Single Transformer Research Articles

Cooperation of the Plasma Reactor with a Converter Power Supply Equipped with a Transformer with Special Design

Plasma generation by means of electrical discharge requires specialized power supply systems. The applicability of plasma for various plasma processes depends on its parameters, and these, in turn, depend on the parameters of power supply systems. Arc plasma can be unstable, generating a lot of electromagnetic interference, overvoltage and overcurrent. The power system of a plasma reactor must guarantee good plasma control characteristics, be immune to disturbances and ensure good cooperation with the power grid. This article analyzes the cooperation of a three-phase plasma reactor, with a gliding arc discharge and a power supply system of a new type. This system integrates an AC/DC/AC converter and a five-column transformer with a special design in a single device. Using the properties of magnetic circuits, it is possible to integrate the functions of ignition and sustaining the burning of the discharge in the reactor in a single transformer. The proper design of the transformer is crucial to achieve the good cooperation of the AC/DC/AC converter with both the plasma reactor and the power supply network. The presented power supply design shows a number of positive features, predisposing it to powering arc plasma reactors.

Distilling Knowledge From an Ensemble of Vision Transformers for Improved Classification of Breast Ultrasound

To develop a deep learning model for the automated classification of breast ultrasound images as benign or malignant. More specifically, the application of vision transformers, ensemble learning, and knowledge distillation is explored for breast ultrasound classification. Single view, B-mode ultrasound images were curated from the publicly available Breast Ultrasound Image (BUSI) dataset, which has categorical ground truth labels (benign vs malignant) assigned by radiologists and malignant cases confirmed by biopsy. The performance of vision transformers (ViT) is compared to convolutional neural networks (CNN), followed by a comparison between supervised, self-supervised, and randomly initialized ViT. Subsequently, the ensemble of 10 independently trained ViT, where the ensemble model is the unweighted average of the output of each individual model is compared to the performance of each ViT alone. Finally, we train a single ViT to emulate the ensembled ViT using knowledge distillation. On this dataset that was trained using five-fold cross validation, ViT outperforms CNN, while self-supervised ViT outperform supervised and randomly initialized ViT. The ensemble model achieves an area under the receiver operating characteristics curve (AuROC) and area under the precision recall curve (AuPRC) of 0.977 and 0.965 on the test set, outperforming the average AuROC and AuPRC of the independently trained ViTs (0.958±0.05 and 0.931±0.016). The distilled ViT achieves an AuROC and AuPRC of 0.972 and 0.960. Both transfer learning and ensemble learning can each offer increased performance independently and can be sequentially combined to collectively improve the performance of the final model. Furthermore, a single vision transformer can be trained to match the performance of an ensemble of a set of vision transformers using knowledge distillation.

Soft-Switching Resonant DC Transformer for Class-D Switching Amplifier With Symmetric Bipolar Outputs

In this letter, soft-switching resonant DC transformer (DCX) for the class-D switching amplifier with the symmetric bipolar outputs is proposed. The proposed DCX consists of a single transformer primary inverter circuit and two transformer secondary dual series resonant half-bridge rectifier circuits, which reduces the voltage stress of the rectifier diodes. In the proposed DCX, the duty cycle of the primary inverter circuit is 0.5, and the symmetric bipolar output voltage gains only depend on the transformer. By dual series resonance between the leakage inductance and the output resonant capacitors, ZVS turn-on of switches and ZCS turn-off of the diodes can be achieved, which improves the overall efficiency. The characteristic of the DCX is elaborated and analyzed in this letter. Finally, a 60 W prototype is established and the experimental verified the theoretical analysis of the proposed DCX.

DBM-ViT: A multiscale features fusion algorithm for health status detection in CXR / CT lungs images

COVID-19 is a severe acute respiratory syndrome caused by SARS-CoV-2. It is highly contagious and spreads rapidly around the world. Although reverse transcription-polymerase chain reaction (RT-PCR) of viral nucleic acid is standard practice in confirmed patients. However, in the early stage of infection, RT-PCR has the disadvantages of low detection sensitivity and long detection cycles. Chest X-ray (CXR) and lung computed tomography (CT) scans combined with intelligent image recognition have the advantages of fast detection speed and low price. However, the traditional convolutional neural network (CNN) has a small receptive field in the image recognition process, which is not conducive to capturing the global features of the image. A single vision transformer (ViT) model lacks the characteristics of inductive bias. These affect the extraction of image fusion features. This paper proposed a DBM-ViT model for deep learning. The model utilized CXR/CT lung images for effective health detection of normal, COVID-19 and other types of pneumonia. The model employed depthwise convolutions with different expansion rates to efficiently captured global information from CXR/CT lung images. Then, the lung feature maps with combined sequences were fed into the ViT module to capture local information. Multi-scale features combined with global and local information ensure maximum feature learning. The results show that the detection accuracy of the DBM-ViT model in the CXR/CT image dataset reached 97.25%/98.36%. This method can effectively capture global and local information in lung images with high detection accuracy and can be used for rapid auxiliary diagnosis of pneumonia types.

Sympathetic Transformer Inrush Detection and Relay Blocking During Power System Restoration

A classic inrush current occurs during normal energization of reactive grid elements like transformers. It can simulate short-circuit currents. Hence inrush currents have to be detected and protection relays need to be blocked for avoiding their overfunction. Conventional detection criteria are the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$2^{\textrm {nd}}$</tex-math></inline-formula> harmonic content and some kind of current waveform analysis. A classic inrush can evolve into a sympathetic inrush event, if the energization of a single transformers drives other transformers into magnetic saturation. The risk of facing a sympathetic inrush current is particularly high during a power system restoration process. Its characteristic waveform differs from the classic inrush. While the classic inrush has unipolar current peaks, the sympathetic inrush is characterized by bipolar peaks. The conventional detection criteria tend to fail in this case. This paper proposes a new approach to detect sympathetic transformer inrush currents. The approach includes the detection of classic inrush currents as well as the distinction against short-circuit currents for ensuring the relay blocking. First it introduces the basic physical mechanism behind the evolution of a classic to a sympathetic inrush. The higher risk of facing a sympathetic inrush during the restoration process is shown by the physical mechanism as well. The proposed detection method is then introduced in detail. Three new differentiation criteria will be applied and merged together to a final decision. The paper closes with practical lab tests of the new approach on simulated sympathetic and measured classic inrush currents. All lab tests showed good results and promise a practical implementation in the grid.

MultiIB-TransUNet: Transformer with multiple information bottleneck blocks for CT and ultrasound image segmentation.

Accurate medical image segmentation is crucial for disease diagnosis and surgical planning. Transformer networks offer a promising alternative for medical image segmentation as they can learn global features through self-attention mechanisms. To further enhance performance, many researchers have incorporated more Transformer layers into their models. However, this approach often results in the model parameters increasing significantly, causing a potential rise in complexity. Moreover, the datasets of medical image segmentation usually have fewer samples, which leads to the risk of overfitting of themodel. This paper aims to design a medical image segmentation model that has fewer parameters and can effectively alleviateoverfitting. We design a MultiIB-Transformer structure consisting of a single Transformer layer and multiple information bottleneck (IB) blocks. The Transformer layer is used to capture long-distance spatial relationships to extract global feature information. The IB block is used to compress noise and improve model robustness. The advantage of this structure is that it only needs one Transformer layer to achieve the state-of-the-art (SOTA) performance, significantly reducing the number of model parameters. In addition, we designed a new skip connection structure. It only needs two 1× 1 convolutions, the high-resolution feature map can effectively have both semantic and spatial information, thereby alleviating the semanticgap. The proposed model is on the Breast UltraSound Images (BUSI) dataset, and the IoU and F1 evaluation indicators are 67.75 and 87.78. On the Synapse multi-organ segmentation dataset, the Param, Hausdorff Distance (HD) andDice Similarity Cofficient (DSC) evaluation indicators are 22.30, 20.04 and 81.83. Our proposed model (MultiIB-TransUNet) achieved superior results with fewer parameters compared to other models.

Monitoring Transformer Condition with MLP Machine Learning Model

Failures of large power transformers in transmission system are always followed by significant costs, which is especially problematic because they present an unplanned expenditure. Aside from derailing financial plans, these events can lead to lower system reliability. This paper describes the development and potential application of transformer model based on multilayer perceptron class of artificial neural networks. Model is built in Python programming language and data collected over the span of one year for a single transformer. Three input features (oil temperature, winding current and outside temperature) are used in the input layer, with the goal of predicting the winding temperature in the transformer. Predicted temperature of the windings can then be compared with the actual winding temperature, which can serve as an indicator of transformers internal condition. Two types of transformer condition degradation are simulated to show the model in action, and certain indicators are explored.

Cuckoo search firefly aided optimization framework for design of medium voltage micro-grid connected soft switching solid-state transformer (MS4T)

A single phase Soft Switching-Solid State Transformer (SS-SST) design is proposed with H-bridge topology as an alternative solution to fulfil the demand of low (or) medium grid power applications. A medium/low frequency transformers fed with H-bridge circuit are incorporate without DC-voltage link, and it’s provided sinusoidal output voltage into the grid. An optimization of Cuckoo Search Firefly (CSF) algorithm was proposed in this research to find optimum switching angle and duty cycle in bridge circuit unit. At present optimum grid power is achieved a maximum efficiency of medium/low power frequency with the help of proposed SS-SST (MS4T) model. For proposed design is used to electric aircraft, ship power systems, battery energy storage systems (BESS) and fast charging electric vehicles (EV). Which are appealing the networks of medium-voltage DC (MVDC). Proposed MS4T design is based on soft-switching transformer with low conduction loss, low EMI and high efficiency via H-bridge converter circuit. The capacitor voltage balancing control between cascade module and design of the component including a medium level voltage frequency transformer that is implement a 1 kV to 0.25 kV MS4T described. Therefore, the efficacy of the present investigations are established with MATLAB platform. The medium voltage Micro Grid (MG) output is estimated under different operation load conditions. A simulation result of the grid power is measured minimum harmonics level by using optimum switching angle, switching frequency and duty cycle arrangements.

Shallow-Guided Transformer for Semantic Segmentation of Hyperspectral Remote Sensing Imagery

Convolutional neural networks (CNNs) have achieved great progress in the classification of surface objects with hyperspectral data, but due to the limitations of convolutional operations, CNNs cannot effectively interact with contextual information. Transformer succeeds in solving this problem, and thus has been widely used to classify hyperspectral surface objects in recent years. However, the huge computational load of Transformer poses a challenge in hyperspectral semantic segmentation tasks. In addition, the use of single Transformer discards the local correlation, making it ineffective for remote sensing tasks with small datasets. Therefore, we propose a new Transformer layered architecture that combines Transformer with CNN, adopts a feature dimensionality reduction module and a Transformer-style CNN module to extract shallow features and construct texture constraints, and employs the original Transformer Encoder to extract deep features. Furthermore, we also designed a simple Decoder to process shallow spatial detail information and deep semantic features separately. Experimental results based on three publicly available hyperspectral datasets show that our proposed method has significant advantages compared with other traditional CNN, Transformer-type models.

SPU-BERT: Faster human multi-trajectory prediction from socio-physical understanding of BERT

Accurately predicting pedestrian trajectories requires a human-like socio-physical understanding of movement, nearby pedestrians, and obstacles. However, traditional methods struggle to generate multiple trajectories in the same situation based on socio-physical understanding and are computationally intensive, making real-time application difficult. To overcome these limitations, we propose SPU-BERT, a fast multi-trajectory prediction model that incorporates two non-recursive BERTs for multi-goal prediction (MGP) and trajectory-to-goal prediction (TGP). First, MGP predicts multiple goals through generative models, followed by TGP generating trajectories that approach the predicted goals. SPU-BERT can simultaneously understand movement, social interaction, and scene context from trajectories and semantic maps using a single Transformer encoder, providing explainable results as evidence of socio-physical understanding. In experiments, SPU-BERT accurately predicted future trajectories (with 0.19 m and 7.54 pixels of ADE20 for the ETH/UCY datasets and SDD) with over 100 times faster computation (0.132 s) than the state-of-the-art method. The code is available at https://github.com/kina4147/SPUBERT.

Dynamic State Evaluation Method of Power Transformer Based on Mahalanobis–Taguchi System and Health Index

Health status assessment is the key link of transformer-condition-based maintenance. The health status assessment method of power transformers mostly adopts the method based on the health index, which has the problems of multiple parameters of each component and strong subjectivity in the selection of weight value, which is easily causes misjudgment. However, the existing online monitoring system for dissolved gas in transformer oil (DGA) can judge the normal or abnormal state of the transformer according to the gas concentration in a monitoring cycle. Still, there are problems, such as fuzzy evaluation results and inaccurate judgment. This paper proposes a dynamic state evaluation method for power transformers based on the Mahalanobis–Taguchi system. First, the oil chromatography online monitoring time series is used to screen key features using the Mahalanobis–Taguchi system to reduce the problem of excessive parameters of each component. Then, a Mahalanobis distance (MD) calculation is introduced to avoid subjectivity in weight selection. The health index (HI) model of a single transformer is built using the MD calculated from all DGA data of a single transformer. Box–Cox transformation and 3 σ criteria determine the alert value and threshold value of all transformer His. Finally, taking two transformers as examples, we verify that the proposed method can reflect the dynamic changes of transformer operation status and give early warning on time, avoiding the subjectivity of parameter and weight selection in the health index, which easily causes misjudgment and other problems and can provide a decision-making basis for transformer condition-based maintenance strategies.

Experimental Assessment of Ground Fault Protection in Frequency-Selective Grounded Systems Fed by a Single Transformer

This paper presents performance assessment of ground fault protection for systems supplied by a transformer, which has frequency-selective grounding. Frequency-selective grounding is a new grounding system that is designed to act as a solid-grounding for high frequency currents, and as a low resistance-grounding for low frequency currents. This grounding system is composed from a parallel <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$R-C$</tex-math></inline-formula> circuit with an equivalent impedance able to affect system zero-sequence currents and ground potentials. Such effects on zero-sequence currents and ground potentials can directly impact the accuracy and response speed of ground fault protection. Effects of frequency-selective grounding on ground fault protection are experimentally assessed for different fault and non-fault events, when the transformer supplies linear and non-liner loads. Experimental results demonstrate responses of various implementations of ground fault protection to fault and non-fault events, when the protected system has frequency selective grounding. Moreover, presented results and discussions aim to demonstrate the ability of frequency-selective grounding to support the operation of ground fault protection during different loading conditions and fault events.

Evaluation method of power loss for distribution network fault handling based on distribution transformer outage and restoration information matrix

To solve the problem of power loss evaluation in distribution network fault handling, an evaluation method of power loss for distribution network fault handling based on distribution transformer outage and restoration information matrix is proposed. Firstly, the distribution transformer outage and restoration information matrix model is established to describe the distribution transformer, its outage time, its restoration time, its outage duration, and its lost power affected by the distribution network failure outage. Then the outage duration in the process of fault handling is discussed in four cases, and the detailed calculation formula is given. At last, the power loss of a single distribution transformer, the total power loss of distribution network failure, and the average power loss of a distribution transformer are described in detail by outage time and predicted power. The example analysis shows that this method can quickly estimate the power loss of a distribution transformer in the process of fault disposal, and the method is simple and reliable, which can meet the needs of field applications and has good promotion value.

High Efficiency and High Power Density Partial Power Regulation Topology With Wide Input Range

Data centers require high efficiency and high power-density dc–dc converter with wide input range and galvanic isolation. To overcome the drawbacks of traditional single-stage <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLC</i> and two-stage solutions, a novel partial power regulation topology with wide input range is proposed in this article. The proposed topology delivers the load power through a three-ports <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLC</i> -based dc transformer regulated DCX (DCX). A pulsewidth modulation (PWM) converter is cascaded with one of the input ports of the DCX and then in series with the other input port to regulate the output voltage. Since a single transformer and a partial power PWM converter are employed, high power density can be achieved. Furthermore, only a portion of the load power is transferred by the PWM converter, the efficiency sacrifice is reduced. By introducing an appropriate negative current, the zero voltage switching of the PWM is also achieved. Furthermore, when the input voltage approaches the lower or the upper limit, the switching frequency of the PWM regulator can be ultralow. As a result, switching losses and core losses are further reduced. Finally, a prototype with 190—475 V input and 12 V/500 W output is demonstrated, which achieves a peak efficiency of up to 97.2% and a power density of 153 W/in <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> .

Differential Power Processing Converter With Active Clamp Structure and Integrated Planar Transformer for Power Generation Optimization of Multiple Photovoltaic Submodules

In recent years, in order to improve the fuel efficiency of electric vehicles, a lot of research has been conducted to adopt photovoltaic (PV) systems to electric vehicles. In addition, various Differential Power Processing (DPP) converters that can mitigate power imbalance among submodules in PV systems are being studied and developed. However, most conventional DPP converters suffer from large volume, low-price competitiveness, and low efficiency. Therefore, in this paper proposes a novel DPP converter. The proposed DPP converter can significantly reduce the volume of converter by integrating multiple transformers into single transformer. Moreover, because of the active clamp structure, it is possible to eliminate secondary side circuits. Furthermore, more than 95% efficiency under 10-60% load and maximum 96.654% efficiency about 20% load were achieved. Thus, the proposed DPP converter can achieve high power density, high price competitiveness, as well as high efficiency. To support the validity of the proposed converter, a prototype of the proposed DPP converter was built, and experiments were proceeded with 8V input and 60W/30V output specifications.

Positional Attention Guided Transformer-Like Architecture for Visual Question Answering

Transformer architectures have recently been introduced into the field of visual question answering (VQA), due to their powerful capabilities of information extraction and fusion. However, existing Transformer-like models, including models using a single Transformer structure and large-scale pre-training generic visual-linguistic models, do not fully utilize both positional information of words in questions and positional information of objects in images, which are shown in this paper to be crucial in VQA tasks. To address this challenge, we propose a novel positional attention guided Transformer-like architecture, which can adaptively extracts positional information within and across the visual and language modalities, and use this information to guide high-level interactions in inter- and intra-modality information flows. In particular, we design and assemble three positional attention modules into a single Transformer-like model MCAN. We show that the positional information introduced in intra-modality interaction can adaptively modulate inter-modality interaction according to different inputs, which plays an important role for visual reasoning. Experimental results demonstrate that our model outperforms the state-of-the-art models and is particularly good at handling object counting questions. Overall, our model achieves the accuracy of 70.10%, 71.27%, 71.52% on the datasets of COCO-QA, VQA v1.0 test-std and VQA v2.0 test-std, respectively. The source code will be publicly available at https://github.com/waizei/PositionalMCAN.

Diagnosis of transformers of electrical complexes using non-contact laser vibrometers

TARGET. The purpose of this work is to improve the method of vibration control of the active part of a power transformer by fractal analysis of its vibration signal. Fractal analysis of the magnitude-time characteristics of the quantitative assessment of the degree of "jagged" vibration signal of the transformer. For a quantitative assessment, the coefficient of fractal analysis (FAC) was introduced based on the determination of the fractal dimension by the Hausdorff-Besikovich method. To test the developed method of application, a non-contact laser measuring diagnostic complex (MDC) with developed software based on LabVIEW, ImageJ and Python. Experimental studies of the technical condition of the transformer TSZ 16 were carried out on the basis of the developed method using a non-contact LCIK, the CFA of the windings and the magnetic circuit of a single transformer was determined, the level of technical condition for the impulse elements of the transformer was determined.METHODS. The vibration control method makes it possible to control a power transformer during its operation under voltage, which makes it possible to switch from a planned system of transformer repairs to a system for taking repairs according to the current technical condition.RESULTS. The improved method of vibration control was tested using the developed noncontact LCIK, the level of technical condition of the power transformer under voltage was determined.CONCLUSION. An improved method of vibration control makes it possible to determine the level of technical condition of an energized power transformer with the possibility of automatically obtaining a decision on the technical condition, as well as to use statistical methods for processing and analyzing signals received from the transformer.

HTMatch: An efficient hybrid transformer based graph neural network for local feature matching

Local feature matching plays a vital role in various computer vision tasks. In this work, we present a novel network that combines feature matching and outlier rejection for finding reliable correspondences between image pairs. The proposed method is a hybrid transformer-based graph neural network (GNN), termed HTMatch, which aims to achieve high accuracy and efficient feature matching. Specifically, we first propose a hybrid transformer that integrates self- and cross-attention together to condition the feature descriptors between image pairs. By doing so, the intra/inter-graph attentional aggregation can be realized by a single transformer layer, which achieves more efficient message passing. Then, we introduce a new spatial embedding module to enhance the spatial constraints across images. The spatial information from one image is embedded into another, which can significantly improve matching performance. Finally, we adopt a seeded GNN architecture for establishing a sparse graph, which improves both efficiency and effectiveness. Experiments show that HTMatch reaches state-of-the-art results on several public benchmarks.

Modeling synthetic power distribution network and datasets with industrial validation

Creating synthetic networks and datasets for power distribution network is challenging due to continuous expansion of networks, integration of new low carbon technologies and large penetration of renewable energy resources in network. In this paper, a practical approach for generating synthetic distribution networks and datasets using public databases and data synthesis algorithms is proposed. A synthetic power distribution network is developed by leveraging the open-data from local government databases, OpenStreetMaps and mapping engines such as Google Street View. New data synthesis algorithms are proposed to obtain the missing network datasets. The proposed algorithms include a topology for designing power lines, a method for computing the lengths of power lines, a hub-line algorithm for determining the number of consumers connected to a single transformer, a virtual layer approach based on FromNode and ToNode for establishing electrical connectivity, and a technique for ingesting raw data into industrial data platforms. The practical feasibility of the proposed solutions is demonstrated by an illustrative case study of the Colac region in Australia. Synthetic network and datasets are created for the distribution feeder, and then evaluated in industry servers. The results are compared using a three-step validation procedure: comparing the synthetic and actual network datasets using geo-based visualizations, by including feedback from industry experts familiar with the analysis, and by testing the generic applicability of the proposed techniques to other regions. The paper compares network elements that include 4714 power lines, 48 distribution transformers, 4155 energy consumers, 609 electrical nodes, and 1 substation. The comparison results demonstrate the efficacy of developed networks and datasets as they show resemblance to real network and datasets while providing the geographical validation of distribution network models.

Fully-Integrated Planar Transformer With a Segmental Shunt for LLC Resonant Converters

This article presents two topologies, which provide high leakage inductance in shunt-inserted integrated magnetic transformers. These differ from conventional designs by replacing the low-permeability magnetic shunt of a planar transformer with high-permeability magnetic shunt segments, separated by many small air gaps. This approach results in a shunt with the same bulk permeability as the conventional design, while using lower cost and readily available magnetic materials such as ferrite. A modeling and design approach, which can estimate the leakage and magnetizing inductances precisely is provided for each topology. Theoretical analysis is presented and verified using finite-element analysis and experimental implementation. Ac resistance analysis for both transformer topologies is also presented. In addition, an <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLC</i> resonant converter is built to verify the performance of the proposed fully-integrated magnetic transformers in practice. It is shown that the proposed topologies can integrate all three magnetic components of an isolated <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">LLC</i> resonant converter in a single planar transformer, which reduces the converter's volume and cost.