Compressor Operability Assessment Using ML Damage Surrogates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational fluid dynamic (CFD) techniques for predicting compressor operability in gas turbine engines are time-consuming, labor-intensive, and not sufficiently accurate, leading to potential unnecessary condemnation of components and increased costs due to inaccurate damage assessments and complex repair requirements.
Innovation Solution
A computer-implemented method using machine learning algorithms, specifically artificial neural networks, is employed to determine compressor operability by training the algorithms with data quantifying damage to compressor blades and operating parameters, allowing for faster and more accurate assessments of compressor damage and operability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computational fluid dynamic (CFD) techniques are used to model the effect of damage on compressor operability, then detailed analysis of compressor performance can be achieved, but the process becomes time-consuming and labor-intensive
Solution Approach 1:
The patent creates a simplified digital copy or surrogate model of the complex CFD system using machine learning algorithms. This surrogate model is trained on CFD simulation data to replicate the behavior of the full CFD analysis, enabling rapid predictions without repeating the computationally expensive original simulations. The machine learning model acts as a lightweight copy that preserves the essential predictive capabilities while eliminating the time-consuming aspects.
Solution Approach 2:
The patent performs preliminary CFD simulations to generate training data before deploying the machine learning model. By pre-computing a comprehensive dataset covering various damage scenarios and using it to train the ML algorithm, the system prepares the simplified model in advance. This preliminary action allows the trained ML model to provide rapid predictions later without needing to run full CFD analyses each time.
2Measurement precision
If computational fluid dynamic (CFD) techniques are used to model the effect of damage on compressor operability, then detailed analysis can be performed, but the process becomes labor-intensive requiring expert time
Solution Approach 1:
The patent replaces the complex mechanical/computational CFD analysis system with a machine learning-based predictive system. Instead of solving complex fluid dynamic equations through iterative numerical methods, the system uses trained ML algorithms that have learned the relationships between damage characteristics and compressor performance. This substitution maintains prediction accuracy while dramatically simplifying the analysis process and reducing the need for expert intervention.
Solution Approach 2:
The patent creates a simplified digital copy or surrogate model of the complex CFD system using machine learning algorithms. This surrogate model is trained on CFD simulation data to replicate the behavior of the full CFD analysis, enabling rapid predictions without repeating the computationally expensive original simulations. The machine learning model acts as a lightweight copy that preserves the essential predictive capabilities while eliminating the time-consuming aspects.
3Reliability
If inaccurate predictions of blade damage impact are made, then components may be condemned unnecessarily, but this results in significant cost penalties
Solution Approach 1:
The patent implements a feedback mechanism where the machine learning model's predictions are continuously refined through training on actual operational data and validation against known outcomes. The system learns from past assessments and adjusts its predictions to improve accuracy over time. This feedback loop ensures that the model becomes increasingly reliable at determining whether damaged components can continue to operate, reducing false condemnations and associated costs.
Data Source
AI summary
A computer-implemented method comprising: controlling input of data quantifying damage received by a compressor of a gas turbine engine into a first machine learning algorithm; receiving data quantifying a first operating parameter of the compressor as an output of the first machine learning algorithm; and determining operability of the compressor by comparing the received data quantifying the first operating parameter of the compressor with a threshold.