A physics-driven deep learning model for process-porosity causal relationship and porosity prediction with interpretability in laser metal deposition

Guo, Weihong; Tian, Qi; Guo, Shenghan; Guo, Y. B.

doi:10.1016/j.cirp.2020.04.049

Cited by 58 publications

(15 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They also must be carefully trained with available experimental data, and often require a lot of data to be trained upon. Such models are difficult to interpret, apply, or generalize for a broader set of process conditions [26].…”

Section: Machine Learning-based Models In Lmdmentioning

confidence: 99%

“…Recently, an emerging research area that integrates the physics-informed (or physicsbased, physics-driven, physics-guided) models and machine learning models is gaining increased interest [26][27][28][29][30]. These methods combine the strengths of machine learning and physical principles to decrease uncertainty from PDEs, data (noise), and learning models themselves (generalizing error) [27].…”

Section: Piml Models In Ammentioning

confidence: 99%

“…Recent research has also shown that incorporating physical data into the machine learning model can enhance porosity-predicting models for LMD processes. For example, Guo et al developed a physics-driven deep learning model for the process-porosity causal relationship and porosity prediction with interpretability [26]. Gawade et al developed simulated and empirical data-driven insight into supervised learning for porosity prediction [23].…”

Section: Piml Models In Ammentioning

confidence: 99%

“…Gawade et al developed simulated and empirical data-driven insight into supervised learning for porosity prediction [23]. In these works [23,26], the physics-informed insights are integrated into the features that are used in the machine learning models.…”

Section: Piml Models In Ammentioning

confidence: 99%

“…The design of the loss function is another important component in the design of machine learning models [33,34]. Different from previous the works [23,26], this paper focuses on the capability to combine simulated physical data with experimental data in the form of physics-informed custom loss functions built into deep learning models in the LMD process. This paper aims to explore the impact of several physics-informed loss functions on a baseline deep learning-only porosity prediction model.…”

Section: Piml Models In Ammentioning

confidence: 99%

See 4 more Smart Citations

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

Gawade

Guo

2022

Sensors

View full text Add to dashboard Cite

Physics-informed machine learning is emerging through vast methodologies and in various applications. This paper discovers physics-based custom loss functions as an implementable solution to additive manufacturing (AM). Specifically, laser metal deposition (LMD) is an AM process where a laser beam melts deposited powder, and the dissolved particles fuse to produce metal components. Porosity, or small cavities that form in this printed structure, is generally considered one of the most destructive defects in metal AM. Traditionally, computer tomography scans measure porosity. While this is useful for understanding the nature of pore formation and its characteristics, purely physics-driven models lack real-time prediction ability. Meanwhile, a purely deep learning approach to porosity prediction leaves valuable physics knowledge behind. In this paper, a hybrid model that uses both empirical and simulated LMD data is created to show how various physics-informed loss functions impact the accuracy, precision, and recall of a baseline deep learning model for porosity prediction. In particular, some versions of the physics-informed model can improve the precision of the baseline deep learning-only model (albeit at the expense of overall accuracy).

show abstract

Section: Machine Learning-based Models In Lmdmentioning

confidence: 99%

Section: Piml Models In Ammentioning

confidence: 99%

Section: Piml Models In Ammentioning

confidence: 99%

Section: Piml Models In Ammentioning

confidence: 99%

Section: Piml Models In Ammentioning

confidence: 99%

See 3 more Smart Citations

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

Gawade

Guo

2022

Sensors

View full text Add to dashboard Cite

show abstract

Review on Fatigue of Additive Manufactured Metallic Alloys: Microstructure, Performance, Enhancement, and Assessment Methods

Liu,

Yu,

Guo

et al. 2023

Advanced Materials

View full text Add to dashboard Cite

Additive manufacturing (AM), which is a process of building objects in a layer‐upon‐layer fashion from designed models, has received unprecedented attention from research and industry because it offers outstanding merits of flexibility, customization, reduced buy‐to‐fly ratio, and cost‐effectiveness. However, the fatigue performance of safety‐critical industrial components fabricated by AM is still far below that obtained from conventional methods. This review discusses the microstructural heterogeneities, randomly dispersed defects, poor surface quality, and complex residual stress generated during the AM process that can negatively impact the fatigue performance of as‐printed parts. The difference in microstructural origin of fatigue failure between conventionally manufactured and printed metals is reviewed with particular attention to the effects of the trans‐scale microstructures on AM fatigue failure mechanisms. Various methods for mitigating the fatigue issue, including pre‐process, inter‐process and post‐process treatments, are illustrated. Empirical, semi‐empirical and microstructure‐sensitive models are presented to predict fatigue strength and lifetime. Summary and outlooks for future development of the fatigue performance of AM materials are provided.This article is protected by copyright. All rights reserved

show abstract

Applications in Data-Driven Additive Manufacturing

Safdar

Lamouche

Paul

et al. 2023

SpringerBriefs in Applied Sciences and Technology

View full text Add to dashboard Cite

A physics-driven deep learning model for process-porosity causal relationship and porosity prediction with interpretability in laser metal deposition

Cited by 58 publications

References 6 publications

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

Review on Fatigue of Additive Manufactured Metallic Alloys: Microstructure, Performance, Enhancement, and Assessment Methods

Applications in Data-Driven Additive Manufacturing

Contact Info

Product

Resources

About