Neural Thermal Prediction for Voxel-Level 3D Printing Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Predicting transient thermal behavior in 3D printing is challenging due to the lack of quantitative knowledge on voxel-level thermal physics, including heat flux, thermal diffusivity, and convective thermal loss, which affects offline print tuning and online printing control.
Innovation Solution
The use of neural networks to compute predicted thermal behavior, such as heat maps, based on contone maps and thermal images, allowing for voxel-level energy control and improved printing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If traditional machining processes are used to create final parts, then material removal is straightforward, but material waste increases and production time extends
Solution Approach 1:
The patent inverts the traditional subtractive manufacturing approach by using additive manufacturing (3D printing) to build parts layer by layer from digital models. This inversion eliminates material removal entirely, achieving near-zero material waste while maintaining production efficiency through automated layer-by-layer deposition processes
Solution Approach 2:
The patent employs parameter changes by transforming the manufacturing paradigm from subtractive to additive processes. By changing the fundamental approach from removing material to adding material selectively, the system achieves both reduced material waste and maintained productivity through precise digital control of deposition parameters
2Manufacturing precision
If neural networks are used to predict thermal behavior, then printing accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by training neural networks offline to predict thermal behavior and generate optimized printing parameters before actual printing begins. This pre-computation stores thermal models and predictions in advance, allowing rapid inference during printing without real-time computational burden, thus improving accuracy while managing complexity
Solution Approach 2:
The patent uses copying by creating simplified thermal models and predictions through neural networks that replicate complex thermal physics behavior. These copied models provide accurate thermal behavior predictions without requiring full computational physics simulations during printing, reducing real-time computational complexity while maintaining manufacturing precision
3Manufacturing precision
If voxel-level thermal control is implemented, then printing quality improves, but process complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the build volume into discrete voxels and controlling thermal parameters at each voxel level independently. This segmentation enables precise local thermal control for improved printing quality while using automated algorithms to manage the complexity of coordinating thousands of voxel-level parameters through systematic layer-by-layer processing
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
Enables precise offline print tuning and online control, enhancing the accuracy and performance of 3D printing by predicting thermal behavior at a voxel level, thereby improving the quality of printed objects.
Implementation Method 1
heat flux (e.g., in-layer heat flux from neighboring voxels (anisotropic conductivity), heat flux from a layer below the voxel, heat flux from a layer above the voxel, etc.)
Implementation Method 2
convective thermal loss
Implementation Method 3
non-discriminative flux (convection, radiation)
Implementation Method 4
thermal energy may be projected over material in a build area, where a phase change and solidification in the material may occur
Implementation Method 5
thermal energy may be projected over material in a build area
Data Source
AI summary
Examples of a thermal behavior prediction method are described herein. In some examples of the thermal behavior prediction method, a predicted heat map of a layer corresponding to a three-dimensional (3D) model is computed using at least one neural network. The predicted heat map is computed based on a contone map corresponding to the 3D model.