基于物理信息的机器学习实时长视野温度场预测金属增材制造中的金属增材制造
Mingxuan Tian1, Haochen Mu2,3,4, Tao Liu1
1School of Mechanical and Power Engineering, Nanjing Tech University, Nanjing, 211816, China.
Communications engineering
|September 29, 2025
概括
这项研究引入了一种新的基于物理的神经网络,用于在电弧增材制造中准确的实时温度预测. 该模型通过最大限度地减少预测错误和减少训练时间来增强过程控制.
科学领域:
- 材料科学 材料科学 材料科学
- 制造业 工程 制造工程
- 计算科学 计算科学
背景情况:
- 准确的实时温度预测对于弧增材制造 (WAAM) 的工艺控制和质量保证至关重要.
- 传统的有限元素方法 (FEM) 是计算密集型的,而现有的数据驱动模型则在错误积累和适应性方面扎.
- 开发高效和准确的预测模型 WAAM 热行为仍然是一个重大挑战.
研究的目的:
- 开发一个基于物理学的几何循环神经网络 (PI-GRNN),用于实时,长时间地平线的温度预测.
- 将几何特征和物理约束集成到深度学习框架中,以提高预测准确性和适应性.
- 利用转移学习来提高 WAAM 实际应用的模型效率.
主要方法:
- 提出了一个基于物理学的几何循环神经网络,其中包含卷积长期短期记忆 (ConvLSTM) 细胞,用于时空特征提取.
- 通过硬编码初始/边界条件和使用物理知情损失函数来强制执行物理一致性.
- 应用转移学习技术以优化模型培训效率.
主要成果:
- 该PI-GRNN模型使用当前数据证明了有效的实时温度场预测,用于未来的时间视界 (例如1.25s).
- 该模型在模拟和实验WAAM数据上实现了4.5-13.9%的最大预测误差.
- 整合几何和物理信息将最大误差降低了约1%,而全PI-GRNN模型将其降低了4%.
- 转移学习将培训时间减少约50%,而不会影响预测准确度.
结论:
- 拟议的基于物理的几何循环神经网络为WAAM的实时温度预测提供了一个计算效率高,准确的解决方案.
- 与纯粹数据驱动的方法相比,整合几何特征和物理定律显著提高了模型性能.
- 转移学习为在工业WAAM环境中部署先进的预测模型提供了实用途径,减少了开发时间和成本.
相关概念视频
01:12Temperature Dependent Deformation
365
In a nonhomogeneous rod made up of steel and brass, restrained at both ends and subjected to a temperature change, several steps are involved in calculating the stress and compressive load. Due to the problem's static indeterminacy, one end support is disconnected, allowing the rod to experience the temperature change freely. Next, an unknown force is applied at the free end, triggering deformations in the rod's steel and brass portions. These deformations are then calculated and added...
365
01:17Theory of Metallic Conduction
1.7K
The conduction of free electrons inside a conductor is best described by quantum mechanics. However, a classical model makes predictions close to the results of quantum mechanics. It is called the theory of metallic conduction.
In this theory, Newton's second law of motion is used to determine the acceleration of an electron in the presence of an applied electric field. Then, its velocity is expressed via this acceleration.
An electron moves through the crystal, containing positive ions,...
In this theory, Newton's second law of motion is used to determine the acceleration of an electron in the presence of an applied electric field. Then, its velocity is expressed via this acceleration.
An electron moves through the crystal, containing positive ions,...
1.7K
02:37Metallic Solids
20.5K
Metallic solids such as crystals of copper, aluminum, and iron are formed by metal atoms. The structure of metallic crystals is often described as a uniform distribution of atomic nuclei within a “sea” of delocalized electrons. The atoms within such a metallic solid are held together by a unique force known as metallic bonding that gives rise to many useful and varied bulk properties.
All metallic solids exhibit high thermal and electrical conductivity, metallic luster, and malleability....
All metallic solids exhibit high thermal and electrical conductivity, metallic luster, and malleability....
20.5K