Laser Weld Fault Monitoring via CNN Height Profile Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fault monitoring systems for laser welding processes are inefficient due to the need for time-consuming 3D scans and inaccurate analysis of weld seam quality using two-dimensional greyscale images, which complicates integration into ongoing manufacturing processes.
Innovation Solution
A system utilizing a convolutional neural network to generate a height profile from two-dimensional image data, allowing for rapid and accurate fault monitoring without the need for continuous 3D scans, using a camera aligned coaxially with the laser processing head and optionally an OCT scanner for training data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D scans are used for weld seam quality monitoring, then measurement precision is improved, but recording time increases and integration into ongoing manufacturing processes becomes difficult
Solution Approach 1:
The patent uses a camera to capture two-dimensional images of the weld seam, creating a visual copy that serves as the basis for height profile reconstruction through neural network processing. This copying approach replaces time-consuming 3D scanning while maintaining sufficient measurement precision for quality control applications.
Solution Approach 2:
The patent replaces the mechanical 3D scanning system with an optical imaging system (camera) combined with computational processing using a convolutional neural network. This substitution eliminates the need for physical 3D measurement apparatus while achieving comparable height profile accuracy through image-based reconstruction.
2Loss of time
If two-dimensional greyscale images are used for weld seam analysis, then recording time is reduced, but height structure analysis accuracy deteriorates
Solution Approach 1:
The patent transforms two-dimensional image data into three-dimensional height profile information through neural network processing. The convolutional neural network learns to predict height values from 2D grayscale intensity patterns, effectively adding the third dimension (height) to the analysis without requiring 3D scanning hardware or increased recording time.
Solution Approach 2:
The convolutional neural network serves as an intermediary that bridges the gap between 2D image data and 3D height profile information. The network learns the complex mapping relationship between grayscale intensity patterns and surface height, enabling accurate height structure analysis from 2D images without direct 3D measurement.
3Measurement precision
If complex systems with multiple cameras and 3D scanners are used, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent makes a single camera perform multiple functions: capturing weld seam images, providing training data for the neural network, and serving as the basis for height profile reconstruction. This multi-functionality eliminates the need for separate 3D scanners, multiple cameras, and complex synchronization systems while maintaining measurement precision.
Solution Approach 2:
The patent changes the approach from direct physical measurement (3D scanning) to computational inference (neural network prediction). By transforming the measurement paradigm from hardware-based to software-based, the system achieves comparable accuracy with significantly reduced device complexity.
Data Source
AI summary
A system for fault monitoring of laser welding processes on a component that is to be processed or has been processed by a laser processing apparatus includes an image recording device for creating two-dimensional image data of the component, and an evaluation unit configured to, based on the two-dimensional image data created by the image recording device, determine associated height values and create a height profile of the component by using a convolutional neural network.