Search

This paper presents a systematic overview of the resilience framework for power electronics converter-based cyber-physical systems (CPSs), emphasizing end-to-end responses to the emerging challenges posed by cyber-physical attacks. While recent advancements in control and communication have enhanced the functionalities of power electronic converters in applications such as electric vehicles and smart grids, such advancements have also introduced new vulnerabilities, particularly to cyber-physical attacks. The existing literature tends to focus on isolated research areas, and a comprehensive review that encompasses strategies applicable before, during, and after an attack remains lacking. To address this gap, this paper categorizes state-of-the-art research into four key stages of an attack: anticipate and prepare, resist and absorb, detect and evaluate, and recover. Topics reviewed in this paper include converter and cyber-physical attack modeling, converter-based CPS testbeds, system hardening at the cyber and physical layers, attack detection, post-attack evaluation, and recovery strategies. Real-world case studies and practical regulations are also analyzed. Additionally, challenges and opportunities of the resilience framework are discussed. Despite notable advancements in modeling and identifying cyber-physical attacks, considerable efforts are still required to improve attack mitigation strategies and recovery mechanisms.

Compound faults easily happen in rolling bearing due to the complex working environment. Diagnosing compound faults accurately is a thorny problem, which can ensure the normal operation of mechanical structure. To tackle this problem, this paper proposes a novel method called double-domain reweighted adaptive sparse representation. The proposed method reweights the weight through the fault information from the wavelet domain and time domain. In the wavelet domain, periodic clustering similarity is proposed to measure fault features by similarity calculation. Then, wavelet domain information is transformed into time domain. In time domain, the Hadamard product of the sub-band signals is calculated to obtain fault features. The fault features in time domain are further transferred into wavelet domain, which is used for reweighting the weight. The proposed model’s optimal parameters are determined through the hippopotamus optimization algorithm, with the proposed frequency correlated kurtosis as the objective function. Frequency correlated kurtosis measures the signal’s frequency domain periodicity. The proposed method is verified by simulation, experimental signals, and comparison experiments, whose results show the effectiveness and superiority of the proposed method.

The rapid advancement of technology has made network security a hot topic of concern for researchers worldwide. Therefore, to improve the accuracy and real-time response capability of network intrusion detection systems, and to effectively detect and analyze network attack forms in complex network environments, starting from the drawbacks of existing network intrusion detection methods and the vulnerable attack modes of networks, this study first introduces variational auto-encoder to improve the semi-supervised intrusion detection algorithm. Subsequently, the labeled dataset is expanded using K-means clustering algorithm and collaborative training algorithm. Finally, a novel network intrusion detection model based on multi-level semi-supervised intrusion detection algorithm is proposed. The experimental outcomes denoted that the model achieved the highest classification accuracy of 93.08%, 92.12%, 91.22%, and 96.38% for four common types of network attacks: denial of service attacks, unauthorized access from remote machines, unauthorized access attacks by ordinary users to local superuser privileges, and cross-site scripting attacks, respectively. The results of the network attack confusion matrix indicated that the proposed model could detect and classify all types of network attacks, with excellent detection applicability and classification efficiency, and its scores were all above 60 points. In addition, the running time of the proposed model was only 25.64 seconds. From this, the proposed method effectively improves the dynamic adaptability and accuracy of network intrusion detection, providing an efficient and accurate solution for network security defense.

Snow depth is a fundamental parameter in hydrological and climate models, which is crucial for studying climate change, hydrological cycles, and ecosystem changes. Polarimetric synthetic aperture radar interferometry (PolInSAR) is one of the most promising snow depth retrieval methods, which is sensitive to the shape, direction, and vertical distribution of targets. The dense medium random-volume-over-ground (DM-RVoG) model for PolInSAR has been shown to be workable for snow depth retrieval, it still suffers from the inaccuracy of the parameters representing the phase center and decorrelation. In this study, based on the backscattering mechanism of snow, a novel snow depth retrieval method is proposed to improve the DM-RVoG model using polarization decomposition and decorrelation optimization. First, the polarization decomposition is extended to obtain the ground scattering phase. Then, the coherence region boundary estimation method is put forward to obtain pure volume decorrelation. Finally, the proposed method is validated using Ku-band UAV SAR data, and the accuracy is assessed using in situ data. The correlation coefficient, root mean square error, and mean absolute error of the proposed method are 0.88, 4.98, and 4.08 cm, respectively, demonstrating significant improvements compared with the original method.

A trajectory planning method for local obstacle avoidance based on an improved artificial potential field (APF) method is proposed, which is aimed at the problem for dual motor driven unmanned tracked vehicles avoiding dynamic and static obstacles in unstructured environments. Firstly, in traditional artificial potential fields, by adding virtual target points, unmanned tracked vehicles can avoid large obstacles and reach the target point in off-road environments. Secondly, a water droplet type repulsive potential field function for static obstacles and an improved dynamic obstacle potential field function including relative velocity function and relative acceleration function are established in the proposed improved APF method to improve the smoothness of lane changing obstacle avoidance paths. The simulation results of overtaking and obstacle avoidance in the same direction show that the change in heading angle is reduced by 42.9%, and the lateral displacement is reduced by 39.5%. Finally, a trajectory planning method based on improved APF for obstacle avoidance and lane changing of the unmanned tracked vehicle is constructed, which also considers the speed planning with kinematic and dynamic constraints. For obstacle avoidance under lateral meeting condition, the collaborative simulation results of Prescan-Adams-Matlab/Simulink show that the change in heading angle is reduced by 84%, and the lateral displacement is almost zero. Under complex working conditions with multiple static and dynamic obstacles, the results of hardware in loop (HIL) simulation testing and vehicle experiments show that the number of drastic changes in turning radius and heading angle of the vehicle is significantly reduced, and the maximum amplitude was reduced by 63.2% and 37.5% respectively, making the vehicle’s obstacle avoidance and lane changing safer, smoother, and more efficient.

Three-dimensional (3D) video with depth information is essential for many applications in the consumer electronics industry. The 3D-high efficiency video coding (3D-HEVC) is the latest 3D video coding standard. Nonetheless, it utilizes various complex coding techniques to create extra intermediate views for better representation of 3D videos, which imposes significant challenges for real-time 3D video applications. Specifically, the high complexity of 3D-HEVC intra coding could be a significant barrier to the adoption of 3D video in consumer electronics. Therefore, in this research, a low-complexity 3D-HEVC intra coding technique is proposed. Firstly, we perform a complexity analysis of 3D-HEVC intra coding. Secondly, we develop a multi-layer features fusion (MLFF) model to estimate the optimal coding tree unit (CTU) depth and prediction unit (PU) mode. Thirdly, to improve the model’s prediction accuracy, we incorporate two external features into the model: the quantization parameter (QP) and texture complexity. Finally, we embed the MLFF model into the 3D-HEVC test platform. The experimental results demonstrate that the suggested method can effectively reduce the 3D-HEVC intra coding time with a small amount of rate-distortion (RD) performance loss while maintaining the subjective quality of the synthesized view.

In this article, a honeycomb vertical surrounding gate access transistor array scheme is proposed to further decrease the DRAM cell area with aggressively shrink bit line (BL) pitch and word line (WL) pitch adopting the ZigZag BL and WL air gap. To verify the process feasibility, process flow emulation is optimized by virtual fabrication with SEMulator3D®. Moreover, the ZigZag BL feasibility in lithography process is evaluated by optical proximity correction (OPC) simulation. In addition, the parasitic capacitance of BL and WL decrease 22.8% and 76.8%, respectively, as compared to reference paper. And the electrical properties of the proposed device are simulated by three-dimensional technology computer aided design (3D TCAD). The GIDL effect is prohibited through prolonging the drain extension and reducing its doping concentration. Finally, the surrounding gate transistor can achieve high on-state current ( $30~\mu \text{A}$ ) at $\text{V}_{\mathrm {g}}=1.5$ V and off-state current below 0.1 fA at $\text{V}_{\mathrm {d}}= -0.2$ V. These results are beneficial for the direction of pathfinding for increasing the density and decreasing the energy consumption of DRAM.

Centralized deepfake service providers have large amounts of computing power and training data, giving them the ability to produce high-quality deepfake content. However, once these service providers are attacked or malfunction, it may lead to the collapse of the entire deepfake ecosystem, making deepfake a potential threat to data security. This monopoly development has led to the uneven distribution of deepfake resources, which in turn has brought about the risk of single points of failure. To deal with the problem, this paper proposes a decentralized deepfake task management algorithm (DD-TMA) based on blockchain and edge computing. The blockchain in this algorithm can provide a decentralized storage and management platform to ensure that the data and models of deepfake tasks will not be tampered with or lost. Edge computing can distribute tasks to edge devices close to the data source for processing, reducing data transmission delays and bandwidth consumption, and improving the efficiency and security of deepfake tasks. The paper innovatively integrates blockchain, federated computing, and edge computing. Firstly, the algorithm establishes a decentralized computing platform based on blockchain. Subsequently, it enhances computing power during the execution of decentralized deepfake tasks through the integration of federated computing and edge computing. Finally, the algorithm increases the active performers of decentralized deepfake tasks through gamification, thereby improving task execution efficiency. Experiments conducted in this study on public data sets demonstrate that the algorithm is efficient, robust, and reusable. Compared with other algorithms, the efficiency of DD-TMA is improved by more than 20% and the stability is improved by more than 13%. This algorithm proves effective in solving the problems encountered in the execution of centralized deepfake tasks. The research provides new ideas for future evaluations of decentralized deepfake effects based on different strategies.

The rapid identification of landslides in mountainous terrain through remote sensing imagery after earthquakes or heavy rainfall is crucial for assessing the severity of the situation and providing timely assistance for post-disaster rescue operations. However, traditional manual interpretation methods and computer vision techniques tend to be inefficient and time-consuming. To address this problem, we propose a novel algorithm called LSI-YOLOv8, which builds upon the lightweight YOLOv8 to enable fast and accurate landslide identification using remote sensing imagery. In this study, we introduce optimization techniques integrated into the YOLOv8 model. Firstly, we incorporate an expandable residual attention Dilation-wise Residual Segmentation (DWRSeg) module to reconstruct its neck structure, reducing parameters and computational load. Additionally, we enhance the positioning ability of the model by replacing the bounding box loss function Complete Intersection over Union (CIoU) with Efficient Intersection over Union (EIoU), optimizing it to improve computing speed. Through comparative experiments with other state-of-the-art models, the accuracy of our LSI-YOLOv8 significantly outperforms Mask-RCNN, YOLOv5, YOLOv7, YOLOv8, and YOLOv9 model, with mAP@0.5 being 9.0%,7.7%, 4.0%, 6.3%, and 2.9% higher than other models, while mAP@0.5 :0.95 is 3.1%,3.6%, 2.7%, 4.8%, and 0.3% higher than others, respectively. Furthermore, the detecting speed of the LSI-YOLOv8 model is 73.2 f/s, which is notably faster by 15.5 f/s, 3.5 f/s, 5.1 f/s, 5.9 f/s, and 4.3 f/s respectively, compared to Mask-RCNN, YOLOv5, YOLOv7, YOLOv8, and YOLOv9. The proposed method provides valuable insights for advancing landslide identification methodologies. The focus of subsequent work will be more on multi-source data fusion, such as Digital Elevation Model (DEM) data and Synthetic Aperture Radar (SAR) data to remove the influence of some background information to improve the accuracy of landslide identification.

Multimodal named entity recognition plays a crucial role in the construction process of knowledge graphs as it directly influences the quality of entity extraction and classification, which in turn affects the overall quality of knowledge graph construction. However, most existing multimodal named entity recognition algorithms do not consider the correlation between text and images. They either use visual features of images as the attention of the text modality or fuse them with textual features. In the case of multimodal tweets containing both text and images, three categories of data can be identified based on the correlation between the two: text that is related to images, text that is partially related to images, and text that is not related to images. Using irrelevant or partially relevant image features as text cross-modal attention can result in incorrect text representation, ultimately leading to misclassification of entities and negatively impacting the model’s performance. To address the problem of uncertainty or negative impact caused by the lack of relevance or partial correlation between text and images, this paper proposes a visually enhanced text representation algorithm based on a hybrid of dynamic gating and correlation coefficient. We conducted experiments on two benchmark datasets, namely Twitter-2015 and Twitter-2017. The experimental results were analyzed comprehensively to showcase the strengths of the proposed model.

In this paper, we investigate the electrical characteristics of AlGaN/GaN HEMTs at the lowest temperature of 20 K. The measurement results indicate that the output current of the device decreases significantly with increasing temperature at temperature ranging from 40 K to 260 K, and the saturation drain current decreases by 19%. The gate leakage current rises slightly when the temperature increases. However, both the transfer and C-V characteristics indicate that the threshold voltage shift slightly in a negative direction as the temperature rises. In order to determine the physical mechanism of electrical characteristics change, the low-frequency noise (LFN) characteristics at different temperatures were measured and the density of traps was extracted. Finally, we consider that there are two competing mechanisms affecting the electrical characteristics of devices. The trap density reduction caused by temperature rise leads to threshold voltage’s negative shift, while the drop of 2DEG mobility is the main reason for the decrease of output current.

With the development of the metro system, an increasing number of new metro lines are being constructed. However, the lack of historical data on the new metro lines leads to the challenge of predicting the origin-destination distribution of the whole network. Based on the advantage of Automatic Fare Collection data, this study proposes a novel method to predict origin-destination distribution over metro networks with expansions. Firstly, the disaggregate model is built to predict the origin-destination distributions of the whole network with consideration of passengers’ destination choice behavior. Then, the aggregate model which can capture the historical passenger distribution patterns is developed to correct the origin-destination distribution in existing metro lines. Finally, this study combines the results of the disaggregate model and aggregate model to obtain the whole network origin-destination distribution. This study applies the method to the Guangzhou Metro network, and the results show the superiority of the proposed model. The results can help to improve the operation and management of urban rail transit.

With the development of the brain-computer interface, it is possible to assess the image quality based on an electroencephalogram (EEG) signal, which is essential to construct a quality assessment system that accords with the characteristics of the human visual system. However, as the complexity of image degradation levels rises, the analysis of brain features becomes progressively more challenging, leading the neural mechanisms still unclear. In this paper, an image quality assessment experiment with three levels of degradation to explore the neural mechanisms of subjects when confronted with images degraded at multiple levels. Subsequently, we proposed a dynamic model combined with Topological Data Analysis (TDA) and Grey Theory to extract the feature of brain response to different distortion-level images, which is called grey-topological data analysis (Grey-TDA) in this paper. The results indicate that the proposed method is effective in identifying brain responses under stimuli of varying image qualities, which contributes to the research of image quality assessment from the perspective of brain cognition.

The active neutral-point-clamped five-level converter has drawn much attention in industrial applications. However, common-mode voltage (CMV) suppression, accurate track current, fixed switching frequencies, neutral-point (NP) voltage balance, and flying capacitor (FC) voltage are compulsory in this configuration. Thus, an improved multiobjective model-predictive control (IMMPC) without weighting factors is proposed in this article. First, to suppress CMV, only 61 low-CMV vectors out of 125 vectors were selected. Second, to ensure current tracking performance and reduce computational burden, based on the gh oblique coordinate, three voltage vectors were calculated online in each control period, which can avoid lookup tables. Third, the optimal switching sequence and duty cycles were presented considering the lowest switching numbers and constant switching frequency. A high-quality current was thus obtained, and the NP voltage and the FC voltage were balanced by selecting switching combinations. The IMMPC can be applied to other multilevel converters for current tracking and CMV suppression with a simple adjustment on parameters. Finally, the feasibility and the effectiveness of the proposed method were verified through simulations and experiments.

3D Skeleton-based human motion prediction, a classic task in computer vision, aims to forecast subsequent motions based on historical motion observations. In particular, precise short-term motion prediction is crucial for the effectiveness of machines designed for real-time human-computer interaction. This study aims to achieve accurate predictions of human motion within timeframes of less than 400 milliseconds, with the goal of improving machine responsiveness and efficiency. Previous research has predominantly relied on sequence models like Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) or Transformers. Transformer-based methods have been relied on temporal forecasting ability of Transformer. In this study, instead, we introduce an alternative perspective for Transformer-based models to comprehend the structure of skeletons, named Joint-Aware Transformer (JAT). Within our model, the attention mechanism is employed to encode inter-joint correlation instead of temporal dependencies. Our approach outperformed the state-of-the-art (SOTA) model in short-term prediction on three types of human motion datasets.

Accuracy of electroencephalography (EEG) source localization relies on the volume conduction head model. A previous analysis of young adults has shown that simplified head models have larger source localization errors when compared with head models based on magnetic resonance images (MRIs). As obtaining individual MRIs may not always be feasible, researchers often use generic head models based on template MRIs. It is unclear how much error would be introduced using template MRI head models in older adults that likely have differences in brain structure compared to young adults. The primary goal of this study was to determine the error caused by using simplified head models without individual-specific MRIs in both younger and older adults. We collected high-density EEG during uneven terrain walking and motor imagery for 15 younger (22±3 years) and 21 older adults (74±5 years) and obtained $\text{T}_{{1}}$ -weighted MRI for each individual. We performed equivalent dipole fitting after independent component analysis to obtain brain source locations using four forward modeling pipelines with increasing complexity. These pipelines included: 1) a generic head model with template electrode positions or 2) digitized electrode positions, 3) individual-specific head models with digitized electrode positions using simplified tissue segmentation, or 4) anatomically accurate segmentation. We found that when compared to the anatomically accurate individual-specific head models, performing dipole fitting with generic head models led to similar source localization discrepancies (up to 2 cm) for younger and older adults. Co-registering digitized electrode locations to the generic head models reduced source localization discrepancies by $\sim $ 6 mm. Additionally, we found that source depths generally increased with skull conductivity for the representative young adult but not as much for the older adult. Our results can help inform a more accurate interpretation of brain areas in EEG studies when individual MRIs are unavailable.

Under the background of dual-carbon goal, renewable energy-oriented power system and integrated energy are the trends of energy system. In order to cope with the uncertainty and volatility of renewable energy output, gas-fired unit can serve as a superior peak-regulation source to effectively solve the supply-demand mismatch problem. Nevertheless, rotor loss of gas-fired unit in the peak-regulation process weakens the operation performance and increases the practical cost of providing peak-regulation service. There is security risk of gas-fired unit in the peak-regulation process, especially during deep peak-regulation and startup/shutdown process, which leads to deficiency in the willingness to provide peak-regulation service. In this context, an optimal scheduling model of integrated electricity and gas system (IEGS) considering peak-regulation loss and renewable energy consumption is proposed in the paper. Firstly, peak-regulation loss and peak-regulation cost model is built under startup/shutdown regulation and variable-load regulation mode of gas-fired unit, respectively. On this basis, the optimal IEGS scheduling model taking gas-fired unit’s peak-regulation loss and renewable energy consumption into account is established, aiming at minimizing the total scheduling cost including operation cost considering peak-regulation loss, renewable energy abandoning penalty, load shedding cost and carbon emission cost under gas network constraints, peak-regulation constraints and other constraints. Finally, a case study on a modified IEGS with a high proportion of renewable energy from Guangdong province, China is conducted to verify the effectiveness of the proposed model, and the simulation results demonstrate that the proposed model could obtain the higher renewable energy consumption level and less carbon emission, and could relieve the rotor loss of gas-fired unit.

The traditional dynamic weapon target assignment model is a combination of multiple static weapon target assignment stages. The assignment of the next stage is carried out after the result of the previous static weapon target assignment is settled. However, between two static weapon target assignment stages, the threat ranking of multiple targets may change with time, and the traditional dynamic weapon target assignment model does not take this time issue into consideration. This paper proposes a “time sampling dynamic weapon assignment model”. This model divides the decision-making stage by setting the time interval of data collection, and it can capture the real-time changes in the target threat degree and make timely decisions. With this model, this study designed a dynamic weapon target assignment method based on the reinforcement learning algorithm. Additionally, according to this method, a comparative experiment with different sampling time divisions was designed, and a better sampling time division method was obtained. Finally, a comparative experiment between the reinforcement learning algorithm and the traditional heuristic algorithm was designed in this study. The simulation results show that, compared with the traditional heuristic algorithm, the proposed assignment model and the reinforcement learning algorithm are better in terms of decision-making timeliness and global considerations.

In this work, a new method for failure analysis of electronic components, high speed camera, is used to investigate burnout failure location of GaN HEMTs under RF overdrive stress. Based on the high speed camera system and the RF test system, we can filter out most of the burn flashes, and clearly locate the weak parts of devices. To further explain the burnout mechanism, a long-term (100 h) RF overdrive stress experiment was carried out and the significant degradation was observed. The drain-source current decreases and the threshold voltage drifts forward. These phenomena show that the degradation of RF overdrive stress is based on hot electron effect (HEE), which is related to the electric field. Besides, Electroluminescence (EL) tests are used and the non-uniform but strong luminescence characteristics of the gate were found, which indicates the strong electric field is the main cause of burnout. We also explore the correlation between burnout and ambient temperature. It was found that the influence of ambient temperature on the burnout was limited. At last, a TCAD simulation is carried out to confirm the temperature and electric field distribution in the device when burnout. It can be found that the electric field inside the device exceeded the breakdown electric field of GaN, which further proves that the burnout caused by RF overdrive is mainly due to electric field rather than temperature.

Small target detection from the infrared remote sensing image is a challenge task. In this article, a novel local adaptive threshold algorithm combined with heterogeneity and compactness filters is proposed to detect the small target from the infrared remote sensing images. First, the infrared image is filtered by a heterogeneity filter to enhance the target saliency. Then, the enhanced image is filtered by a compactness filter to generate a target candidate region map. Finally, for each pixel in the target candidate region, a local adaptive threshold is calculated from the enhanced image to determine whether it is a target pixel or not, and thus, the targets are extracted out. The designed heterogeneity filter and compactness filter can effectively suppress the background clutter, enhance the target, and generate target candidate regions. The proposed adaptive thresholding is a local threshold method, which is calculated in a small local window and can effectively reduce the false alarm and missing alarm. Qualitative and quantitative experiments are conducted on synthetic images and real images. The experiment results show that, with good target enhancement and background suppression, and high detection accuracy, the proposed method outperforms other state-of-the-art methods.

Many thermal power plants use selective catalytic reduction (SCR) systems to complete flue gas denitration. The system has the characteristics of large inertia, large delay and nonlinearity. At the same time, it is difficult to use the controller under a single working condition to meet the control task of the system under the full working conditions. In order to solve the above control difficulties, This paper proposes a two-degree-of-freedom internal model control (2-DOF-IMC) method based on controller parameter adaptation and multi-model adaptation to realize the control of the full working conditions of the SCR system. First, the intelligent identification method based on Differential evolution quantum particle swarm optimization (DEQPSO) is used to obtain the sub-model of ammonia injection-SCR outlet Nitrogen Oxide (NOx) concentration in typical working conditions. A 2-DOF-IMC controller under each sub-model is established. Then, aiming at system set point tracking performance, anti-disturbance performance, and control energy consumption, the Pareto optimal solution of each controller is obtained by using the Non-dominated Sorting Genetic Algorithms-II (NSGA-II) algorithm based on chaotic mutation. Finally, the controller parameter adaptive strategy is designed based on the Takagi-Sugeno (T-S) fuzzy model, and the multi-model adaptive strategy is designed based on the recursive bayesian probability weighting. Through the dynamic performance test and the anti-disturbance performance test, it is verified that the 2-DOF-IMC based on controller parameter adaptation and multi-model adaptation has good control performance and adaptability to all working conditions. The method proposed in this paper combines the advantages of multi-parameter and multi-model strategies, further improves the control quality, and provides a new solution for the control of thermal power plant SCR system.

The neuromorphic event cameras, which capture the optical changes of a scene, have drawn increasing attention due to their high speed and low power consumption. However, the event data are noisy, sparse, and nonuniform in the spatial-temporal domain with an extremely high temporal resolution, making it challenging to process for traditional deep learning algorithms. To enable convolutional neural network models for event vision tasks, most methods encode events into point-cloud or voxel representations, but their performance still has much room for improvement. Additionally, as event cameras can only detect changes in the scene, relative movements can lead to misalignment, i.e., the same pixel may refer to different real-world points at different times. To this end, this work proposes the aligned compressed event tensor (ACE) as a novel event data representation, and a framework called branched event net (BET) for event-based vision under both static and dynamic scenes. We apply them for various object classification and action recognition tasks, and show that they surpass state-of-the-art methods by significant margins. Specifically, our method achieves 98.88% accuracy for the DVS128 action recognition task, and outperforms the second-best method by large margins of 4.85%, 9.56% and 2.33% on N-Caltech101, DVSAction and NeuroIV datasets, respectively. Furthermore, the proposed ACE-BET is efficient, and achieves the fastest inference speed among various methods being tested.

To solve the matching problems caused by the large intensity difference between the multi-source images and the nonlinear radiation distortion, we present a multi-source image matching approach that considers the orientation of the phase sharpness. First, the scale-space of the image pyramid was constructed, and the phase consistency of the image frequency domain was solved to obtain the maximum moment feature, and the KAZE operator is used to extract the feature points. Next, the Log-Gabor even symmetric filter was used to perform Fourier transform, and the improved local phase sharpness feature and phase orientation feature were constructed respectively to replace the gradient amplitude and gradient direction feature of the image. A descriptor of local phase sharpness orientation (LPSO) was established through the log-polar description template, and finally the Euclidean distance was used to measure the similarity to obtain the corresponding points. Multiple sets of typical multi-source images were used as data sources, and the LPSO algorithm was compared with SIFT, LGHD, RIFT and HAPCG algorithms in datasets. Evaluation of the algorithm performances indicates that the proposed method is more accurate and robust in the task of multi-source image matching.

At present, the problem of the personal information security of apps has attracted widespread attention from the government and society, and there is an urgent need for multiple subjects to participate in the regulation together. Based on this, we first considered the information feedback from users. Secondly, we constructed an evolutionary game model of three-party collaborative supervision among local governments, app distribution platforms, and users. Then the stability of the equilibrium points and their stability conditions are analyzed based on Lyapunov’s first law. Finally, MatlabR2021a software is used to analyze the influence of key parameter changes on the strategy choice of game players. The conclusions are as follows: (1) Increasing the penalty amount for local governments and app distribution platforms, increasing the reputation value premium or reputation loss from users’ feedback, and increasing synergistic benefits can promote local governments to adopt the strategies of supervision and app distribution platforms to adopt the strategies of review. (2) Reducing the cost of supervision for local governments can promote their adoption of supervisory strategies, and reducing the cost of review for app distribution platforms can promote their adoption of review strategies. (3) Reducing users’ feedback cost not only improves users’ enthusiasm to participate in supervision, but also promotes local governments and app distribution platforms to adopt positive strategies. (4) When the coefficient of synergistic benefit distribution is more even, it is more favorable for local governments and app distribution platforms to adopt positive strategies. Based on the simulation results, we propose some feasible strategies.

This work carried out the experimental investigation and model analysis on a GaN electrostatic discharge (ESD) clamp, which features a discharge high electron mobility transistor (HEMT) and a voltage divider formed by a lateral field effect rectifier (L-FER) chain and a resistor in series. Firstly, it is found through the experimental investigation, the turn-on voltage ( $V_{\mathrm {T}}$ ) of GaN ESD clamp’s static current is increased with the increase in L-FER’s number and the decrease in resistor’s value. Then, an analytical model is proposed to analyze the dependence of $V_{\mathrm {T}}$ on L-FER’s number and resistor’s value. And the analytical model exhibits a good consistency with the experimentally measured results, which confirms the validity of the presented analytical model. Lastly, the dependence of GaN ESD clamp’s transmission line pulsing characteristics on the L-FER’s number and resistor’s value is also analyzed. Similarly, GaN ESD clamp’s triggering voltage is also increased with the increase in L-FER’s number and the decrease in resistor’s value. The experimental investigation and model analysis of this work can be used to direct the design of GaN ESD clamp, and help to save a lot of manpower, material resources and time costs.

A micro vibration signal extraction method based on deep neural network is proposed. Rough surface of vibrating object modulates the illuminating laser wave front and generates speckle pattern, which is recorded by a linear array CMOS and preprocessed and input into a 16-layer convolution neural network (CNN) trained with specially prepared data. The optical experimental setup is analyzed to fulfil the temporal and spatial Shannon sampling theorem. The output audio signals are evaluated with standard algorithms and show enhanced segmental SNR and intelligibility. The effects of different input audio types and quality of raw audio signals are investigated, and the results show that the neural network is robust to the input. The CNN structure is optimized and the results show the performance decrease with the reduction of convolution layers. The performances of three popular deep neural networks are compared and the performance of CNN is better.

The ship on the sea has the phenomenon of swing under the combined action of sea waves, wind and its stability system in the case of synthetic aperture radar imaging, which makes it difficult to focus. In order to correct the two-dimensional spatial variance and accomplish the imaging of swing ship, this article proposes a new imaging algorithm based on spatial variant phase gradient autofocus (PGA). First, the entropy difference is used to determine the specific position of the ship. Then, the ship image is extracted and segmented by a variable-size segmentation strategy. PGA is used to focus each subimage, and a well-focused ship image is obtained. The results of simulation experiments and real data show that the algorithm has good focusing performance.

The motor imagery (MI) based brain-computer interfaces (BCIs) have been proposed as a potential physical rehabilitation technology. However, the low classification accuracy achievable with MI tasks is still a challenge when building effective BCI systems. We propose a novel MI classification model based on measurement of functional connectivity between brain regions and graph theory. Specifically, motifs describing local network structures in the brain are extracted from functional connectivity graphs. A graph embedding model called Ego-CNNs is then used to build a classifier, which can convert the graph from a structural representation to a fixed-dimensional vector for detecting critical structure in the graph. We validate our proposed method on four datasets, and the results show that our proposed method produces high classification accuracies in two-class classification tasks (92.8% for dataset 1, 93.4% for dataset 2, 96.5% for dataset 3, and 80.2% for dataset 4) and multiclass classification tasks (90.33% for dataset 1). Our proposed method achieves a mean Kappa value of 0.88 across nine participants, which is superior to other methods we compared it to. These results indicate that there is a local structural difference in functional connectivity graphs extracted under different motor imagery tasks. Our proposed method has great potential for motor imagery classification in future studies.

It is difficult to identify optimal cut-off frequencies for filters used with the common spatial pattern (CSP) method in motor imagery (MI)-based brain-computer interfaces (BCIs). Most current studies choose filter cut-frequencies based on experience or intuition, resulting in sub-optimal use of MI-related spectral information in the electroencephalography (EEG). To improve information utilization, we propose a SincNet-based hybrid neural network (SHNN) for MI-based BCIs. First, raw EEG is segmented into different time windows and mapped into the CSP feature space. Then, SincNets are used as filter bank band-pass filters to automatically filter the data. Next, we used squeeze-and-excitation modules to learn a sparse representation of the filtered data. The resulting sparse data were fed into convolutional neural networks to learn deep feature representations. Finally, these deep features were fed into a gated recurrent unit module to seek sequential relations, and a fully connected layer was used for classification. We used the BCI competition IV datasets 2a and 2b to verify the effectiveness of our SHNN method. The mean classification accuracies (kappa values) of our SHNN method are 0.7426 (0.6648) on dataset 2a and 0.8349 (0.6697) on dataset 2b, respectively. The statistical test results demonstrate that our SHNN can significantly outperform other state-of-the-art methods on these datasets.

Accurately predicting patient expenditure in healthcare is an important task with many applications such as provider profiling, accountable care management, and capitated medical payment adjustment. Existing approaches mainly rely on manually designed features and linear regression-based models, which require massive medical domain knowledge and show limited predictive performance. This paper proposes a multi-view deep learning framework to predict future healthcare expenditure at the individual level based on historical claims data. Our multi-view approach can effectively model the heterogeneous information, including patient demographic features, medical codes, drug usages, and facility utilization. We conducted expenditure forecasting tasks on a real-world pediatric dataset that contains more than 450,000 patients. The empirical results show that our proposed method outperforms all baselines for predicting medical expenditure. These findings help toward better preventive care and accountable care in the healthcare domain.

Mining individual emotions from the data is extremely challenging work in many fields, such as opinion monitoring, business decisions, and information prediction. In previous studies, many studies recognized texts containing positive, negative and neutral sentiments, but few studies focus on emotions according to multiple categories, such as happiness, sadness, anger, fear, disgust and surprise. In this paper, we propose an individual emotion-identifying model called semantic emoticon emotion recognition (SEER). First, we divide the input short text into four categories by using the emotion dictionary and emoticons. Second, we combine a bidirectional gated recurrent unit (Bi-GRU) network with an attention mechanism to capture the emotion vectors of input text from the word aspect. Third, we construct an emoticon distribution model to obtain the emotion vectors in a large quantity of social network data. Fourth, according to different types of input short texts, we select different fusion weights to fuse the emoticon emotion features in text and the semantic emotion features of the texts. Finally, we classify the short text emotions into six categories depending on the final emotion vector. The experimental results show that the SEER is an effective method for improving the accuracy of emotion recognition. In addition, our proposed approach achieves the highest accuracy of 86.35% in emotion recognition, which indicates that the accuracy of our proposed approach achieves improvements of 12.66%, 14.56%, 5.38%, 4.48%, 4.9%, 2.68%, and 3.17% compared to SVM, WL, LSTM, BiLSTM, GRU, Bi-GRU, and CNN+BiLSTM, respectively.

We report the development of a novel soft body touch sensor capable of measuring contact force along its length. The sensor can be embedded in soft objects and on curved surfaces for smart furniture, interactive electronics, and general robotics applications. The sensor uses pneumatic transduction principle and achieves both high sensitivity and impact loading tolerance. The development is motivated by the needs of consumer electronics industries for smart furniture and interactive toys. The current work fills a gap of existing soft-bodied touch sensors. The main sensor body is completely polymeric and contains no conductive or semiconducting material elements. It exhibits high pass frequency response to reject common mode contact forces or shape variations. The central question addressed by this paper is dynamic behavior of this sensor under normal and impact touch force. In this study we developed a theoretical model for the sensor output and validated with experimental measurements. We modeled sensor behavior under two operational regimes of different unloading speed. Depending on the rate of change, the sensor could be in force limited relaxation mode or elastic relaxation (free relaxation) mode. Measurements have been performed to capture time constants of occupant behavior in a smart cushion application.

It is significant for restoration and protection of natural resources and ecological services in coastal wetlands to map different land cover types with satellite remote sensing data. Considering difficulties of wetland species classification, hyperspectral images (HSIs) with high spectral resolution and multispectral images (MSI) with high spatial resolution are considered to achieve complementary advantages of multisource data. An effective approach, named as multistream convolutional neural network, is proposed to achieve fine classification of coastal wetlands. First, regression processing is adopted to make chaotically scattered coastal wetland data more compact and different. Second, through appropriate feature extraction and feature fusion strategies, high-level information of multisource data in regression domain is fused to distinguish different land cover. Experiments on GF-5 HSIs and Sentinel-2 MSIs are carried out in order to validate the classification performance of the proposed approach in two coastal wetlands of research value in China, i.e., Yellow River Estuary and Yancheng coastal wetland. Experimental results demonstrate the effectiveness of the proposed method compared with the state-of-the-art methods in the field, especially when the number of sample size is extremely small.

Many deep learning models for fault diagnosis have not considered the prior diagnosis knowledge of the rolling bearing. Moreover, some measuring locations cannot collect adequate data to diagnose due to equipment size or installation space problems. This paper proposes a wavelet convolutional deep transfer learning model for rolling bearing fault detection on cross-measurement points. A new convolution layer includes a redesigned convolution kernel and a new energy pooling layer. The convolution kernel is designed based on wavelet construction for mining time-frequency characteristics. The energy pooling layer is proposed to extract the energy of different frequency bands. The associated location’s fault information has been transferred to domain knowledge to enhance the target features. Different domain features have adaptive matches based on multiple kernel variants of maximum mean discrepancy. The experimental results demonstrate that the accuracy of fault detection can reach 99.73%, and the robustness of the proposed method is also verified.

Snow cover is an important component of the cryosphere. Clouds have a large influence on optical remote sensing satellites when recognizing snow cover. Geostationary satellites, due to their high-frequency observations over coverage areas, can effectively compensate for the drawback of snow cover recognition from polar orbit optical satellites under cloud-covered conditions. However, past geostationary satellites have relatively few band settings to produce sensitive factors for snow cover recognition. The FY-4A Advanced Geostationary Radiation Imager (AGRI) satellite has the advantage of high temporal resolution with a wealth of bands, which highlights its potential in reducing the impact of clouds and accurately obtaining snow cover information. Based on the advantages of FY-4A AGRI data and the flow characteristics of clouds, we developed an improved maximum brightness temperature image synthesis algorithm, which can greatly reduce the probability of cloud and snow cover misclassification. Combining the features of FY-4A AGRI data, we reorganized the snow cover recognition factor and developed a new snow cover recognition method. The results show that the proposed method can reduce cloud cover by 57.172% compared with MOD10A1 data. After evaluating the proposed method using meteorological ground observation datasets and MOD10A1 data, we found that the overall accuracy of the proposed method can reach 94.11% and 98.55%, respectively, and the F-score (FS) can reach 73.05% and 85.40%, respectively.

The rapid and accurate identification of spider sex is the first step in spider image recognition. The traditional artificial method used to identify the sex of mature spiders is mainly based on their genital structures (male palps or female epigynum) and highly dependent on the professional background of the identifiers. This article uses computer-based deep learning and transfer learning to identify the sex of spider, explores the design and application of convolutional neural networks in deep learning for spider sex recognition from images, and establishes a neural network model that displays excellent performance in experiments. In addition to optimizing the network model, we select appropriate hyperparameters to improve the accuracy of recognition and reduce the influence of human factors in the identification process. Through a comparison of multiple sets of experiments based on existing sample data collected in the laboratory, we find that the transfer learning method based on Xception can obtain better prediction accuracy than ResNet-152. After data augmentation, the optimization of a combined activation function and the fine-tuning of frozen layers, the prediction accuracy reaches 98.02%, and for an actual measurement of independent samples, the recognition accuracy reaches 92.38%. Therefore, the proposed method can basically replace manual identification and provide a reference for the artificial intelligence-based identification of spider species. Additionally, the model results indicate that male and female dimorphism may exist beyond the non-genital characteristics of spiders.

Expression recognition in the wild is easily distorted by nonfrontal and asymmetry faces. In nonfrontal faces, some areas are compressed and distorted. Even after frontalization, these compressed areas may still be blurred and distort expression recognition. Additionally, asymmetrical expressions are common on half or local face areas and produce incorrect expression features. Therefore, this paper presents a half-face frontalization and pyramid Fourier frequency conversion method. Despite the location, range and intensity of incorrect expressions in nonfrontal faces being unknown, according to discrete Fourier transform, it can be proven that the frequency band of the correct expression is much larger than that of incorrect expression on the same face. This can be taken advantage of by pyramid frequency conversion, which is designed based on Fourier frequency conversion. It can adjust incorrect expression frequency in multiscales to take them out off the band-pass of the convolution operations of deep learning and be eliminated completely, whereas correct expression information is reserved. Thus, expressions can be recognized effectively.

Traditional methods for change detection in synthetic aperture radar images have difficulty in obtaining results from the generated differential image (DI) owing to speckle noise. In recent years, many deep learning-based methods have emerged because of their outstanding anti-noise and self-learning ability. However, they are limited by the requirement of abundant high-precision labels. Therefore, in this article, we propose a novel unsupervised method based on improved non-subsampled shearlet transform (NSST) and multi-scale feature fusion convolutional neural network for change detection. First, this method improves the traditional NSST algorithm and proposes a novel pseudo-label generator to obtain more pseudo-labels with higher confidence. It is noteworthy that the more accurate the pseudo-labels are, the better the change detection results will be. Second, this method designs a multi-scale feature fusion block in the network to make the feature images contain more complete information and reduces the number of pooling layers to avoid losing feature image details. The main idea of this method is to eliminate the step of generating the DI and directly obtain results from the original images. The theoretical analysis and final results conducted on three real datasets prove its validity. Furthermore, to verify the generality and potential of the proposed method, we apply it to the cross-region change detection and compare it with the supervised method, which achieve satisfactory results.

Facial Expression Recognition (FER) has long been a challenging task in the field of computer vision. Most of the existing FER methods extract facial features on the basis of face pixels, ignoring the relative geometric position dependencies of facial landmark points. This article presents a hybrid feature extraction network to enhance the discriminative power of emotional features. The proposed network consists of a Spatial Attention Convolutional Neural Network (SACNN) and a series of Long Short-term Memory networks with Attention mechanism (ALSTMs). The SACNN is employed to extract the expressional features from static face images and the ALSTMs is designed to explore the potentials of facial landmarks for expression recognition. A deep geometric feature descriptor is proposed to characterize the relative geometric position correlation of facial landmarks. The landmarks are divided into seven groups to extract deep geometric features, and the attention module in ALSTMs can adaptively estimate the importance of different landmark regions. By jointly combining SACNN and ALSTMs, the hybrid features are obtained for expression recognition. Experiments conducted on three public databases, FER2013, CK+, and JAFFE, demonstrate that the proposed method outperforms the previous methods, with the accuracies of 74.31%, 95.15%, and 98.57%, respectively. The preliminary results of Emotion Understanding Robot System (EURS) indicate that the proposed method has the potential to improve the performance of human-robot interaction.

The high-head pumped storage power station (PSPS) has complex working conditions and severe transient processes. Under load rejection conditions, the turbine speed and the flow channel pressure will threaten the unit. By adjusting the wicket gate closing law (WGCL), we can effectively alleviate the adverse effects of the transient process. In this study, a full flow channel refined model of a high-head PSPS was constructed, and the influence mechanism of load rejection conditions on key parameters was analyzed by tracking the operating trajectory on the S characteristic curves and simulating the transient flow field inside the pumped turbine by three-dimensional computational fluid dynamics (3D-CFD). Given the contradiction between different parameters in the wicket gate closing process, a third-generation non-dominated genetic algorithm based on reference point selection (NSGA-III) was introduced, and a high-dimensional multi-objective WGCL optimization model was constructed. We compared parameter changes in the two-phase and three-phase WGCL optimized elite solutions and explained the rationality of adopting the three-phase WGCL in high-head PSPSs. By proposing an improved three-phase WGCL, a new scheme—the most suitable WGCL under load rejection—was obtained through precise optimization. The research embodied the excellent effect of the heuristic multi-objective (MO) optimization algorithm based on the Pareto strategy in solving the operation problem of the large fluctuation transient process of hydraulic machinery. It revealed the theoretical basis of engineering adjustment for the safe production and operation of hydropower.

In this work, the effect of gate voltage sampling interval ( $\Delta \text{V}_{\mathrm {gs}}$ ) and various difference approximation methods on extracting the value of subthreshold swing (SS) have been investigated in detail for InGaZnO thin film transistor (TFT). The values of extracted SS under the $\Delta \text{V}_{\mathrm {gs}}$ altered from 1 V to 0.03125 V for the same InGaZnO TFT exhibit a large disparity and the minimum value of SS with $\Delta \text{V}_{\mathrm {gs}}$ of 0.03125 V has 3~4 times less than the $\Delta \text{V}_{\mathrm {gs}}$ of 1 V by the same approximation method. Furthermore, the impact of various difference approximation methods on extracting SS have also been involved. Moreover, an empirical formula is proposed to estimate the proper sampling interval for accurate approximating the SS. The proposed phenomenon and method in this work could be a meaningful reference for extracting the SS and other parameters in InGaZnO TFTs.

Parallel magnetic resonance (MR) imaging is an important acceleration technique based on the spatial sensitivities of array receivers. The recently proposed Parallel low-rank modeling of local k-space neighborhoods (PLORAKS) approach uses the low-rank matrix model based on local neighborhoods of undersampled multichannel k-space data for reconstruction purposes. The joint total variation (JTV) regularization term was then combined with the PLORAKS model to improve the quality of reconstructed images. To further improve the quality of parallel MR imaging, we propose combining adaptive transform learning and joint sparsity with the PLORAKS model to obtain two algorithms, and reconstruction problems are solved by using the alternating direction method of multipliers (ADMM) and conjugate gradient techniques. The experimental results show that the two proposed algorithms can achieve higher performance than the PLORAKS algorithm and the PLORAKS-JTV algorithm with the JTV regularization term in terms of the signal-to-noise ratio (SNR), normalized root mean square error (NRMSE), high-frequency error norm (HFEN), and structural similarity index measure (SSIM).

The traditional energy system has the disadvantages of high energy consumption and high emission. Through multiple energy networks interconnection and flexible energy conversion, the regional integrated energy system (RIES) can satisfy various energy demands and achieve clean and efficient energy utilization. However, the demand and supply relationship of RIES is complex. Without a reasonable energy supply structure, it's difficult for RIES to give full play to the economic and efficient operation. At the same time, the multi-energy trading and service mechanisms are still unclear. The promotion of RIES brings challenges and opportunities to energy sector. In order to address the issues in the promotion of RIES, a day-ahead energy trading strategy of RIES considering energy cascade utilization is proposed in this paper. Firstly, the relationship between energy suppliers (ESs), integrated energy service provider (IESP) and energy users (EUs) is clarified. And their benefits are clarified in the energy trading strategy. Furthermore, the day-ahead energy trading model, which is a bi-level optimization model is built. In the upper model, the IESP purchases electric energy and natural gas from ESs at wholesale prices and set energy retail prices. According to energy cascade utilization theory, the IESP make decisions on the energy retail plan and energy scheduling. In the lower model, optimal energy consuming plan are given for EUs. In this model, energy grades of energy production, conversion and storage devices output are classified and the energy grade demands of different energy loads are divided. And the energy cascade utilization trading model is simulated based on actual data. The simulation results show that the trading strategy is feasible and the application of energy cascade utilization theory can further improve the energy utilization efficiency of RIES.

Recently, tag-to-tag (T2T) backscattering technique in a passive RFID system has received broad attention due to its superiority for large-scale network applications. If used to implement a Network of Tags, use of T2T communication allows inherent communication parallelism, thus supporting orders of magnitude larger capacity than centralized RFID reader-based systems. To unleash the potential of T2T communication, turbo backscattering operation enables the implementation of a multi-hop network of tags, which supports larger network coverage of a Network of Tags. However, due to asymmetric communication links and interferences among tags' transmissions in such a T2T backscattering based network, the routing protocol design has become one of the main technical challenges, especially for large-scale networks. Furthermore, the computation time of T2T routing protocols increases exponentially with the number of tags, greatly limiting the practicability of such large-scale backscattering networks. In this paper, we present the design of a Network of Tags model to address these challenges, and we propose novel routing protocols for three distinct types of tags with different hardware capabilities. To address the issue of computational processing time of the routing protocol for large-scale T2T networks, we propose a new scheme with linear time complexity. We evaluate and compare the performance of the proposed protocols, as well as investigate the impact of network parameters on the performance.

3D visualization of digital human becomes a key tool for the medical visualization, especially for medical education. Web3D technology has been commonly applied in this field. However, the quality of rendering is not expected for the medical purpose. Nowadays, global illumination (GI) map is an efficient tool for real-time lighting and shadow rendering. On the cloud baking server, a large number of rendered GI maps are generated under variety of configuration in the scene on the Web3D interface end. GI tree works on organizing these baked maps for reusing in the Web3D client. Meanwhile, it dispatches the existing baked maps directly in the case that the viewpoint appears in the duplicate positions in the Web3D client. This is the main stream solution of the cloud pre-rendering. However, it is a challenge to store and manage excessive rendered maps. In this paper, we propose a light-weight collaborative machine learning method for lighting and shadow rendering in medical applications. In this system, the conditional generative adversarial networks (GAN) works for generating the GI map instead of finding out the similar from number of stored maps, and we propose structure-aware 3D image warping method to improve the system performance. The experiments demonstrated that our proposed system not only guarantees the resolution of the GI map in the Web3D client, but also significantly reduces the rendering computational needs so as to improve the system performance.

Flicker noise contributed by active mixers usually compromises the overall sensitivity level of direct-conversion receivers (DCRs). In this article, a qualitative quasi-analytical model has been developed to in-depth explain the flicker noise mechanism existing in the switched-gm active mixer in which an improved current source switch is presented for high mixer common-mode rejection ratio (MCMRR). The built model simply explains how frequency translations take place within the mixer. Compact equations are derived to estimate the flicker noise contribution of individual stages at the output. Simulations validate the accuracy of the predictions, and the dependence of flicker noise on local oscillator (LO) slope ratio, period, and other circuit parameters. The high-frequency limitation of the mixer is further estimated by investigating into the tail parasitic capacitance charging and discharging behavior. Furthermore, a switched-gm pMOS mixer prototype with low flicker noise is implemented in a 0.18-μm CMOS process. It operates at an RF input frequency of 1 GHz and provides a maximal conversion gain of 12.9 dB and an NF of 11.4 dB while a flicker noise corner of 220 kHz and an IIP3 of 3.6 dBm are measured, respectively. The mixer core only consumes 3.3 mW from a 1.8 V supply.

While electromagnetic forming is widely used for aluminum alloy materials, the non-uniformity of plate radial deformation is still a key issue that restricts the wide development of this technology. This article presents a new electromagnetic-forming topology. The proposed topology comprises outer ring double-coil plates along with an inner loop driving coil of a small number of turns whose geometrical parameters are optimized to adjust the induced eddy current of the plate. Numerical simulation results show that the presence of an inner ring driving coil shapes the magnetic field in a manner that the radial electromagnetic force is concentrated close to the edge of the die which affects the center of the plate. With optimum design for the geometrical parameters of the proposed inner ring driving coil, the electromagnetic force distribution can be adjusted over the plate, resulting in an improved plate deformation with an increase in the maximum uniform deformation area from 24mm to 65mm. Results also show that the proposed inner and outer ring double coil loading method possesses superior advantages over the traditional flat spiral coil loading method.

Improving the forming efficiency at a reasonable cost has been the main challenge in the electromagnetic forming technology. Towards this aim, this paper proposes a simultaneous cost-effective double-sheet electromagnetic forming method using single driving coil. In order to highlight the features of the proposed technique, comparison of the performance of five electromagnetic forming models is presented. These models include a no-sheet model, single-sheet stationary model, single-sheet moving model, double-sheet stationary model and double-sheet moving model. For each model, electromagnetic force distribution and forming efficiency of the workpiece are analyzed. Furthermore, the influence of the pulse current width of the driving coil on the electromagnetic forming efficiency of the plate is investigated for all models. Results show that for the same system discharge parameters, electromagnetic forming efficiency of double-sheet is 21.14% whereas the single-sheet electromagnetic forming model only achieves 14.68% efficiency. Thus, the double-sheet electromagnetic forming technology can improve the low forming efficiency of a single-sheet electromagnetic forming method and at the same time, the method can form two workpieces simultaneously which makes it more cost effective than the single-sheet method.

Uneven axial deformation is the current main issue of the tubes electromagnetic bulging technology which calls for innovative and cost-effective solutions. This paper presents a new method to overcome the above-mentioned issue based on employing a magnetic field shaper. The proposed magnetic field shaper is employed in the middle of the pipe and the driving coil. Induced eddy current control is realized by changing the structural parameters of the magnetic field shaper; thereby regulating the electromagnetic force distribution. The effectiveness of the proposed method is investigated through numerically comparing the electromagnetic performance of two EMF setups; traditional coil-based loading model and magnetic field shaper driving coil-based loading model proposed in this paper. The influence of the magnetic field shaper structural parameters on the tube electromagnetic forming performance is examined for optimum structure design. Results reveal that the concave profile of the radial electromagnetic force in the traditional loading method can be effectively improved using the proposed loading in this paper. The introduction of a magnetic field shaper not only changes the induced eddy current distribution, but also regulates the electromagnetic force profile in the axial-direction of the pipe which effectively enhances the uniform range of the pipe electromagnetic bulging, the maximum uniform deformation area is increased from 9mm to 35mm.

The MOD10A1 daily snow cover product offers a fine temporal resolution, which is particularly important for time-critical applications. Although the reliability of this product was mostly verified by ground measurements before use, the inadequate representativeness of point-scale observations may lead to errors in the assessment of MOD10A1. In this study, based on the Google Earth Engine, MOD10A1 was thoroughly evaluated by a total of 20 131 binary snow maps generated from 30 m Landsat imagery for the 2000-2016 snow seasons in three typical snow regions across China. The three typical snow regions included Northern Xinjiang, Northeast China, and the Tibetan Plateau. In general, the accuracy of MOD10A1 under clear sky conditions is 86.5% based on ground observations from meteorological stations and 90.3% based on Landsat imagery. Compared with ground measurements, snow cover derived from Landsat images is in better agreement with MOD10A1 due to a better spatial match. The error analyses also indicate that 1) compared with land cover, topography has larger effects on the accuracy of MOD10A1 over the three snow regions with statistically significant negative relationships, particularly in Northern Xinjiang (R = -0.604, P