Commercial aircraft engines have a maintenance process that includes overhauling approximately every six years. Hundreds of different components must be disassembled, checked, repaired (if necessary), and then reassem...
详细信息
ISBN:
(纸本)9781728103563
Commercial aircraft engines have a maintenance process that includes overhauling approximately every six years. Hundreds of different components must be disassembled, checked, repaired (if necessary), and then reassembled. this includes undoing fasteners, cleaning, checking, refitting, and tightening them. Prior to refitting the fasteners, they must be checked for damages. In this paper, we propose an automatic damage inspection of the fasteners, using computer vision and machine learning. We built a setup to automatically record and preprocess the data and compared multiple supervised and unsupervised machine learning models for detecting damages of 12 different fasteners. Using our automatic approach, we can determine the type of fastener, its status (damaged or intact) and visualize the anomalies to aid the understanding of the decisions of the automatic detection. this can be the first step towards a fully automated fastener damage detection in overhaul processes.
Representation learning of knowledge graphs has gained wide attention in the field of natural language processing. Most existing knowledge representation models for knowledge graphs embed triples into a continuous low...
详细信息
ISBN:
(纸本)9783030336073;9783030336066
Representation learning of knowledge graphs has gained wide attention in the field of natural language processing. Most existing knowledge representation models for knowledge graphs embed triples into a continuous low-dimensional vector space through a simple linear transformation. In spite of high computation efficiency, the fitting ability of these models is suboptimal. In this paper, we propose a multi-scale capsule network to model relations between embedding vectors from a deep perspective. We use convolution kernels with different sizes of windows in the convolutional layer inside a Capsule network to extract semantic features of entities and relations in triples. these semantic features are then represented as a continuous vector through a routing process algorithm in the capsule layer. the modulus of this vector is used as the score of confidence of correctness of a triple. Experiments show that the proposed model obtains better performance than state-of-the-art embedding models for the task of knowledge graph completion over two benchmarks, WN18RR and FB15k-237.
We propose a deep learning-based method that simultaneously determines a target object to be picked up by an autonomous manipulator and the velocity of an automated guided vehicle (AGV) that passes in front of the man...
详细信息
ISBN:
(纸本)9781728103563
We propose a deep learning-based method that simultaneously determines a target object to be picked up by an autonomous manipulator and the velocity of an automated guided vehicle (AGV) that passes in front of the manipulator while the AGV carries a carton case containing the target and other objects. Our method can efficiently perform automated piece-picking operations in warehouses without the AGV needing to pause in front of the manipulator. In our method, for preparing supervised data sets with color images of objects that are randomly piled up in the carton case, a simulator checks whether each object is "pickable" or not by trying to plan the manipulator's motion to have its hand reach the object while avoiding surrounding obstacles by using the depth images in consideration of the carton case's movement and velocity. then, we make each of multiple deep convolutional neural networks (DCNNs) corresponding to multiple levels of velocity learn to detect grasp points for only pickable objects from an RGB image. In our experimental test, using our method, a prototype of the system successfully picked ordered objects up without the AGV pausing while the AGV changed its velocity depending on the layout of the objects in the carton case.
In the increasingly automated world of biotech manufacturing, the inspection of lyophilized drug products remains a cumbersome and manual process, relying on human operators to inspect finished vials for glass defects...
详细信息
ISBN:
(纸本)9781728103563
In the increasingly automated world of biotech manufacturing, the inspection of lyophilized drug products remains a cumbersome and manual process, relying on human operators to inspect finished vials for glass defects, contaminants, and other possible issues. Automating this procedure is challenging due to "fogging" on glass vials and low occurrence of defective samples for training decision models. In this work, we apply deep neural networks to automated classification of lyophlized product vials via computer vision. the proposed approach utilizes shared convolutional layers to account for multiple images acquired of the same vial at various rotations, and transfer learning is examined as a tool to partially overcome the lack of defective data in industrial manufacturing applications. the method is tested using a real-world industrial product as a representative case study. We find that 85-90% of defects can be detected from a single camera angle, demonstrating the potential of multi-input neural networks in lyophilized drug product inspection. Moreover, the results suggest that transfer learning enhances the generalization ability of learned models, even when using data designed for a very different task.
Additive manufacturing (AM) assisted by a digital twin is expected to revolutionize the realization of high-value and high-complexity functional parts on a global scale. Machine learning (ML) introduced in digitized A...
详细信息
ISBN:
(纸本)9781728103563
Additive manufacturing (AM) assisted by a digital twin is expected to revolutionize the realization of high-value and high-complexity functional parts on a global scale. Machine learning (ML) introduced in digitized AM provides potential to transform AM data into knowledge continuously and automatically;hence AM products will be designed and manufactured with improved quality. Conventional product development systems, however, fail to fully adopt ML algorithms on the increasingly available AM data for knowledge acquisition. To address the limitation, this paper proposes an algorithmic framework for constructing AM knowledge automatically and continuously from data. the proposed algorithm develops predictive models to correlate process parameters with part structure and properties using ML algorithms on design, process control, in-situ monitoring, and ex-situ measurement AM data. Based on the predictive models, the algorithm constructs prescriptive rules necessary for decision-making in AM product developments. the constructed rule knowledge is stored in a knowledge base by the algorithm for further AM knowledge query and constructions. then, the algorithm provides feedback to a knowledge-query formulation phase, which forms the algorithm into a closed loop. through the algorithm, AM knowledge can self-evolve continuously, while reflecting dynamically generated AM data in an automated and autonomous manner. A case study is presented to illustrate the proposed algorithm and a software prototype is developed to demonstrate the algorithm capability.
this paper demonstrates a place recognition and localization method designed for automated guidance of mobile robots. Collecting and annotating sufficient images for a supervised deep learning model is often an exhaus...
详细信息
ISBN:
(纸本)9781728103563
this paper demonstrates a place recognition and localization method designed for automated guidance of mobile robots. Collecting and annotating sufficient images for a supervised deep learning model is often an exhausting work. Devising an effective visual detection scheme for a mobile robot location detection job in a feature-barren environment such as the indoor corridors of buildings is also quite challenging. To address these issues, a supervised deep learning model for the spatial coordinate detection of a mobile robot is proposed here. Specifically, a novel technique is introduced involving structuring and collaging of the surrounding views obtained by the on-board cameras for the training data preparation. A system linking robot kinematics and image processing provides automatic data annotation, which significantly reduces the need for human work on data preparation. Experimental evidence showed that the precision and recall rates of the location coordinate detection are 0.91 and 0.85, respectively. Also, the detection appeared to be effective over a path width of 0.75 m, which is sufficient to cover the possible deviations from the target path. Furthermore, it took averagely 0.14 s for each visual detection performed by an ordinary PC on-board the mobile robot;thus, real-time navigation using the proposed method is achievable.
the aim of this study is to develop and validate a machine learning (ML) model for predicting survival after liver transplantation based on pre-transplant donor and recipient characteristics. For this purpose, we cons...
详细信息
ISBN:
(纸本)9783030336172;9783030336165
the aim of this study is to develop and validate a machine learning (ML) model for predicting survival after liver transplantation based on pre-transplant donor and recipient characteristics. For this purpose, we consider a database from the United Network for Organ Sharing (UNOS), containing 29 variables and 39,095 donor-recipient pairs, describing liver transplantations performed in the United States of America from November 2004 until June 2015. the dataset contains more than a 74% of censoring, being a challenging and difficult problem. Several methods including proportional-hazards regression models and ML methods such as Gradient Boosting were applied, using 10 donor characteristics, 15 recipient characteristics and 4 shared variables associated withthe donor-recipient pair. In order to measure the performance of the seven state-of-the-art methodologies, three different evaluation metrics are used, being the concordance index (ipcw) the most suitable for this problem. the results achieved show that, for each measure, a different technique obtains the highest value, performing almost the same, but, if we focus on ipcw, Gradient Boosting outperforms the rest of the methods.
In computer vision, when labeled images of the target domain are highly insufficient, it is challenging to build an accurate classifier. Domain adaptation stands for an effective solution to address it by utilizing av...
详细信息
Heart sounds are essential components of cardiac diagnosis, in which heart conditions can be detected using phonocardiogram (PCG) signals. PCG signals provide useful information and can help in the early detection and...
详细信息
In the recent past Deep learning (DL) has been used to develop intelligent systems that perform surprisingly well in a large variety of tasks, e.g. image recognition, machine translation, and self-driving cars. the hu...
详细信息
暂无评论