What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use Physics - Informed Machine Learning (PIML) technology to improve the accuracy, transparency and generalization ability of the model in the process of metal additive manufacturing (AM), especially in Laser Metal Deposition (LMD). Specifically, the paper focuses on: 1. **Limitations of data - driven machine - learning models**: The application of traditional data - driven machine - learning models in intelligent manufacturing is limited by the high cost of data acquisition and the "black - box" nature, and it is difficult to explain their prediction results. 2. **Insufficient effective utilization of physical laws**: Existing methods fail to effectively use physical laws to develop data - efficient machine - learning algorithms. 3. **Development of PIML models**: Develop PIML models by integrating neural networks and physical laws to overcome the above problems. This model can predict the temperature field in the LMD process without labeled training data, thereby achieving intelligent monitoring of the manufacturing process. ### Specific problem description - **Prediction of thermal behavior in the LMD process**: In the LMD process, the temperature distribution of materials is crucial for understanding phenomena such as molten pool formation, thermal stress and deformation. Although traditional numerical simulation methods such as the Finite Element Method (FEM) and Computational Fluid Dynamics (CFD) can provide detailed thermal behavior analysis, the computational cost is high and real - time monitoring cannot be achieved. - **Model training without labeled data**: The PIML model can be trained without labeled training data. By embedding physical laws (such as the energy conservation equation) into the loss function, the learning process of the model is guided. ### Research objectives - **Improve model accuracy**: Improve the accuracy of the model in predicting the temperature field in the LMD process by combining physical laws and neural networks. - **Enhance model transparency**: By introducing physical laws, make the prediction results of the model more transparent and easier to understand and explain. - **Improve model generalization ability**: Make the model better adapt to the LMD process under different conditions and improve its generalization ability. ### Method overview - **Embedding of physical laws**: Embed physical laws such as the energy conservation equation, initial conditions and boundary conditions into the loss function of the PIML model. - **Neural network architecture**: Design a neural network with multiple hidden layers, with the input being space - time coordinates and the output being temperature values. - **Model training**: Use the Adam optimizer and the L - BFGS optimizer for model training. By adjusting the weights and biases, make the model gradually adapt to the physical laws and boundary conditions of the system. ### Conclusions - **Feasibility of unlabeled data**: The PIML model can accurately predict the temperature field in the LMD process without using any labeled training data. - **Model performance**: The PIML model performs particularly well in predicting the top - surface temperature, with a relative error of only 2.1%, while the prediction error inside the domain is slightly higher, at 3.7%. - **Future research directions**: Further integrate more physical laws, predict multi - physical phenomena (such as residual stress), and evaluate the transfer learning ability of the model under parameter - changing conditions. In conclusion, this paper aims to solve the problem of temperature - field prediction in the LMD process through PIML technology and provide an efficient, accurate and transparent solution for intelligent manufacturing.

Physics-Informed Machine Learning for Smart Additive Manufacturing

Physics-Informed Machine Learning of Argon Gas-Driven Melt Pool Dynamics

Machine learning for metal additive manufacturing: Predicting temperature and melt pool fluid dynamics using physics-informed neural networks

Physics-Informed Machine Learning for Accurate Prediction of Temperature and Melt Pool Dimension in Metal Additive Manufacturing

Physics-informed machine learning and mechanistic modeling of additive manufacturing to reduce defects

A framework driven by physics-guided machine learning for process-structure-property causal analytics in additive manufacturing

Physics-Informed and Hybrid Machine Learning in Additive Manufacturing: Application to Fused Filament Fabrication

Thermal-Mechanical Physics Informed Deep Learning For Fast Prediction of Thermal Stress Evolution in Laser Metal Deposition

Accelerated and Inexpensive Machine Learning for Manufacturing Processes with Incomplete Mechanistic Knowledge

Towards a generic physics-based machine learning model for geometry invariant thermal history prediction in additive manufacturing

Progress and Opportunities for Machine Learning in Materials and Processes of Additive Manufacturing

When Physics Meets Machine Learning: A Survey of Physics-Informed Machine Learning

Intelligent laser-based metal additive manufacturing: A review on machine learning for process optimization and property prediction

A Comparative Study of Machine Learning Methods for Computational Modeling of the Selective Laser Melting Additive Manufacturing Process

Predicting melt track geometry and part density in laser powder bed fusion of metals using machine learning

Machine learning in additive manufacturing: State-of-the-art and perspectives

Physics-informed machine learning and its structural integrity applications: state of the art

Perspectives of using machine learning in laser powder bed fusion for metal additive manufacturing

Physics-Informed Machine Learning for Modeling and Control of Dynamical Systems

Machine learning techniques in additive manufacturing: a state of the art review on design, processes and production control