Browsing by Subject "defect detection"
Now showing 1 - 9 of 9
- Results Per Page
- Sort Options
Item A Deep Learning Approach to Defect Detection in Additive Manufacturing of Titanium Alloys(University of Texas at Austin, 2021) Liu, X.; Mileo, A.In Additive Manufacturing (AM) of titanium alloys, the formation of defects in parts is typically related to the stability of the melt pool. With increased instability and size of the melt pool comes an increase in the level of emissions generated as the laser processes the material. Recent developments with in-situ monitoring and process control allows the collection of large amounts of data during the printing process. This includes data about emissions, which are made available as 2D representations in the form of colour images. However, it is still a manual process to inspect these 2D representations to identify defects, which hinders scalability. Given recent advances in Deep Learning for computer vision and the availability of large amounts of data collected from in-situ monitoring, our approach leverages Deep Learning techniques for characterizing abnormal emissions to automatically identify defects during the printing process. One of the challenges to apply deep learning in AM is the lack of proper labelled data for training the models. In this paper, we tackle this challenge by proposing an approach that uses transfer learning and fine-tuning on a pre-trained Convolutional Neural Network (CNN) model called VGG 16 to successfully train the deep model with a small labelled dataset. Results show good classification accuracy on the emission images obtained from the in-situ monitoring system, and improvements in classification of defects on a public industrial benchmark datasets named DAGM (Deutsche Arbeitsgemeinschaft für Mustererkennung e.V., German chapter of the IAPR).Item Development of a Standalone In-Situ Monitoring System for Defect Detection in the Direct Metal Laser Sintering Process(University of Texas at Austin, 2019) Quinn, Paul; O'Halloran, Sinead; Ryan, Catriona; Pamell, Andrew; Lawlor, Jim; Raghavendra, RameshDirect metal laser sintering (DMLS) is a powder bed fusion (PBF) additive manufacturing process commonly used within the medical device and aerospace industries where regulations drive the requirement for stringent quality control. Using in-situ monitoring, the identification of defects, as well as the geometric and dimensional measurement of the layers throughout the build allows for greater quality control, as well as a reduction in the requirement for ex-situ measurement. A standalone monitoring system for the EOS M280 is presented in this research, allowing for the build process to be monitored layer-by-layer. The system images the build area after powder deposition and after laser exposure allowing for the identification of inefficiencies in both the powder deposition and the laser exposure. The system has proven to be capable to identify in build defects and work is ongoing to develop an automated program to identify these defects and notify the operator in real time.Item Dynamic Defect Detection in Additively Manufactured Parts using FEA Simulation(University of Texas at Austin, 2019) Johnson, Kevin; Allen, Aimee; Blough, Jason; Barnard, Andrew; Labyak, David; Hartwig, Troy; Brown, Ben; Soine, David; Cullom, Tristan; Kinzel, Edward; Bristow, Douglas; Landers, RobertThe goal of this paper is to evaluate internal defects in additively manufactured (AM) parts using FEA simulation. The resonant frequencies of parts are determined by the stiffness and mass involved in the mode shape at each resonant frequency. Voids in AM parts will change the stiffness and mass therefore shift the resonant frequencies from nominal. This paper will investigate the use of FEA to determine how much a void size, shape, and location will change the resonant frequencies. Along with where the optimal input and response locations are in order to find these frequency changes. The AM part evaluated in this work includes a common tensile bar and hammer shaped part evaluated individually and as a set of parts that are still attached to the build plate.Item IN-SITU DEFECT DETECTION FOR LASER POWDER BED FUSION WITH ACTIVE LASER THERMOGRAPHY(University of Texas at Austin, 2023) Breese, P.P.; Becker, T.; Oster, S.; Metz, C.; Altenburg, S.J.Defects are still common in metal components built with Additive Manufacturing (AM). Process monitoring methods for laser powder bed fusion (PBF-LB/M) are used in industry, but relationships between monitoring data and defect formation are not fully understood yet. Additionally, defects and deformations may develop with a time delay to the laser energy input. Thus, currently, the component quality is only determinable after the finished process. Here, active laser thermography, a nondestructive testing method, is adapted to PBF-LB/M, using the defocused process laser as heat source. The testing can be performed layer by layer throughout the manufacturing process. We study our proposed testing method along experiments carried out on a custom research PBF-LB/M machine using infrared (IR) cameras. Our work enables a shift from post-process testing of components towards in-situ testing during the AM process. The actual component quality is evaluated in the process chamber and defects can be detected between layers.Item Machine Learning for Defect Detection for PBFAM Using High Resolution Layerwise Imaging Coupled with Post-Build CT Scans(University of Texas at Austin, 2017) Petrich, Jan; Gobert, Christian; Phoha, Shashi; Nassar, Abdalla R.; Reutzel, Edward W.This paper develops a methodology based on machine learning to detect defects during Powder Bed Fusion Additive Manufacturing (PBFAM) processes using data from high resolution images. The methodology is validated experimentally using both a support vector machine (SVM) and a neural network (NN) for binary classification. High resolution images are collected each layer of the build, and the ground truth labels necessary for supervised machine learning are obtained from a 3D computed tomography (CT) scan. CT data is processed using image processing tools—extended to 3D—in order to extract xyz position of voids within the component. Anomaly locations are subsequently transferred from the CT domain into the image domain using an affine transformation. Multi-dimensional features are extracted from the images using data surrounding both anomaly and nominal locations. Using cross-validation strategies for machine learning and testing, accuracies of close to 90% could be achieved when using a neural network for in-situ anomaly detection.Item An Online Surface Defects Detection System for AWAM Based on Deep Learning(University of Texas at Austin, 2017) Tang, Shangyong; Wang, Guilan; Zhang, Haiou; Wang, RuiDefects detecting layer by layer in arc welding based additive manufacturing (AWAM) is a big challenge as it affects the successive layer quality of the products. Most of the work on layer quality defection were focused on 3D profile measurement and X-ray spectroscopy method but it is inefficient, expensive with poor adaptability. In this work, an online intelligent surface defects detection system for AWAM was developed through deep learning algorithm and support vector machine method. To achieve a reliable surface feature of the welding beads, a vision sensor was used to get the image of the shaped surface synchronously. The system was trained offline and online to acquire knowledge of the welding beads which were classified into five patterns as normal, pore, hump, depression and undercut. An defection test result showed 95.29% accuracy. The system was verified to be practical with high accuracy and efficiency for the surface defects.Item A Passive On-Line Defect Detection Method for Wire and Arc Additive Manufacturing Based on Infrared Thermography(University of Texas at Austin, 2019) Chen, Xi; Zhang, Haiou; Hu, Jiannan; Xiao, YuAccording to the additive manufacturing process, this paper comes up with a passive infrared thermography non-destructive testing method based on the stack temperature field and pixel width curve. The temperature field of the arc-melting layer is collected in real-time, and the multi-frame temperature data stream is stacked for maximum value, and the region where the maximum value is greater than 800 °C is intercepted to obtain the current molten layer profile. The AlexNet model is used to classify the profile of molten layers, such as normal, deviation, flow and hump. Determine whether the current layer has a shape defect based on the model and the pixel width curve and the processing such as milling and repair welding will be taken in time. This method detects in real-time during the manufacturing process which will cause irreversible losses, and the current layer detection information is also the basis for adjusting the processing parameters of the next layer and realizes the closed-loop feedback of the additive manufacturing process.Item Sensing Defects during Directed-Energy Additive Manufacturing of Metal Parts using Optical Emissions Spectroscopy(University of Texas at Austin, 2014) Nassar, A.R.; Spurgeon, T.J.; Reutzel, E.W.Critical components produced via additive manufacturing must be free of unwanted defects. While defects may be detectable after deposition using nondestructive testing techniques, detecting defects during the deposition process offers many benefits: it may enable users to interrupt deposition to repair the part, or to abort deposition to minimize further loss of time and material. Here, we present a method for real-time defect detection during directed-energy additive manufacturing of metals. The method utilized optical emission spectroscopy and a custom-built data acquisition and control infrastructure. It was implemented on a LENS MR-7 machine, and employed during manufacturing of Ti-6Al-4V components in which defects were intentionally introduced. Emission spectra were correlated with defect locations, determined via computed tomography and metallographic cross-sectioning. Preliminary results indicated that defect formation was correlated with atomic titanium (Ti I) and Vanadium (V I) emissions and that measurement of the line-to-continuum ratio for line emissions could be used for defect detection. Based on these findings, sensing strategies for defect detection and, potentially, in-situ-defect repair may be realizable.Item Towards Defect Detection in Metal SLM Parts Using Modal Analysis "Fingerprinting"(University of Texas at Austin, 2017) Urban, James; Capps, Nick; West, Brian; Hartwig, Troy; Brown, Ben; Landers, Robert; Bristow, Douglas; Kinzel, EdwardThe validation of Additively Manufactured (AM) materials is a difficult and expensive process because the local engineering properties are a function of the thermal history. The thermal history varies with the process parameters, as well as the part geometry. This paper presents a case study using modal testing to identify defects in realistic AM parts. A setup consisting of a Scanning Laser Doppler Vibrometer (LDV) was used to identify the resonant frequencies for several geometrically identical parts on a build plate. Parts with suboptimal process parameters from purposely varying the process parameters, are identified by a shift in the mode peak frequency. Results from this study are compared to Finite Element Analysis (FEM) models and generalized for identifying defects in parts created with AM on the basis vibration/modal “fingerprinting.”