Article
Chemistry, Analytical
Tobias Mitterer, Christian Lederer, Hubert Zangl
Summary: In robotics, it is important to properly and securely identify new sensor or actuator modules added to a robot environment. This study presents a workflow that uses electronic datasheets and near field communication (NFC) to automate the identification process and establish trust in the new modules. By exchanging security information through NFC, the device can be easily identified and additional security information stored in the datasheet can be used to establish trust. The workflow was tested with prototype tactile sensors mounted onto a robotic gripper.
Article
Biochemical Research Methods
Xia-an Bi, Wenyan Zhou, Sheng Luo, Yuhua Mao, Xi Hu, Bin Zeng, Luyun Xu
Summary: This study proposes a novel deep learning method (FAGCN) to efficiently capture the development pattern of Alzheimer's disease (AD). By establishing a brain region-gene network and using a feature aggregation graph convolutional network, this method performs well in identifying AD-related brain regions and genes.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Chandrashekhar Goswami, V. K. Senthil Ragavan, Janjhyam Venkata Naga Ramesh, J. Balajee, A. Ronald Doni, T. R. Saravanan, S. Siva Shankar
Summary: Wireless sensor networks (WSNs) are used for reliable remote monitoring and tracking of objects, but there are privacy concerns when transmitting sensitive health data. Encrypted personal health records (PHRs) have been proposed, but challenges remain in preventing data leakage during deep learning and training model leaks. This study introduces a method of using homomorphic encryption to secure health data and a deep convolutional neural network (DCNN)-based algorithm to predict disease (COVID-19) and issue timely alerts. An assured data deletion method is also deployed to protect privacy, and the proposed strategy is supported by comprehensive and empirical studies.
Article
Biochemical Research Methods
Yuning Yang, Zilong Hou, Yansong Wang, Hongli Ma, Pingping Sun, Zhiqiang Ma, Ka-Chun Wong, Xiangtao Li
Summary: This study presents a novel end-to-end framework called HCRNet for identifying circRNA-RBP binding events. By fusing multiple sources of biological information to represent circRNAs and using a deep temporal convolutional network, HCRNet performs well across various circRNA datasets and exhibits good interpretability.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Engineering, Electrical & Electronic
Tian Meng, Yang Tao, Ziqi Chen, Jorge R. Salas Avila, Qiaoye Ran, Yuchun Shao, Ruochen Huang, Yuedong Xie, Qian Zhao, Zhijie Zhang, Hujun Yin, Anthony J. Peyton, Wuliang Yin
Summary: The article addresses the challenges of automatic depth evaluation of metal surface defects using deep learning techniques. It introduces a highly integrated portable ECT device, constructs a dataset named MDDECT, and trains various state-of-the-art 1-D residual convolutional neural networks. The results show high accuracy in discriminating surface defects in a stainless steel sheet with depths from 0.3 to 2.0 mm.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Automation & Control Systems
Maximilian Lorenz, Robert J. Martin, Thomas Bruecklmayr, Christian Donhauser, Bernd R. Pinzer
Summary: Neural networks prove to be capable of inline segmentation for measuring the burnish surface of punching parts, providing a more efficient and accurate method compared to traditional approaches.
INTERNATIONAL JOURNAL OF ADVANCED MANUFACTURING TECHNOLOGY
(2023)
Article
Computer Science, Information Systems
Muhammed Zahid Ozturk, Chenshu Wu, Beibei Wang, K. J. Ray Liu
Summary: This article presents GaitCube, a high-accuracy gait recognition system using a single commodity millimeter-wave radio. By introducing a novel feature representation called gait data cube and utilizing a pipeline of signal processing, GaitCube can automatically detect and segment human walking and effectively extract gait data. Experimental results show that GaitCube achieves high accuracy in different locations and times, enabling practical and ubiquitous gait-based identification.
IEEE INTERNET OF THINGS JOURNAL
(2022)
Article
Computer Science, Information Systems
Juxiang Zhou, Jianhou Gan, Wei Gao, Antoni Liang
Summary: A new robust global descriptor for image retrieval is proposed in this paper by effectively aggregating local deep convolutional features, aiming to improve the ability of deep features to be described and discerned. Experimental results show that the method can achieve retrieval results comparable to popular approaches for deep feature aggregation.
INFORMATION SCIENCES
(2021)
Article
Materials Science, Characterization & Testing
Jakub Roemer, Hassan Khawaja, Mojtaba Moatamedi, Lukasz Pieczonka
Summary: This paper proposes a new data processing scheme for laser spot thermography (LST) in nondestructive testing (NDT) of composite laminates. The scheme utilizes parameterization and machine learning to overcome difficulties in LST signal processing and provide valuable diagnostic information. The effectiveness of the proposed approach is demonstrated on an experimental dataset of a laminated composite sample with simulated delaminations. The paper discusses the theoretical aspects of the signal processing and inference algorithms, as well as the necessary experimental arrangements.
JOURNAL OF NONDESTRUCTIVE EVALUATION
(2023)
Article
Geochemistry & Geophysics
Xin Zhang, Yongcheng Wang, Ning Zhang, Dongdong Xu, Huiyuan Luo, Bo Chen, Guangli Ben
Summary: The development of deep learning has improved the classification ability of hyperspectral images, but problems such as limited labeled samples and imbalanced categories still exist. To address these issues, a spectral-spatial fractal residual convolutional neural network with data balance augmentation is proposed, which effectively learns spectral-spatial information and ensures the integrity of the data.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Engineering, Multidisciplinary
Jie Chen, Yi Gao, Yongming Liu
Summary: This study presents a novel framework called MDA-CNN for multi-fidelity modeling. The framework utilizes convolutional neural networks to process multi-fidelity data, enabling high accuracy estimation and low computational cost. The method has advantages in handling high-dimensional data and nonlinear mapping, and can handle data from different scales and sources.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Engineering, Multidisciplinary
Yuliang Wu, Xuelei Fu, Jiapu Li, Pengyu Zhang, Honghai Wang, Zhengying Li
Summary: Nondestructive testing (NDT) is crucial for ensuring equipment safety. Ultrasonic NDT, known for its high sensitivity, fast speed, and accurate defect location, has gained significant attention. Photoacoustic NDT, a promising field within ultrasonic NDT, is especially attractive due to its immunity to electromagnetic interference. However, current photoacoustic NDT systems face challenges such as insufficient excitation intensity and complex ultrasonic signal characteristics, which hinder large-area NDT and accurate crack visualization. In this study, an all-fiber photoacoustic system for large-area NDT is presented. The system addresses these challenges by developing a photoacoustic generator unit to generate stronger ultrasonic signals and using mode decomposition to simplify detected ultrasonic signals. The technology allows for large-area crack monitoring with improved resolution, paving the way for high-resolution equipment crack monitoring with enhanced accuracy in various environments.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2023)
Article
Engineering, Electrical & Electronic
Frederik Elischberger, Joachim Bamberg, Xiaoyi Jiang
Summary: Ultrasonic testing (UT) has been widely used in the industry to detect internal defects in bulk material. This study focuses on the inspection of IN718 superalloy, commonly used for turbine components. A new type of defect, called discrete Clean White Spot Segregation, poses challenges to conventional UT due to its different material characteristics. This study is the first to use deep learning techniques in combination with conventional UT for the detection of this defect.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Chemistry, Physical
Pawel Karol Frankowski, Tomasz Chady
Summary: This paper presents the method of multisensory spatial analysis (MSA) for quick and simultaneous identification of concrete cover thickness, rebar diameter, and alloys of reinforcement in large areas of reinforced concrete structures. The method divides the complex problem into three simple tasks based on separate premises, and achieves high accuracy in real-time area testing.
Article
Remote Sensing
Panle Li, Xiaohui He, Mengjia Qiao, Disheng Miao, Xijie Cheng, Dingjun Song, Mingyang Chen, Jiamian Li, Tao Zhou, Xiaoyu Guo, Xinyu Yan, Zhihui Tian
Summary: In this study, a multi-map integration model (MMIM) is developed using multiple crowdsourced data, including OpenStreetMap (OSM), Zmap, and GPS, to improve the performance of DCNNs in road extraction tasks. The proposed method shows great performance in experiments, with smoother and more complete road extraction results.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2021)
Article
Engineering, Electrical & Electronic
Grzegorz Psuj
INTERNATIONAL JOURNAL OF APPLIED ELECTROMAGNETICS AND MECHANICS
(2018)
Article
Chemistry, Analytical
Grzegorz Psuj
Article
Chemistry, Analytical
Cesar Camerini, Joao Marcos Alcoforado Rebello, Lucas Braga, Rafael Santos, Tomasz Chady, Grzegorz Psuj, Gabriela Pereira
Article
Chemistry, Analytical
Michal Maciusowicz, Grzegorz Psuj
Article
Chemistry, Analytical
Michal Maciusowicz, Grzegorz Psuj
Article
Chemistry, Physical
Michal Maciusowicz, Grzegorz Psuj
Article
Chemistry, Physical
Jakub Kowalczyk, Przemyslaw Lopato, Grzegorz Psuj, Dariusz Ulbrich
Article
Chemistry, Analytical
Grzegorz Psuj, Przemyslaw Lopato, Michal Maciusowicz, Michal Herbko
Summary: This paper introduces a system based on the ferromagnetic resonance method for monitoring changes in the magnetization dynamics of steel elements subjected to deformations. The system operation was verified using a set of samples made of low carbon steel film, showing successful differentiation of changes in magnetic conditions caused by the straining process.
Article
Chemistry, Physical
Barbara Szymanik, Grzegorz Psuj, Maryam Hashemi, Przemyslaw Lopato
Summary: This paper proposes a new method based on active infrared thermography for assessing the state of 3D-printed structures. The technique involves using an external energy source to create temperature differences between undamaged and defective areas, which may be hard to measure in materials with low thermal conductivity. A dedicated algorithm and deep convolutional neural network are used for signal analysis to enhance contrast between background and defect areas. Experimental results demonstrate the effectiveness of this hybrid signal analysis method in visualizing inner structure and determining defect parameters such as depth and diameter.
Article
Chemistry, Multidisciplinary
Michal Maciusowicz, Grzegorz Psuj
Summary: Magnetic Barkhausen Noise (MBN) is a method being considered by research and development centers for its ability to provide knowledge about the properties of examined materials. The evaluation of magnetic anisotropy is a key application area, and various time-frequency transformations are used for analyzing MBN data.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Physical
Michal Maciusowicz, Grzegorz Psuj, Pawel Kochmanski
Summary: This paper presents a new approach that combines the Short-Time Fourier Transform and deep convolutional neural networks (DCNN) to analyze magnetic Barkhausen noise (MBN) and evaluate electrical steels. The proposed methodology allows for the extraction of hidden information and building of knowledge, enabling a more effective quantification of the characteristics of magnetic materials.
Article
Engineering, Electrical & Electronic
Michal Herbko, Przemyslaw Lopato, Grzegorz Psuj, Prabhu Rajagopal
Summary: This article investigates the use of various fractal patch geometries in microstrip strain sensors. By comparing the performance of fractal geometries with rectangular patches, it is found that the best solution to this problem is to use a combination of the Koch curve and the Sierpinski carpet, which can reduce the patch size by 70% while still maintaining a high sensitivity.
IEEE SENSORS JOURNAL
(2022)
Article
Chemistry, Multidisciplinary
Michal Maciusowicz, Grzegorz Psuj
Summary: In this paper, a combination of Magnetic Barkhausen Noise (MBN) and classical machine learning (ML) methods were used to evaluate the grade and magnetic directions of electrical sheets. The results showed that deep learning models outperformed classical ML methods in terms of accuracy.
APPLIED SCIENCES-BASEL
(2022)
Article
Chemistry, Physical
Dariusz Ulbrich, Grzegorz Psuj, Artur Wypych, Dariusz Bartkowski, Aneta Bartkowska, Arkadiusz Stachowiak, Jakub Kowalczyk
Summary: This study presents a novel method for assessing spot welded joints using ultrasound technology. By measuring the amplitude of ultrasonic pulses, the quality of the welds can be determined. The size of the weld nugget affects the transfer of wave energy, resulting in different amplitudes of ultrasonic pulses.
Editorial Material
Chemistry, Multidisciplinary
Grzegorz Psuj, Barbara Szymanik
APPLIED SCIENCES-BASEL
(2023)