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: A deep convolution neural network was applied to detect three typical types of PBDs in a selective laser sintering (SLS) process, namely warpage, part shifting, and short feed, which were intentionally generated by varying the process conditions.
Abstract: The presence of defects in a powder bed fusion (PBF) process can lead to the formation of flaws in consolidated parts. Powder bed defects (PBDs) have different sizes and shapes and occur in different locations in the built area. Those variations pose great challenges to their detection. In this study, a deep convolution neural network was applied to detect three typical types of PBDs in a selective laser sintering (SLS) process, namely warpage, part shifting, and short feed, which were intentionally generated by varying the process conditions. Images of the powder bed were captured using a digital camera, which were split into three single-channel images corresponding to the color channels in the color image. A deep residual neural network was then used to extract multiscale features, and a region proposal network was adopted to detect the object-level defect bounding box. In the final stage, a fully convolutional neural network was proposed to generate instance-level defect regions in the bounding box. Our results demonstrated that this method had higher accuracy and efficiency and was able to cope with geometrical distortion and image blurring, in comparison to the defect detection methods reported previously. Also, the detection system was cost-effective and could be easily installed outside the chamber of a PBF system. This study lays the groundwork for the development of a variety of automated technologies for additive manufacturing, such as real-time powder layer quality inspection and 3D quality certificate generation for finish parts.
21 citations
TL;DR: In this article, the authors present a survey of the state-of-the-art metal-based additive manufacturing (MAM) process monitoring and control systems, and discuss the advantages and disadvantages of their algorithmic implementations and applications.
Abstract: Compared with other additive manufacturing processes, the metal-based additive manufacturing (MAM) can build higher precision and higher density parts, and have unique advantages in the applications to automotive, medical, and aerospace industries. However, the quality defects of builds, such as dimensional accuracy, layer morphology, mechanical and metallurgical defects, have been hindering the wide applications of MAM technologies. These decrease the repeatability and consistency of build quality. In order to overcome these shortcomings and to produce high-quality parts, it is very important to carry out online monitoring and process control in the building process. A process monitoring system is demanded which can automatically optimize the process parameters to eliminate incipient defects, improve the process stability and the final build quality. In this paper, the current representative studies are selected from the literature, and the research progress of MAM process monitoring and control are surveyed. Taking the key components of the MAM monitoring system as the mainstream, this study investigates the MAM monitoring system, measurement and signal acquisition, signal and image processing, as well as machine learning methods for the process monitoring and quality classification. The advantages and disadvantages of their algorithmic implementations and applications are discussed and summarized. Finally, the prospects of MAM process monitoring researches are advised.
21 citations
TL;DR: In this article, a statistical process monitoring approach is proposed to detect the onset of geometrical distortions during the build by comparing the slice contour reconstruction with the nominal geometry in each layer.
Abstract: In-situ layerwise imaging in laser powder bed fusion (L-PBF) has been implemented by many system developers to monitor the powder bed homogeneity. Increasing attention has been recently devoted to the possibility of using the same sensing approach to detect also in-plane and out-of-plane geometrical distortions of the part while it is being produced. To this aim, seminal works investigated the suitability of various image segmentation algorithms and assessed the accuracy of layerwise dimensional and geometrical measurements. Nevertheless, there is a lack of automated methods to identify, in-situ and in-process, geometrical defects and out-of-control deviations from the nominal geometry. This study presents a methodology that combines an active contours methodology for image segmentation with a statistical process monitoring approach suitable to deal with complex geometries that change layer by layer. The proposed approach enables a data-driven and automated alarm rule to detect the onset of geometrical distortions during the build by comparing the slice contour reconstruction with the nominal geometry in each layer. Moreover, by coupling edge-based and region-based segmentation techniques, the method is sufficiently robust to be applied to imaging and illumination setups that are already available on industrial L-PBF systems. The effectiveness of the proposed approach was tested on a real case study involving the L-PBF of complex Ti6Al4V parts that exhibited local geometrical distortions.
21 citations
TL;DR: In this article, the porosity resulting from powder bed fusion-electron beam melting (PBF-EB) was characterized over a series of 30 build cycles (consisting of ~ 480h cumulative build time) using X-ray Micro Computed Tomography (μCT).
20 citations
TL;DR: A weak supervision machine vision detection method based on artificial defect simulation for mobile phone screen defects detection and well-retrained model is used for defect recognition.
Abstract: During a practical detection process, insufficient defect data, unbalanced defect types and the high cost of defect labeling can present problems. Therefore, it often takes considerable time and labor to collect actual samples to improve the accuracy of defect classification and recognition. In this paper, we propose a weak supervision machine vision detection method based on artificial defect simulation. First, four typical mobile phone screen defects – scratches, floaters, light stains and severe stains – are simulated by the proposed synthesis algorithms, and an artificial defect database is created. Next, the artificial dataset is applied to a deep learning recognition algorithm, and an initial model is trained. Then, the collected actual defects are augmented due to the insufficient training quantity. The augmented actual defects are then applied as the training data, and the initial model is retrained by fine tuning. Finally, the well-retrained model is used for defect recognition. The experimental results demonstrate that satisfactory performance is achieved with the proposed detection method.
20 citations
References
More filters
TL;DR: High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated and the performance of the support- vector network is compared to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Abstract: The support-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.
High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
37,861 citations
01 Jun 2010
TL;DR: A brief overview of clustering is provided, well known clustering methods are summarized, the major challenges and key issues in designing clustering algorithms are discussed, and some of the emerging and useful research directions are pointed out.
Abstract: Organizing data into sensible groupings is one of the most fundamental modes of understanding and learning. As an example, a common scheme of scientific classification puts organisms into a system of ranked taxa: domain, kingdom, phylum, class, etc. Cluster analysis is the formal study of methods and algorithms for grouping, or clustering, objects according to measured or perceived intrinsic characteristics or similarity. Cluster analysis does not use category labels that tag objects with prior identifiers, i.e., class labels. The absence of category information distinguishes data clustering (unsupervised learning) from classification or discriminant analysis (supervised learning). The aim of clustering is to find structure in data and is therefore exploratory in nature. Clustering has a long and rich history in a variety of scientific fields. One of the most popular and simple clustering algorithms, K-means, was first published in 1955. In spite of the fact that K-means was proposed over 50 years ago and thousands of clustering algorithms have been published since then, K-means is still widely used. This speaks to the difficulty in designing a general purpose clustering algorithm and the ill-posed problem of clustering. We provide a brief overview of clustering, summarize well known clustering methods, discuss the major challenges and key issues in designing clustering algorithms, and point out some of the emerging and useful research directions, including semi-supervised clustering, ensemble clustering, simultaneous feature selection during data clustering, and large scale data clustering.
6,601 citations
TL;DR: The state-of-the-art of additive manufacturing (AM) can be classified into three categories: direct digital manufacturing, free-form fabrication, or 3D printing as discussed by the authors.
Abstract: This paper reviews the state-of-the-art of an important, rapidly emerging, manufacturing technology that is alternatively called additive manufacturing (AM), direct digital manufacturing, free form fabrication, or 3D printing, etc. A broad contextual overview of metallic AM is provided. AM has the potential to revolutionize the global parts manufacturing and logistics landscape. It enables distributed manufacturing and the productions of parts-on-demand while offering the potential to reduce cost, energy consumption, and carbon footprint. This paper explores the material science, processes, and business consideration associated with achieving these performance gains. It is concluded that a paradigm shift is required in order to fully exploit AM potential.
4,055 citations
TL;DR: Future directions such as the "print-it-all" paradigm, that have the potential to re-imagine current research and spawn completely new avenues for exploration are pointed out.
Abstract: Additive manufacturing (AM) is poised to bring about a revolution in the way products are designed, manufactured, and distributed to end users. This technology has gained significant academic as well as industry interest due to its ability to create complex geometries with customizable material properties. AM has also inspired the development of the maker movement by democratizing design and manufacturing. Due to the rapid proliferation of a wide variety of technologies associated with AM, there is a lack of a comprehensive set of design principles, manufacturing guidelines, and standardization of best practices. These challenges are compounded by the fact that advancements in multiple technologies (for example materials processing, topology optimization) generate a "positive feedback loop" effect in advancing AM. In order to advance research interest and investment in AM technologies, some fundamental questions and trends about the dependencies existing in these avenues need highlighting. The goal of our review paper is to organize this body of knowledge surrounding AM, and present current barriers, findings, and future trends significantly to the researchers. We also discuss fundamental attributes of AM processes, evolution of the AM industry, and the affordances enabled by the emergence of AM in a variety of areas such as geometry processing, material design, and education. We conclude our paper by pointing out future directions such as the "print-it-all" paradigm, that have the potential to re-imagine current research and spawn completely new avenues for exploration. The fundamental attributes and challenges/barriers of Additive Manufacturing (AM).The evolution of research on AM with a focus on engineering capabilities.The affordances enabled by AM such as geometry, material and tools design.The developments in industry, intellectual property, and education-related aspects.The important future trends of AM technologies.
1,792 citations
TL;DR: In this article, a review of additive manufacturing (AM) techniques for producing metal parts are explored, with a focus on the science of metal AM: processing defects, heat transfer, solidification, solid-state precipitation, mechanical properties and post-processing metallurgy.
Abstract: Additive manufacturing (AM), widely known as 3D printing, is a method of manufacturing that forms parts from powder, wire or sheets in a process that proceeds layer by layer. Many techniques (using many different names) have been developed to accomplish this via melting or solid-state joining. In this review, these techniques for producing metal parts are explored, with a focus on the science of metal AM: processing defects, heat transfer, solidification, solid-state precipitation, mechanical properties and post-processing metallurgy. The various metal AM techniques are compared, with analysis of the strengths and limitations of each. Only a few alloys have been developed for commercial production, but recent efforts are presented as a path for the ongoing development of new materials for AM processes.
1,713 citations