Deep Learning Fatigue Crack Propagation Measurement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional fatigue crack propagation rate test methods struggle to accurately measure crack length and propagation rate, especially for non-standard samples, due to reliance on operator experience and limited adaptability to geometric changes, leading to challenges in repeatability and scalability.
Innovation Solution
A deep learning-based fatigue crack propagation rate test device and method utilizing a dual scale Faster-RCNN network for real-time crack detection and measurement, incorporating global and local scale identification modules to accurately measure crack length and propagate rate across various geometric sizes, overcoming limitations of traditional methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If visual measurement method with low-power microscope is used, then operator can manually read surface crack length, but measurement accuracy deteriorates due to operator fatigue, sense of responsibility and experience factors
Solution Approach 1:
The patent replaces the mechanical visual measurement method with an optical imaging system combined with image processing algorithms. A digital camera captures crack images, and image processing automatically measures crack length, eliminating operator fatigue and subjectivity. This substitution transforms manual mechanical reading into automated optical-digital measurement, significantly improving measurement precision while maintaining ease of operation.
2Extent of automation
If non-visual measurement method is used, then automation is achieved, but adaptability to non-standard test-pieces deteriorates due to reliance on calibration curves for standard geometric sizes
Solution Approach 1:
The patent changes the measurement parameters from calibration curve-based (requiring standard geometry) to image processing-based (geometrically flexible). By capturing crack images and processing them computationally, the system adapts to various test-piece geometries without requiring predefined calibration curves. This parameter change enables both automation and versatility, allowing measurement of non-standard test-pieces while maintaining automated operation.
3Extent of automation
If crack propagation extensometer is used, then automated measurement is achieved, but applicability to non-standard test-pieces deteriorates due to standard specification requirements
Solution Approach 1:
The patent creates a universal measurement system that can handle both standard and non-standard test-pieces. The image processing approach serves multiple functions: it measures crack length on standard specimens like extensometers do, but also adapts to non-standard geometries. This multi-functionality eliminates the limitation of extensometers while preserving automated measurement capability, making the system universally applicable across different test-piece types.
4Device complexity
If traditional measurement methods are used, then equipment complexity is low, but productivity deteriorates due to inability to meet repeatability and large scale test requirements
Solution Approach 1:
The patent implements a self-service measurement system where the imaging and image processing automatically perform crack length measurement without operator intervention. The system captures images, processes them through algorithms, and outputs measurements autonomously. This self-service capability ensures consistent repeatability across multiple tests and enables large-scale testing, while the added complexity is justified by the significant productivity improvement and elimination of manual measurement variability.
Data Source
AI summary
A fatigue crack propagation rate test device and method based on deep learning, comprises a dual scale Faster Region-based Convolutional Neural Network (Faster-RCNN) to accurately measure a crack length. The device can be used for tracking a crack propagation length of a non-standard test-piece having any geometric size. The method comprises: firstly, acquiring crack data sets of different scales by means of a camera; secondly, training the crack data sets by using the Faster-RCNN; then, constructing a global and local dual scale fast convolutional neural network, and predicting crack lengths under whole times of load cycle; and finally, fusing fracture mechanics to obtain a relationship between the fatigue crack propagation rate and a crack tip stress intensity factor.