Feature Transfer Research Articles

Residual deep subdomain adaptation network: A new method for intelligent fault diagnosis of bearings across multiple domains

The existing transfer learning-based fault diagnosis methods basically make use of fault knowledge learnt from one source domain to another target domain. Due to variable operating conditions, transfer tasks in the fault diagnosis need be inevitably performed many times, which limit their industrial applicability. To address the problem, a new residual deep subdomain adaptation network is proposed for intelligent fault diagnosis of bearings across multiple domains. Its remarkable advantage is that only one transfer task need be executed no matter how the operating conditions change. In this method, a residual network is constructed to extract transferable features from source domain and target domains. And, local maximum mean discrepancy is introduced to accurately align the distribution of the related subdomains within the same category in the source domain and the target domains. Comprehensive experimental results confirm that the proposed method can make use of fault knowledge learnt from the single source domain for fault diagnosis in the multiple target domains. The classification accuracy has a significant improvement as compared with the existing popular methods.

Boosting Photothermal Theranostics via TICT and Molecular Motions for Photohyperthermia Therapy of Muscle‐Invasive Bladder Cancer (Adv. Healthcare Mater. 24/2021)

Photothermal Therapies In article number 2101063 by Qian Liu, Guangxue Feng, Dan Ding, and co-workers, a small photothermal molecule with active intramolecular motions and a dark state twisted intramolecular charge transfer feature is designed to boost the photothermal conversion ability. Such a design strategy enables the effective low-temperature photo-hyperthermia treatment of muscle-invasive bladder cancers at low drug and laser dosages with minimal side effects.

A deep transfer learning method for system-level fault diagnosis of nuclear power plants under different power levels

The system-level fault diagnosis technology of nuclear power plants (NPPs) can provide operating status information, and further assist operators in making decisions to support the safe operation of NPPs. In traditional intelligent fault diagnosis research, the required assumption that the training set and test set follow the same distribution limits the application of the model. The fault diagnosis methodology integrating deep learning and transfer learning is proposed in this paper to solve the problem of poor generalization performance caused by inconsistent probability distribution under different power levels. First, the monitoring parameter preprocessing is carried out to suit the input of deep convolutional neural network (DCNN), which can effectively extract the transferable features between source and target power levels. More importantly, a promising transfer learning approach is introduced to sufficiently reduce the feature distribution discrepancy between different power levels, and eventually realize the fault diagnosis of the target power level. Finally, the simulation data test verifies the effectiveness of the transfer learning method for fault diagnosis under different power levels. Based on the feature visualization technology, the data discrepancy reduction effect is intuitively present, consequently demonstrating the potential application value in improving the generalization ability of NPP fault diagnosis.

High Performance Near-Infrared Emitters with Methylated Triphenylamine and Thiadiazolo[3,4-g]quinoxaline-Based Fluorophores.

Three near-infrared emitters (2TPA-QBT, 2MeTPA-BT and TPA-QBT-MeTPA) were rationally designed and synthesized. Density functional theory (DFT) and time-dependent density functional theory (TDDFT) calculations showed that the introduction of mono- or di-methyl groups between the donors and acceptor could result in the spatial configuration changing greatly for 2MeTPA-QBT and TPA-QBT-MeTPA compared to their parent compound 2TPA-QBT. The emission of TPA-QBT-MeTPA had a more obvious hybridized local and charge transfer feature (HLCT) based on the influence of the steric hindrance of the methyl substituent. Attributed to their different spatial configurations and luminescence mechanisms, different emission wavelengths with photoluminescent quantum yields of 26%, 38% and 34% in toluene, as well as 24%, 27% and 31% in 4,4′-bis(N-carbazolyl)-1,1′-biphenyl (CBP) doped film, were observed for 2TPA-QBT, 2MeTPA-QBT and TPA-QBT-MeTPA, respectively. The constructed organic light-emitting devices (OLEDs) displayed electroluminescence with emission peaks at 728, 693 and 710 nm, with maximum external quantum efficiencies of 1.58%, 1.33% and 3.02% for the 2TPA-QBT, 2MeTPA-QBT and TPA-QBT-MeTPA-doped OLEDs, respectively. This work illustrated the effect of spatial configuration changes on the luminescence properties of donor-acceptor-type organic emitters.

SAR Target Detection Based on Domain Adaptive Faster R-CNN with Small Training Data Size

It is expensive and time-consuming to obtain a large number of labeled synthetic aperture radar (SAR) images. In the task of small training data size, the results of target detection on SAR images using deep network approaches are usually not ideal. In this study, considering that optical remote sensing images are much easier to be labeled than SAR images, we assume to have a large number of labeled optical remote sensing images and a small number of labeled SAR images with the similar scenes, propose to transfer knowledge from optical remote sensing images to SAR images, and develop a domain adaptive Faster R-CNN for SAR target detection with small training data size. In the proposed method, in order to make full use of the label information and realize more accurate domain adaptation knowledge transfer, an instance level domain adaptation constraint is used rather than feature level domain adaptation constraint. Specifically, generative adversarial network (GAN) constraint is applied as the domain adaptation constraint in the adaptation module after the proposals of Faster R-CNN to achieve instance level domain adaptation and learn the transferable features. The experimental results on the measured SAR image dataset show that the proposed method has higher detection accuracy in the task of SAR target detection with small training data size than the traditional Faster R-CNN.

Quasi-one-dimensional thermal transport in trigonal selenium crystal

The lattice thermal conductivity in van der Waals crystal selenium is investigated by solving the phonon Boltzmann transport equation combined with the first-principles calculations. The lattice thermal conductivity of selenium along the perpendicular to the chain direction is extremely low, which is 5 times lower than the conductivity in the parallel direction, showing a clear quasi-one-dimensional heat transfer feature. The small phonon group velocity, acoustic-optical phonon branches mixing and bonding anharmonicity along the perpendicular to the chain direction contribute to the high anisotropy in thermal conductivity. Moreover, particles with several nanometers can introduce boundary scattering to a large portion of phonons modes in selenium, where the lattice thermal conductivity is greatly reduced. Both the unique one-dimensional thermal transport property and low thermal conductivity make selenium to be a potential thermal management in thermoelectric and other electronic technologies.

Cross-domain fault diagnosis of rotating machinery in nuclear power plant based on improved domain adaptation method

ABSTRACT Bearings are widely applied in rotating machinery of nuclear power plants (NPPs). Data-driven fault diagnosis technology is critical to ensuring the reliable operation of rotating machinery. Aiming at the problem of poor model generalization ability caused by discrepant data distribution of monitoring signals under various working conditions, a deep transfer learning method based on fully categorized alignment subdomain adaptation (FCA-SAN) is proposed in this paper. Firstly, the bearing vibration signals of the source and target operating conditions are preprocessed and converted into time-frequency domain images suitable for model input. Subsequently, a pre-trained deep convolutional neural network (DCNN) model is adopted as the feature extractor, which is combined with FCA-SAN to extract transferable features across different working conditions. The subdomain adaptation method reduces the data distribution discrepancy more fine-grained by aligning the feature distribution of different working conditions, thereby effectively improving the model generalization ability. Finally, the experimental results show that, compared with the traditional method, the proposed subdomain adaptation method reaches the highest fault diagnosis accuracy in different transfer tasks, which demonstrates the potential application value in rotating machinery of NPPs.

Deep Discriminative Domain Adaptation

Domain adaptation studies the problem of how to transfer knowledge across different domains where the source domain with rich labeled source samples and target domain with less or even no labeled target samples are drawn from different probability distribution. A prevailing strategy is to generate transferable features cross-domain, which induces deficient knowledge transfer in learning process, especially close to the input end. Moreover, when target samples distributed far their corresponding class centers, or near the edge of the clusters, only using learned source samples features to predict target samples class which may easily brings about misclassification. Targeting to deal with these issues, we propose Deep Discriminative Domain Adaptation (DDDA) method, which jointly minimizes the supervised classification loss of annotated source examples, the unsupervised center alignment and correlation alignment losses measured on both convolutional layers and fully connected layers with help of attention mechanism. The multi-layer transfer mechanism complementary strengthens each individual transfer component, and markedly improves the generalization ability of transfer models. A series of experiments conducted on several standard datasets validate that the proposed method consistently outperforms contemporary adaptation approaches.

Heat transportation enrichment and elliptic cylindrical solution of time-dependent flow

The main focus of this article is to examine the effects of heat transfer for a compressible time-dependent laminar flow pass the two distinctly positioned elliptic cylinders. The Mac number is chosen below 0.3 to keep the flow laminar. The heat transfer feature has been added and coupled with the laminar flow. The heat transfer feature adjusted for constant pressure. The arising Naiver-Stokes equations have been addressed numerically. The mesh has also been created and its entities have been elaborated statistically. The outcomes of velocity distribution, pressure distribution, 2D temperature plots, isothermal contours, drag coefficient, streamlines, and surface volume of fluid are discussed. The BDF technique has been employed to tackle the problem numerically. It was observed that the velocity profile at the boundaries of the elliptic cylinder has a maximum value, 3.85 m/s. The pressure distribution is observed maximum around elliptic cylinders. The heat transfer coefficient has maximum values at the upper and lower boundaries, the maximum temperature value observed is 290K. The isothermal contours, streamlines, and velocity volume were also studied. The drag coefficient is observed increasing but the drag force is decreasing. The mathematical modeling of the current problem has been designed in COMSOL.

Multiscale domain adaption models and their application in fault transfer diagnosis of planetary gearboxes

In recent years, various deep domain adaption (DDA) models, such as deep domain confusion (DDC) and deep adaption network (DAN), are proposed. These models can adapt a trained model in the source domain to new classification tasks in the target domain. However, these classical DDA models suffer from some inherent drawbacks. For example, traditional DDA models can output only one transfer feature (TF) with high dimension and fixed scales, thus possibly losing important information while performing domain confusion operation. Multiscale domain adaption (MSDA) is proposed in this paper to remold and improve the classical DDA models to solve the aforementioned problems. MSDA is a universally applicable strategy that can be embedded into most existing classical DDA models. An MSDA block is designed and constructed on the basis of multiscale convolution networks to replace the last convolutional layer of the original DDA model. The MSDA block has four parallel pipelines consisting of multiscale convolutional and global average pooling operations. Therefore, MSDA can extract more domain-invariant features than the original feature extractor, meanwhile the four low-dimension TFs can simplify the calculation of domain confusion losses. The four TFs are concatenated into a feature vector, and then it is input into the top classifier for fault identification. MSDA can be effectively applied to five classical DDA models and enhance their abilities of domain adaption. The effectiveness and advantage of the proposed MSDA are verified through 18 fault transfer diagnosis tasks of planetary gearboxes.

Transtrack: Online meta-transfer learning and Otsu segmentation enabled wireless gesture tracking

Individual diversity poses a cross-user performance variance challenge that stumbles the practicality, especially for the wireless gesture tracking systems. Since the difficulty of annotating low-semantic wireless data limits constructing a big dataset, the recognizer should quickly adjust to different individuals via small datasets. To this end, we present TransTrack, an accurate wireless indoor gesture tracking system that can adjust to different users quickly. The key insight is that each unlabeled gesture contains learnable individual features that can help the gesture tracking model learning how to adapt to different users. Specifically, TransTrack uses recursive Otsu segmentation to separate gesture-induced signals with the background noise inspired by image segmentation. It then augments training data to learn the transferable features by leveraging the redundant information. A datum-based alignment method is proposed to unlock the limitation of classifier selection without distortion. Finally, TransTrack proposes an online meta-transfer learning method that collects unlabeled data transparently to train the tracking model for different tasks. Extensive experiments show that TransTrack can quickly adapt to different users and conditions.

Numerical investigation on performance improvement of latent heat exchanger with sextant helical baffles

• Numerical investigation on a novel latent heat exchanger with sextant helical baffles was implemented. • Helix angle exhibits an opposite effect on the thermal performance and comprehensive performance of shell side. • The thermal conductivity and pr number of the HTF are important factors affecting the shell-side heat transfer feature. • Internal-external fin structure shows a large impact on the melting process of PCM. • Nano-additives have a remarkable influence on the thermal conductivity and latent heat of Nano-PCMs. Latent heat thermal energy storage is a promising technology, which is becoming more and more widespread in industrial fields. A novel latent heat exchanger with sextant helical baffles is thus investigated numerically in this paper. The different helix angles (10°, 20°, 30°, 40°) of sextant helical baffle are employed to explore the shell-side thermal and comprehensive performances in latent heat exchanger. The geometric parameters of internal-external fins, including the height, thickness and shape, are studied to improve the melting rate of phase change material (PCM). Water, fuel oil JP-4, ethylene glycol and hydraulic oil are chosen as heat transfer fluids (HTFs). Nano-PCMs with various volume concentrations of TiO 2 (2%, 4% and 6%) are used to study the discharging characteristic of latent heat exchanger. The results indicate that shell-side heat transfer and integrated behaviors need to be considered comprehensively for the choice of helix angle. The longitudinal-shaped fin with 7.5 mm height and 1 mm thickness is studied as the optimal scheme. Smaller Pr number and larger thermal conductivity are applied to HTF to further enhance the heat transfer between HTF and PCM. Furthermore, the selection of Nano-PCM needs to comprehensively consider the requirements of thermal conductivity and phase change latent heat in practical application.

A simple chalcone molecular rotor for specific fluorescence imaging of mitochondrial viscosity changes in living cells

Mitochondrial viscosity is related to numerous physiological processes such as solute diffusion, enzyme catalysis, protein folding, translocation, and conformation change. It is significant to evaluate and monitor the mitochondrial viscosity changes in living cells. In this work, we developed a simple chalcone molecular rotor Mito-HCA for specific fluorescence imaging of mitochondrial viscosity changes in living cells. The chalcone molecular rotor Mito-HCA consists of a donor-π-acceptor (D-π-A) chalcone skeleton with a typical twisted intramolecular charge transfer (TICT) feature and a triphenylphosphonium unit for mitochondrial targeting. The fluorescent intensity and fluorescence lifetime of the probe Mito-HCA reveal good sensitivity and high selectivity to solution viscosity. It can be conveniently used for the specific fluorescence imaging of mitochondria and mitochondrial viscosity detection benefit from the simple synthetic procedure, large Stokes shift, high pH stability and low cytotoxicity. The fluorescent intensity and fluorescence lifetime of probe Mito-HCA exhibits good sensitivity to mitochondrial viscosity changes in living cells under osmotic shock, starvation stress and ionophore (monensin) incubation.

WDA: An Improved Wasserstein Distance-Based Transfer Learning Fault Diagnosis Method.

With the growth of computing power, deep learning methods have recently been widely used in machine fault diagnosis. In order to realize highly efficient diagnosis accuracy, people need to know the detailed health condition of collected signals from equipment. However, in the actual situation, it is costly and time-consuming to close down machines and inspect components. This seriously impedes the practical application of data-driven diagnosis. In comparison, the full-labeled machine signals from test rigs or online datasets can be achieved easily, which is helpful for the diagnosis of real equipment. Thus, we introduced an improved Wasserstein distance-based transfer learning method (WDA), which learns transferable features between labeled and unlabeled signals from different forms of equipment. In WDA, Wasserstein distance with cosine similarity is applied to narrow the gap between signals collected from different machines. Meanwhile, we use the Kuhn–Munkres algorithm to calculate the Wasserstein distance. In order to further verify the proposed method, we developed a set of case studies, including two different mechanical parts, five transfer scenarios, and eight transfer learning fault diagnosis experiments. WDA reached an average accuracy of 93.72% in bearing fault diagnosis and 84.84% in ball screw fault diagnosis, which greatly surpasses state-of-the-art transfer learning fault diagnosis methods. In addition, comprehensive analysis and feature visualization are also presented.

Daya Tarik Dompet Digital

This research was conducted on the Ovo digital wallet which aims to find out what is the attraction of Ovo, so people use Ovo more than other digital wallets in payment transactions. This study uses a qualitative method, by using triangulation of data sources namely observation, interviews and documentation. Informants in this study are people who have knowledge in the topic under study. And data analysis used with analysis techniques in the form of data reduction, data presentation and drawing conclusions. The results of this study indicate that in the midst of a lot of competition at this time Ovo is intensively conducting sales promotions in the form of cashback and vouchers, so consumers feel they benefit more by using Ovo as their payment instrument compared to other digital wallets. Ovo is superior because it has an easy to fill balance feature without additional administrative costs that are not owned by other digital wallets. In addition, the attraction of Ovo is because in Ovo there is a transfer feature to the bank without additional administrative costs. This is a favorite feature of Ovo users themselves. At present hundreds of merchants have become Ovo's partners. Ovo offers convenience in its features so that it makes consumers feel easier in carrying out activities. More practical, fast and easy transactions, Ovo's presence provides benefits such as shortening their payment time, no need to carry or save large amounts of cash.

Novel, Stable Catholyte for Aqueous Organic Redox Flow Batteries: Symmetric Cell Study of Hydroquinones with High Accessible Capacity.

Owing to their broad range of redox potential, quinones/hydroquinones can be utilized for energy storage in redox flow batteries. In terms of stability, organic catholytes are more challenging than anolytes. The two-electron transfer feature adds value when building all-quinone flow battery systems. However, the dimerization of quinones/hydroquinones usually makes it difficult to achieve a full two-electron transfer in practical redox flow battery applications. In this work, we designed and synthesized four new hydroquinone derivatives bearing morpholinomethylene and/or methyl groups in different positions on the benzene ring to probe molecular stability upon battery cycling. The redox potential of the four molecules were investigated, followed by long-term stability tests using different supporting electrolytes and cell cycling methods in a symmetric flow cell. The derivative with two unoccupied ortho positions was found highly unstable, the cell of which exhibited a capacity decay rate of ~50% per day. Fully substituted hydroquinones turned out to be more stable. In particular, 2,6-dimethyl-3,5-bis(morpholinomethylene)benzene-1,4-diol (asym-O-5) displayed a capacity decay of only 0.45%/day with four-week potentiostatic cycling at 0.1 M in 1 M H3PO4. In addition, the three fully substituted hydroquinones displayed good accessible capacity of over 82%, much higher than those of conventional quinone derivatives.

Glucose-oxidase like catalytic mechanism of noble metal nanozymes

Au nanoparticles (NPs) have been found to be excellent glucose oxidase mimics, while the catalytic processes have rarely been studied. Here, we reveal that the process of glucose oxidation catalyzed by Au NPs is as the same as that of natural glucose oxidase, namely, a two-step reaction including the dehydrogenation of glucose and the subsequent reduction of O2 to H2O2 by two electrons. Pt, Pd, Ru, Rh, and Ir NPs can also catalyze the dehydrogenation of glucose, except that O2 is preferably reduced to H2O. By the electron transfer feature of noble metal NPs, we overcame the limitation that H2O2 must be produced in the traditional two-step glucose assay and realize the rapid colorimetric detections of glucose. Inspired by the electron transport pathway in the catalytic process of natural enzymes, noble metal NPs have also been found to mimic various enzymatic electron transfer reactions including cytochrome c, coenzymes as well as nitrobenzene reductions.

A multi-representation-based domain adaptation network for fault diagnosis

Deep learning-based domain adaptation algorithms with various representations have been recently developed to address the domain shift problem in mechanical fault diagnosis. However, few research have focused on potential improvements through multiple representations. Thus, a multi-representation-based domain adaptation network is proposed in this paper. Three complementary time–frequency representations are first proposed to serve as input-based multiple representations for the subsequent parallel models. Then, parallel models with improved inception modules are trained to obtain feature-based multiple domain-invariant representations. Finally, ensemble learning through majority voting is used to obtain the final results. Comprehensive experimental results on two test rigs reveal that the proposed algorithm outperforms state-of-the-art single-representation-based domain adaptation algorithms in terms of cross-domain fault diagnosis. Furthermore, visualization results demonstrate that the proposed algorithm extracts transferable features and takes advantage of ensemble learning to achieve high-precision diagnosis.

Tracklet style transfer and part-level feature description for person reidentification in a camera network

Tracklet style transfer and part-level feature description for person reidentification in a camera network

Classification of COVID-19 images based on transfer learning and feature fusion

ABSTRACT Rapid and accurate identification of COVID-19 images is particularly important in the context of the COVID-19 epidemic persisting. As COVID-19CT image acquisition data is limited, a novel pneumonia image classification method based on transfer learning and feature fusion of VGG16, DenseNet121 and ResNet50 is proposed in this paper. Firstly, the three basic network models are used to extract the features from multiple levels, respectively, and then the pairwise fusion models and the proposed three-backbone network feature convolution model (VGG16-DenseNet121-ResNet50) are utilized to fuse the extracted features. The fused features are extracted through a convolutional pooling module and finally, a layer of full connection and SoftMax are used for classification. Results show that the accuracy, sensitivity and specificity of the proposed model are 97.8%, 98.3% and 97.2%, respectively. Compared with the basic transfer learning model and pairwise fusion models, the proposed model has a better classification effect and can effectively improve the intelligent recognition effect of COVID-19 images.