Blade Counting via HMM Chains for Rotating Jet Engine Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatic blade counting in aircraft jet engines is challenging due to the lack of reliable visual cues, self-similarity between blades, limited field of view, instability under varying illumination, and subtle tip movement causing motion blurring and drifting of the borescope pose during inspection.
Innovation Solution
A method using Hidden Markov Model (HMM) chains to represent the rotation cycles of blades, with initial and subsequent states on closed curves, and down-sampling, Gaussian smoothing, and principal component analysis to stabilize the counting process and account for pose changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional visual inspection methods are used with borescope, then inspection can be performed on rotating blades, but reliable visual cues are lacking due to smooth surface and self-similarity between blades
Solution Approach 1:
The patent applies artificial lighting to illuminate the blades during rotation, creating visible contrast and visual cues on the smooth blade surfaces. The lighting system projects light patterns that enhance the visibility of blade features and defects, transforming the otherwise featureless smooth surfaces into detectable visual targets.
Solution Approach 2:
The patent introduces an intermediary processing system that captures images from the borescope, processes them through algorithms to enhance features, and reconstructs blade positions. This intermediary system bridges the gap between the limited visual cues available and the need for reliable blade detection and counting.
2Ease of operation
If manual rotation of blades is performed during inspection, then operator can examine blades closely, but fast rotation causes motion blurring and only a few frames are captured per cycle
Solution Approach 1:
The patent employs periodic illumination patterns that synchronize with the blade rotation cycle. By projecting light at specific intervals matching the rotation period, the system ensures that blades are illuminated and captured at optimal moments, reducing motion blur effects and ensuring sufficient frames are captured per rotation cycle.
Solution Approach 2:
The system uses feedback from captured frames to adjust processing parameters in real-time. By analyzing the quality and clarity of captured frames, the system adapts its processing algorithms to compensate for motion blur and optimize the capture rate, ensuring adequate frame resolution even during fast rotation.
3Ease of operation
If borescope tip position is adjusted manually during inspection, then orientation can be optimized, but subtle hand shaking and mechanical vibration cause tip pose to drift throughout the rotation
Solution Approach 1:
The patent replaces manual mechanical adjustment of the borescope with an automated image processing system. Instead of relying on stable manual positioning, the system uses computational algorithms to track and compensate for pose drift in each captured frame, effectively substituting mechanical stability requirements with computational correction.
Solution Approach 2:
The system performs self-correction by automatically detecting and compensating for borescope pose drift through image processing. The algorithm continuously monitors frame characteristics and adjusts the reconstructed blade positions accordingly, allowing the system to maintain accuracy despite mechanical instability without external intervention.
4Measurement precision
If automatic blade counting is implemented, then defect localization can be improved, but random manual rotation and subtle tip movement make counting challenging
Solution Approach 1:
The patent applies preliminary processing steps to the captured images, including enhancement algorithms and feature extraction, before performing blade counting. By preparing the data in advance through systematic processing of illumination patterns and motion compensation, the system simplifies the subsequent counting task despite the challenging rotation conditions.
Solution Approach 2:
The system transitions from analyzing individual 2D frames to reconstructing 3D blade positions and trajectories over time. By adding the temporal dimension and reconstructing spatial relationships across multiple frames, the system can accurately count blades even when individual frames are blurred or partially obscured due to rotation and vibration.
Data Source
AI summary
This disclosure relates to method and system for visual inspection of rotating components. The method includes representing rotation cycles of a rotating component as spatial features based on video or image frames, ascertaining and/or evolving Hidden Markov Model (HMM) chains for the cycles, ascertaining a count of the rotating component in the frames and/or labelling the frames with ascertained states of the HMM chains.