and operational parameter integration. This segmentation allows each component to specialize in specific aspects of wear prediction while working together as an integrated system, achieving high measurement precision without overwhelming complexity in any single component.
Data Source
AI summary
A tool for monitoring a part condition includes a computerized device having a processor and a memory. The computerized device includes at least one of a camera and an image input and a network connection configured to connect the computerized device to a data network. The memory stores instructions for causing the processor to perform the steps of providing an initial micrograph of a part to a trained model, providing a data set representative of operating conditions of the part to the trained model, and outputting an expected state of the part from the trained model based at least in part on the input data set and the initial micrograph.