Photopolymerized Ceramic Slicing Correction for Geometry Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The use of additive manufacturing machines for producing ceramic parts introduces inaccuracies and artifacts due to limitations in resolution and interaction between light energy and ceramic particles, leading to discrepancies in geometry.
Innovation Solution
A method utilizing neural networks to correct slicing images by training networks on initial and test piece images, minimizing differences to achieve precise geometric reconstruction, and applying optimized vectors to the additive manufacturing process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If slicing images are used by the additive manufacturing machine, then the three-dimensional geometry can be reproduced layer by layer, but discretization introduces artifacts and inaccuracies in the final part geometry
Solution Approach 1:
The method applies preliminary correction to the slicing images before manufacturing by training neural networks on reference images and test piece images. The corrected slicing images compensate for expected errors in advance, allowing the additive manufacturing machine to produce parts with geometry that more closely matches the reference design while maintaining the layer-by-layer manufacturing approach
Solution Approach 2:
The method implements feedback by comparing the actual geometry of manufactured test pieces (obtained through tomography or measurement) with the reference slicing images. This feedback loop trains neural networks to learn the discrepancy patterns, which are then used to correct future slicing images, continuously improving manufacturing precision
2Manufacturing precision
If the resolution of the polymerization element is increased to reduce artifacts, then geometry accuracy improves, but the machine complexity and cost increase
Solution Approach 1:
The method replaces the need for higher mechanical/optical resolution in the polymerization element with a computational approach using neural networks. Instead of upgrading the physical hardware to achieve better precision, the system uses software-based correction of slicing images to compensate for the limitations of the existing polymerization element resolution
Solution Approach 2:
The method changes the parameter being optimized from hardware resolution to image correction parameters. By adjusting the neural network correction parameters applied to slicing images, the system achieves improved geometry accuracy without modifying the physical resolution characteristics of the polymerization element
3Manufacturing precision
If light energy interacts with ceramic particles for photopolymerization, then the ceramic part is formed, but diffraction of light on particles causes geometric inaccuracies
Solution Approach 1:
The method converts the harmful effect of light diffraction on ceramic particles into a beneficial correction factor. By measuring and analyzing the geometric deviations caused by diffraction in test pieces, the neural networks learn to pre-correct slicing images in the opposite direction, transforming the diffraction effect from a source of error into a predictable and compensable phenomenon
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method ensures precise geometry matching, minimizes manufacturing errors, and enhances the ability to print complex geometries and small cavities with sub-pixel precision, addressing irregularities and diffraction issues.
Implementation Method 1
An additive manufacturing machine deposits several layers of a material to be photopolymerized according to slicing images
Implementation Method 2
the diffraction of light on the ceramic particles suspended in the polymerization bath
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to a method for manufacturing a part (P) of photopolymerized ceramic material using an additive manufacturing machine, comprising the following steps: training (E2) of a first neural network on the images (XAi), obtaining (E3) N third images (PRi) of the material of test parts (P'), training (E4) of a second neural network, correction, from the first and second networks having been trained, of one of the images (XAi) into a corrected image (Xopt(i)) of reference, the machine manufacturing (E10) the part (P) by depositing and photopolymerizing the N photopolymerization slices (Ti) in the N superimposed planes (Hi) of ceramic material deposition according to the first N two-dimensional reference images (XAi), the fourth image (xi) of which has been replaced by the corrected image (Xopt(i)).