Gas Turbine Fan Blade Abnormality Detection from Operating Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for inspecting fan blades of gas turbine engines for abnormalities are unreliable and cumbersome, often requiring the engine to be shut down and relying on human inspection, which is prone to errors due to variations in camera angles, lighting, and noise.
Innovation Solution
Utilizing image analytics techniques, such as neural networks and dimensional reduction methods, to analyze photographic images of fan blades for abnormalities without the need for blueprints or computer model renderings, and providing automated notifications for defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual inspection methods are used to examine fan blades, then human experts can identify abnormalities, but the process becomes unreliable and cumbersome due to variations in camera angles, lighting conditions, and engine vibration
Solution Approach 1:
The patent replaces manual human inspection with an automated image analytics system using neural networks. The system captures images of fan blades during engine operation and uses machine learning models to automatically detect abnormalities, eliminating the need for human inspectors to manually examine blades under varying conditions. This substitution of mechanical/manual processes with automated intelligent systems directly resolves the contradiction by improving reliability through consistency while reducing operational complexity.
Solution Approach 2:
The neural network system performs self-learning and self-improvement through continuous training with augmented image data. The system automatically adapts to different camera angles, lighting conditions, and vibration patterns without requiring human intervention for each inspection scenario. This self-service capability enables the system to maintain high reliability across varying operational conditions while keeping the inspection process simple and automated.
2Measurement precision
If traditional inspection methods require the engine to be shut down and removed from the aircraft, then detailed examination can be performed, but significant loss of time occurs
Solution Approach 1:
The system performs preliminary inspection actions by continuously monitoring fan blades during normal engine operation. Images are captured and analyzed in real-time or near-real-time, allowing abnormalities to be detected before the engine requires shutdown for maintenance. This preliminary detection capability eliminates the need to wait for scheduled maintenance windows, thereby reducing inspection time while maintaining detection accuracy through advanced image analytics.
Solution Approach 2:
The inspection system operates continuously during engine operation rather than requiring periodic shutdowns. The image capture and analysis process runs continuously or at regular intervals, ensuring that fan blades are constantly monitored for abnormalities. This continuous inspection approach maintains high detection accuracy while minimizing time loss by eliminating the need to stop the engine for routine examinations.
3Extent of automation
If image analytics with neural networks are implemented, then automated and reliable abnormality detection is achieved, but device complexity increases
Solution Approach 1:
The system uses image copies and digital representations of fan blades instead of requiring physical access to the actual blades during inspection. Digital images are captured, stored, and analyzed through neural networks, creating a virtual copy of the inspection process. This copying approach enables full automation of the inspection workflow while managing system complexity by working with digital data rather than complex mechanical inspection apparatus.
Solution Approach 2:
The neural network inspection system is designed to be universal, handling multiple inspection scenarios, abnormality types, and engine configurations through a single integrated platform. The system can process images from different camera angles, lighting conditions, and blade types using the same core algorithm, reducing the need for multiple specialized systems and thereby managing overall device complexity while maintaining high automation levels.
Data Source
AI summary
Methods of inspecting fan blades of a gas turbine engine for abnormalities are disclosed. One method utilizes a neural network to determine whether a photographic image of at least one fan blade depicts an abnormality of the at least one fan blade. Another method uses a dimensional reduction technique on one or more first photographic images that depict at least one first fan blade to obtain basis vectors, utilizes the basis vectors to obtain a reconstructed version of a second photographic image that depicts at least one second fan blade, and determines whether the at least one second fan blade includes an abnormality based on a difference between the second photographic image and the reconstructed version of the second photographic image. Another method determines whether a fan blade includes an abnormality based on contrast differences between adjacent areas of an image.