Application of supervised machine learning for defect detection during metallic powder bed fusion additive manufacturing using high resolution imaging.
TL;DR: In this article, an in- situ defect detection strategy for powder bed fusion (PBF) AM using supervised machine learning is described, where multiple images were collected at each build layer using a high resolution digital single-lens reflex (DSLR) camera.
Abstract: Process monitoring in additive manufacturing (AM) is a crucial component in the mission of broadening AM industrialization. However, conventional part evaluation and qualification techniques, such as computed tomography (CT), can only be utilized after the build is complete, and thus eliminate any potential to correct defects during the build process. In contrast to post-build CT, in situ defect detection based on in situ sensing, such as layerwise visual inspection, enables the potential for in-process re-melting and correction of detected defects and thus facilitates in-process part qualification. This paper describes the development and implementation of such an in situ defect detection strategy for powder bed fusion (PBF) AM using supervised machine learning. During the build process, multiple images were collected at each build layer using a high resolution digital single-lens reflex (DSLR) camera. For each neighborhood in the resulting layerwise image stack, multi-dimensional visual features were extracted and evaluated using binary classification techniques, i.e. a linear support vector machine (SVM). Through binary classification, neighborhoods are then categorized as either a flaw, i.e. an undesirable interruption in the typical structure of the material, or a nominal build condition. Ground truth labels, i.e. the true location of flaws and nominal build areas, which are needed to train the binary classifiers, were obtained from post-build high-resolution 3D CT scan data. In CT scans, discontinuities, e.g. incomplete fusion, porosity, cracks, or inclusions, were identified using automated analysis tools or manual inspection. The xyz locations of the CT data were transferred into the layerwise image domain using an affine transformation, which was estimated using reference points embedded in the part. After the classifier had been properly trained, in situ defect detection accuracies greater than 80% were demonstrated during cross-validation experiments.
Citations
More filters
TL;DR: In this paper, the authors survey and assemble the knowledge existing in the literature regarding residual stresses in powder bed fusion (PBF) processes, highlighting the anisotropic nature of the stress fields.
Abstract: Metal additive manufacturing (AM) has garnered tremendous research and industrial interest in recent years; in the field, powder bed fusion (PBF) processing is the most common technique, with selective laser melting (SLM) dominating the landscape followed by electron beam melting (EBM). Through continued process improvements, these methods are now often capable of producing high strength parts with static strengths exceeding their conventionally manufactured counterparts. However, PBF processing also results in large and anisotropic residual stresses (RS) that can severely affect fatigue properties and result in geometric distortion. The dependence of RS formation on processing variables, material properties and part geometry has made it difficult to predict efficiently and has hindered widespread acceptance of AM techniques. Substantial investigations have been conducted with regards to RS in PBF processing, which have illuminated a number of important relationships, yet a review encompassing this information has not been available. In this review, we survey and assemble the knowledge existing in the literature regarding RS in PBF processes. A discussion of background mechanics for RS development in AM is provided along with methods of measurement, highlighting the anisotropic nature of the stress fields. We then review modeling efforts and in-process experimental measurements made to advance process understanding, followed by a thorough analysis and summary of the known relationships of both material properties and processing variables to resulting RS. The current state of knowledge and future research needs for the field are discussed.
307 citations
TL;DR: A comprehensive review on the state-of-the-art of ML applications in a variety of additive manufacturing domains can be found in this paper, where the authors provide a section summarizing the main findings from the literature and provide perspectives on some selected interesting applications.
Abstract: Additive manufacturing (AM) has emerged as a disruptive digital manufacturing technology. However, its broad adoption in industry is still hindered by high entry barriers of design for additive manufacturing (DfAM), limited materials library, various processing defects, and inconsistent product quality. In recent years, machine learning (ML) has gained increasing attention in AM due to its unprecedented performance in data tasks such as classification, regression and clustering. This article provides a comprehensive review on the state-of-the-art of ML applications in a variety of AM domains. In the DfAM, ML can be leveraged to output new high-performance metamaterials and optimized topological designs. In AM processing, contemporary ML algorithms can help to optimize process parameters, and conduct examination of powder spreading and in-process defect monitoring. On the production of AM, ML is able to assist practitioners in pre-manufacturing planning, and product quality assessment and control. Moreover, there has been an increasing concern about data security in AM as data breaches could occur with the aid of ML techniques. Lastly, it concludes with a section summarizing the main findings from the literature and providing perspectives on some selected interesting applications of ML in research and development of AM.
274 citations
TL;DR: In the authors’ perspective, in situ monitoring of AM processes will significantly benefit from the object detection ability of ML, and data sharing of AM would enable faster adoption of ML in AM.
Abstract: Additive manufacturing (AM) or 3D printing is growing rapidly in the manufacturing industry and has gained a lot of attention from various fields owing to its ability to fabricate parts with complex features. The reliability of the 3D printed parts has been the focus of the researchers to realize AM as an end-part production tool. Machine learning (ML) has been applied in various aspects of AM to improve the whole design and manufacturing workflow especially in the era of industry 4.0. In this review article, various types of ML techniques are first introduced. It is then followed by the discussion on their use in various aspects of AM such as design for 3D printing, material tuning, process optimization, in situ monitoring, cloud service, and cybersecurity. Potential applications in the biomedical, tissue engineering and building and construction will be highlighted. The challenges faced by ML in AM such as computational cost, standards for qualification and data acquisition techniques will also be discussed. In the authors’ perspective, in situ monitoring of AM processes will significantly benefit from the object detection ability of ML. As a large data set is crucial for ML, data sharing of AM would enable faster adoption of ML in AM. Standards for the shared data are needed to facilitate easy sharing of data. The use of ML in AM will become more mature and widely adopted as better data acquisition techniques and more powerful computer chips for ML are developed.
229 citations
Additional excerpts
...The feature vector is then fed to SVM image claissification algorithm to learn the defects such as under-melting, keyholing, and balling (Scime and Beuth 2019)....
[...]
...For instance, a hybrid ML algorithm was devised which uses hierarchical clustering to classify AM design features and support vector machine (SVM) to enhance the hierarchical clustering result in pursuit of finding the recommended AM design features (Yao et al....
[...]
...2019), and support vector machine (SVM) (Gobert et al....
[...]
...%) and the combination of SVM and principle component analysis (PCA) (90.1%)(Zhang et al....
[...]
...%) was found to have higher classification accuracy as compared to SVM (89.6...
[...]
TL;DR: It is shown that experiment- and simulation-based data mining in combination with machine leaning tools provide exceptional opportunities to enable highly reliant identification of fundamental interrelations within materials for characterization and optimization in a scale-bridging manner.
Abstract: Machine learning tools represent key enablers for empowering material scientists and engineers to accelerate the development of novel materials, processes and techniques. One of the aims of using such approaches in the field of materials science is to achieve high-throughput identification and quantification of essential features along the process-structure-property-performance chain. In this contribution, machine learning and statistical learning approaches are reviewed in terms of their successful application to specific problems in the field of continuum materials mechanics. They are categorized with respect to their type of task designated to be either descriptive, predictive or prescriptive; thus to ultimately achieve identification, prediction or even optimization of essential characteristics. The respective choice of the most appropriate machine learning approach highly depends on the specific use-case, type of material, kind of data involved, spatial and temporal scales, formats, and desired knowledge gain as well as affordable computational costs. Different examples are reviewed involving case-by-case dependent application of different types of artificial neural networks and other data-driven approaches such as support vector machines, decision trees and random forests as well as Bayesian learning, and model order reduction procedures such as principal component analysis, among others. These techniques are applied to accelerate the identification of material parameters or salient features for materials characterization, to support rapid design and optimization of novel materials or manufacturing methods, to improve and correct complex measurement devices, or to better understand and predict fatigue behavior, among other examples. Besides experimentally obtained datasets, numerous studies draw required information from simulation-based data mining. Altogether, it is shown that experiment- and simulation-based data mining in combination with machine leaning tools provide exceptional opportunities to enable highly reliant identification of fundamental interrelations within materials for characterization and optimization in a scale-bridging manner. Potentials of further utilizing applied machine learning in materials science and empowering significant acceleration of knowledge output are pointed out.
222 citations
TL;DR: In this paper, the authors examined advances in metal printing focusing on metallurgy, as well as the use of mechanistic models and machine learning and the role they play in the expansion of the additive manufacturing of metals.
Abstract: Additive manufacturing enables the printing of metallic parts, such as customized implants for patients, durable single-crystal parts for use in harsh environments, and the printing of parts with site-specific chemical compositions and properties from 3D designs. However, the selection of alloys, printing processes and process variables results in an exceptional diversity of microstructures, properties and defects that affect the serviceability of the printed parts. Control of these attributes using the rich knowledge base of metallurgy remains a challenge because of the complexity of the printing process. Transforming 3D designs created in the virtual world into high-quality products in the physical world needs a new methodology not commonly used in traditional manufacturing. Rapidly developing powerful digital tools such as mechanistic models and machine learning, when combined with the knowledge base of metallurgy, have the potential to shape the future of metal printing. Starting from product design to process planning and process monitoring and control, these tools can help improve microstructure and properties, mitigate defects, automate part inspection and accelerate part qualification. Here, we examine advances in metal printing focusing on metallurgy, as well as the use of mechanistic models and machine learning and the role they play in the expansion of the additive manufacturing of metals. Several key industries routinely use metal printing to make complex parts that are difficult to produce by conventional manufacturing. Here, we show that a synergistic combination of metallurgy, mechanistic models and machine learning is driving the continued growth of metal printing.
190 citations
References
More filters
TL;DR: It was observed that, in addition to static mechanical properties, the fatigue properties of the porous biomaterials are highly dependent on the type of unit cell as well as on porosity.
Abstract: Meta-materials are structures when their small-scale properties are considered, but behave as materials when their homogenized macroscopic properties are studied. There is an intimate relationship between the design of the small-scale structure and the homogenized properties of such materials. In this article, we studied that relationship for meta-biomaterials that are aimed for biomedical applications, otherwise known as meta-biomaterials. Selective laser melted porous titanium (Ti6Al4V ELI) structures were manufactured based on three different types of repeating unit cells, namely cube, diamond, and truncated cuboctahedron, and with different porosities. The morphological features, static mechanical properties, and fatigue behavior of the porous biomaterials were studied with a focus on their fatigue behavior. It was observed that, in addition to static mechanical properties, the fatigue properties of the porous biomaterials are highly dependent on the type of unit cell as well as on porosity. None of the porous structures based on the cube unit cell failed after 106 loading cycles even when the applied stress reached 80% of their yield strengths. For both other unit cells, higher porosities resulted in shorter fatigue lives for the same level of applied stress. When normalized with respect to their yield stresses, the S-N data points of structures with different porosities very well (R2>0.8) conformed to one single power law specific to the type of the unit cell. For the same level of normalized applied stress, the truncated cuboctahedron unit cell resulted in a longer fatigue life as compared to the diamond unit cell. In a similar comparison, the fatigue lives of the porous structures based on both truncated cuboctahedron and diamond unit cells were longer than that of the porous structures based on the rhombic dodecahedron unit cell (determined in a previous study). The data presented in this study could serve as a basis for design of porous biomaterials as well as for corroboration of relevant analytical and computational models.
304 citations
TL;DR: Multisensor data fusion in dimensional metrology is used in order to get holistic, more accurate and reliable information about a workpiece based on several or multiple measurement values from one or more sensors.
271 citations
TL;DR: This paper proposes a classification technique, which it is called the Genetic Kernel SVM (GK SVM), that uses Genetic Programming to evolve a kernel for a SVM classifier.
Abstract: The Support Vector Machine (SVM) has emerged in recent years as a popular approach to the classification of data. One problem that faces the user of an SVM is how to choose a kernel and the specific parameters for that kernel. Applications of an SVM therefore require a search for the optimum settings for a particular problem. This paper proposes a classification technique, which we call the Genetic Kernel SVM (GK SVM), that uses Genetic Programming to evolve a kernel for a SVM classifier. Results of initial experiments with the proposed technique are presented. These results are compared with those of a standard SVM classifier using the Polynomial, RBF and Sigmoid kernel with various parameter settings
157 citations
TL;DR: In this article, the authors studied the fatigue behavior of as-built Ti-6Al-4V (Ti64) samples and found that lack-of-fusion (LOF) defects were primarily responsible for fatigue crack initiation.
Abstract: Additive Manufacturing (AM) technologies are considered revolutionary because they could fundamentally change the way products are designed. Selective Laser Melting (SLM) is a metal based AM process with significant and growing potential for the manufacture of aerospace components. Traditionally a material needs to be listed in the Metallic Materials Properties Development and Standardization (MMPDS) handbook if it is to be considered certified. However, this requires a considerable amount of test data to be generated on the materials mechanical properties. Therefore, the MMPDS certification process does not lend itself easily to the certification of AM components as the final component can have similar mechanical properties to wrought alloys combined with the defects associated with traditional casting and welding technologies. These defects can substantially decrease the fatigue life of a fabricated component. The primary purpose of this investigation was to study the fatigue behaviour of as-built Ti-6Al-4V (Ti64) samples. Fatigue tests were performed on the Ti-6Al-4V specimens built using SLM with a variety of layer thicknesses and build (vertical or horizontal) directions. Fractography revealed the presence of a range of manufacturing defects located at or near the surface of the specimens. The experimental results indicated that Lack-of-Fusion (LOF) defects were primarily responsible for fatigue crack initiation. The reduction in fatigue life appeared to be affected by the location, size and shape of the LOF defect.
151 citations
TL;DR: In this paper, the authors investigate the possibility of in situ flaw detection for powder bed, beam-based additive manufacturing processes using a thermal imaging system and compare infrared images (IR) which were taken during the generation of Ti•6Al•4V parts in a selective electron beam melting system (SEBM) with metallographic images taken from destructive material investigation.
Abstract: Purpose – The purpose of this paper is to investigate the possibility of in situ flaw detection for powder bed, beam‐based additive manufacturing processes using a thermal imaging system.Design/methodology/approach – The authors compare infrared images (IR) which were taken during the generation of Ti‐6Al‐4V parts in a selective electron beam melting system (SEBM) with metallographic images taken from destructive material investigation.Findings – A good match is found between the IR images and the material flaws detected by metallographic techniques.Research limitations/implications – First results are presented here, mechanisms of flaw formation and transfer between build layers are not addressed in detail.Originality/value – This work has important implications for quality assurance in SEBM and rapid manufacturing in general.
127 citations