Adaptive Thermal Diffusivity Kernel for Additive Manufacturing Phase Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In additive manufacturing, inferring material phase conditions from voxel temperature is inaccurate due to varying phase conditions at the same temperature, and thermal diffusivity is anisotropic and heterogeneous, complicating the detection of sintering conditions and geometrical accuracy.
Innovation Solution
An adaptive thermal diffusivity kernel is predicted using a machine learning model constrained by a physical heat transfer model, which accounts for neighboring voxel properties and phase conditions, enabling in-situ detection of particle sintering conditions and improving geometrical accuracy by modifying thermal applications in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If voxel temperature is used to infer material phase conditions, then temperature measurement is simple, but phase detection accuracy deteriorates because voxels at the same temperature can have different phase conditions
Solution Approach 1:
The patent introduces thermal diffusivity as an intermediary parameter between temperature and phase condition. Instead of directly inferring phase from temperature, the system uses thermal diffusivity (which captures heat transfer characteristics) as a mediator that provides more nuanced information about material phase, thereby improving detection accuracy while maintaining practical measurability
Solution Approach 2:
The patent replaces direct temperature-based phase inference with a machine learning model that processes thermal diffusivity data. This substitution transforms the simple temperature measurement approach into a more sophisticated system that uses learned patterns from thermal behavior to accurately determine phase conditions, resolving the contradiction between measurement simplicity and detection accuracy
2Measurement precision
If thermal diffusivity is used to detect sintering conditions, then phase detection accuracy is improved, but device complexity increases due to anisotropic and heterogeneous properties requiring machine learning models
Solution Approach 1:
The patent changes the detection parameter from simple temperature to thermal diffusivity, which is a derived parameter that captures the dynamic heat transfer behavior of the material. This parameter change enables more accurate phase detection because thermal diffusivity reflects the actual thermal behavior during sintering, and the machine learning model learns to interpret this parameter's variations to determine phase conditions accurately
Solution Approach 2:
The system uses the existing thermal sensing infrastructure to gather temperature data, then automatically processes this data through machine learning models to extract thermal diffusivity information. The system serves itself by transforming readily available temperature measurements into more informative thermal diffusivity metrics without requiring additional complex hardware, thereby managing complexity while improving accuracy
3Manufacturing precision
If real-time thermal applications are modified based on material phase detection, then geometrical accuracy is improved, but processing time increases due to active closed-loop control requirements
Solution Approach 1:
The patent implements a closed-loop control system where material phase detection provides real-time feedback to the thermal application process. The machine learning model continuously monitors thermal diffusivity to determine phase conditions, and this information feeds back to adjust thermal parameters dynamically. This feedback mechanism enables the system to maintain optimal sintering conditions throughout the manufacturing process, improving geometrical accuracy by preventing defects while managing processing time through efficient real-time control
Solution Approach 2:
The system uses the machine learning model to predict phase conditions and thermal behavior in advance, allowing proactive adjustments to thermal applications before defects occur. By detecting phase changes early and anticipating future thermal requirements, the system can prepare and apply appropriate thermal parameters ahead of time, thereby improving accuracy without significant delays in the manufacturing process
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
This approach enhances the detection of material phase changes and sintering conditions, improving the geometrical accuracy of 3D objects during manufacturing by providing real-time feedback for active closed-loop control and reducing sintering errors.
Implementation Method 1
An adaptive thermal diffusivity kernel may be predicted based on a machine learning model that is constrained with a physical model
Implementation Method 2
thermal energy may be projected over material in a build area, where a phase change and solidification in the material may occur
Implementation Method 3
An example of a physical model is a heat transfer model
Data Source
AI summary
Examples of methods for detecting a material phase are described herein. In some examples, a kernel is predicted based on an input corresponding to an object and based on a machine learning model. In some examples, the machine learning model is constrained with a physical model. In some examples, a material phase is detected based on the kernel.