Search

We revisit the design of convolutional kernels in lightweight convolutional neural networks, and inspired by the recent advances in RepLKNet, we design a Variable Kernel Convolutional Network module VarKNet, which solves the problem of the imbalance between depthwise convolution and pointwise convolution in the case of depthwise separable convolution when the network width is large, and enriches the model’s receptive field. The VarKNet module adopts a multi-branch structure during training and is re-parameterized and fused into a single-path structure during inference to maintain the strong expressive ability of the model and improve the inference speed. In order to further enhance the information exchange between channels, VarKNet adds channel shuffling in the fused branches. Built on VarKNet, we designed a large-scale face recognition network VarKFaceNet. VarKFaceNet achieved A great achievement of 99.5% accuracy on the LFW dataset with 0.7M parameters and 0.24 GFLOPS. At the same time, the measured speed on the NVIDIA Jetson Nano platform is 159 times, 4.2 times, and 2.4 times that of ResNet-50, EfficientNet, and MobileFaceNet, respectively. VarKFaceNet excels in balancing speed and accuracy and is quite suitable for embedded devices with limited resources.

This paper tackles the exact consensus issue in nonlinearly parameterized multi-agent systems with unknown control directions, employing an adaptive learning control approach. The nonlinearly parameterized function is effectively compensated using neural networks and Fourier cardinal expansions. A cooperative consensus protocol is meticulously designed to achieve exact consensus through the application of iterative learning laws. The convergence of the consensus algorithm is rigorously analyzed using a composite energy function. Additionally, the robustness of the consensus algorithm is examined by introducing input disturbances to the controlled subsystem and bounded inputs to the leader. The scalability of the consensus protocol is explored through the context of formation control problems. Finally, the effectiveness of the constructed learning protocol is illustrated by simulations.

To improve the assimilation accuracy of satellite microwave observations, it is necessary to conduct precipitation detection before assimilation and exclude the data contaminated by the scattering of hydromorphic particles. The induction of microwave scattering is closely related to the effective radius of hydromorphic particles, and the motion state of hydromorphic particles at different scales can directly affect the internal dynamic structure of clouds, which will lead to precipitation. To further analyze and validate the impact of microwave scattering in a precipitation environment on the assimilation of satellite microwave soundings within the temperature-sensitive band, this article utilized the on-orbit observations of Microwave Temperature Sounder-III (MWTS-III) and Medium Resolution Spectral Imager onboard the Fengyun-3E (FY-3E) satellite to achieve multispectral fusion observation and precipitation detection over oceans, combined with the vertical dynamic characteristics of global numerical forecast in the cloud environment of China Meteorological Administration (CMA) Global Forecast System (CMA-GFS). By conducting the dynamic analysis of the hydrometeors in the cloud, the maximum possible effective particle radius of hydromorphic particles under the no-phase-transition condition and internal vertical motion is obtained. The consistency of the cloud liquid water path (CLWP) retrieved by MWTS-III and multispectral fusion precipitation detection is evaluated subsequently. The results show that the distribution of hydrometeors in the CMA-GFS forecast is broadly in line with the CLWP distribution obtained from the MWTS-III retrieval, regardless of Mie scattering or Rayleigh scattering. However, compared with Mie scattering, the weak convection of CMA-GFS shows better adaptability in the Rayleigh environment.

Traditional grid analytics heavily rely on accurate power system models, especially dynamic ones for generators, controllers, and loads. However, obtaining comprehensive models is impractical in real operations due to inaccessible parameters and consumer privacy. This necessitates dynamic equivalencing for unknown subsystems, which employs physics-informed machine learning and neural ordinary differential equations (ODE-NET) to preserve dynamic behaviors post-disturbances. The contributions include: 1) A neural dynamic equivalence (NeuDyE) formulation enabling continuous-time, data-driven dynamic equivalence, eliminating the need for acquiring inaccessible system details; 2) Introduction of Physics-Informed NeuDyE learning (PI-NeuDyE) to actively control NeuDyE’s closed-loop accuracy; 3) Driving Port NeuDyE (DP-NeuDyE), a practical application of NeuDyE, reducing the number of inputs required for training. Extensive case studies on the 140-bus NPCC system validate the generalizability and accuracy of both PI-NeuDyE and DP-NeuDyE. These analyses cover various scenarios, including limitations in data accessibility. Test results demonstrate the scalability and practicality of NeuDyE, showcasing its potential application in ISO and utility control centers for online transient stability analysis and planning purposes.

Due to the interference of multiplicative speckles, it is challenging to accurately detect changes in polarimetric synthetic aperture radar (PolSAR) images. Convolutional neural network has been proven to learn rich local features from PolSAR data. However, convolution kernels with limited receptive fields have difficulty in exploring global information. Here, a dual attention-based global-local feature extraction network (DA-GLN) is developed for unsupervised PolSAR image change detection (CD). First, we use fuzzy C-means clustering on the enhanced Shannon entropy difference image to automatically generate pseudolabeled samples required for unsupervised CD. Subsequently, our DA-GLN utilizes a deep residual shrinkage network that incorporates channel attention mechanisms and soft-thresholding to weaken the influence of speckle noise and capture local features. Meanwhile, a pooling-based vision transformer is adopted in DA-GLN to extract global features, which introduces pooling layers to complete self-attention spatial information interaction with higher efficiency than the visual transformer. Furthermore, a global-local constraint feature fusion strategy is designed to effectively fuse local and global features. Finally, we employ a feature constraint-focal loss function including feature constraint loss and focal loss as the objective function of DA-GLN. Specifically, the feature constraint loss function is constructed to eliminate feature redundancy and fully exploit the complementarity between features, while the focal loss function is introduced to balance the impact of the inequality between changed and unchanged samples on the network. Numerical experiments on five real spaceborne PolSAR datasets demonstrate that our DA-GLN is more competitive than other state-of-the-art methods.

Voltage sag is one of the reasons that hinders the high-quality and efficient development of high-tech manufacturing users. Custom power device (CPD) is one of the effective ways to solve voltage sag. The reason why most high-tech manufacturing users are hesitant to invest CPD is that the economic benefits of CPD are vague. To solve this problem, a new premium power service (PPS) for CPD based on trial-purchase model is proposed. By trialing CPD, users can choose to purchase or cancel the devices after trialing, so as to improve the enthusiasm of users to invest premium power and ensure the smooth progress of PPS. According to the operation process of the proposed service, an optimization model of service scheme is established with the goal of maximizing the interests of power utility and user, to optimize the return ratio of users’ mitigation benefits and the length of trial service. The revenue distribution model based on Game theory (Nash negotiation) is established to obtain the revenue distribution strategies of power utility and devices company respectively. The feasibility of proposed service is verified by a semiconductor user in China. The trial-purchase premium power service could be replicable and applicable, which provides a new choice both for power utility and users.

Remote sensing semantic change detection (SCD) involves extracting information about changes in land cover/land use (LCLU) within the same area at different times. This issue is of crucial significance in many Earth observation tasks, such as precise urban planning and natural resource management. However, the current methods primarily focus on spatial feature extraction, lacking awareness of temporal features. Consequently, there are challenges in extracting change features, making distinguishing intraclass and interclass differences difficult. This also contributes to pseudochange, posing challenges for SCD tasks. To overcome the limitations of existing methods, we present a dual-dimension feature interaction network (DFINet) for SCD. First, to enhance the assessment and perceptual abilities related to intraclass and interclass differences, we introduce a temporal difference feature enhancement (TDFE) module. This module comprehensively captures features from the temporal dimension. Then, to address the interrelation between multitemporal and multilevel features, we investigate the feature selection interaction (FSIA) and interaction attention modules (IAM), which enable multidimensional deep fusion and interaction of change features. This enhances the capacity for information transfer and integration among the features within multitemporal remote sensing images (RSIs). The experimental results demonstrate that, compared to existing methods, the proposed architecture achieves a significant improvement in accuracy. Additionally, the design enhancements added to DFINet boost the practicality of remote sensing SCD, underscoring its substantial research value.

Vision transformer (ViT) provides new ideas for polarization synthetic aperture radar (PolSAR) image classification due to its advantages in learning global-spatial information. However, the lack of local-spatial information within samples and correlation information among samples, as well as the complexity of network structure, limit the application of ViT in practice. In addition, dual-frequency PolSAR data provide rich information, but there are fewer related studies compared to single-frequency classification algorithms. In this article, we adopt ViT as the basic framework, and propose a novel model based on mixed patch interaction for dual-frequency PolSAR image adaptive fusion classification (PolSAR-MPIformer). First, a mixed patch interaction (MPI) module is designed for the feature extraction, which replaces the high-complexity self-attention in ViT with patch interaction intra- and intersample. Besides the global-spatial information learning within samples by ViT, the MPI module adds the learning of local-spatial information within samples and correlation information among samples, thereby obtaining more discriminative features through a low-complexity network. Subsequently, a dual-frequency adaptive fusion (DAF) module is constructed as the classifier of PolSAR-MPIformer. On the one hand, the attention mechanism is utilized in DAF to reduce the impact of speckle noise while preserving details. On the other hand, the DAF evaluates the classification confidence of each band and assigns different weights accordingly, which achieves reasonable utilization of the complementarity between dual-frequency data and improves classification accuracy. Experiments on four real dual-frequency PolSAR datasets substantiate the superiority of the proposed PolSAR-MPIformer over other state-of-the-art algorithms.

Synthetic aperture radar (SAR) target classification methods based on convolutional neural networks (CNNs) are susceptible to overfitting due to limited samples. In addition, the position and deformation variations of SAR targets can also affect the feature extraction capabilities of CNN. To address these issues, this article proposes a novel few-shot learning (FSL) method, called position-aware graph neural network (PA-GNN). The PA-GNN introduces a PAM to enhance the feature representation of the target. Specifically, deformable convolution is incorporated into the embedding network to accommodate the various shapes of SAR targets. Then, the self-attention network is brought in to capture the spatial dependence of any two positions in the feature maps, and the cross-attention network calculates the cross-attention between support and query feature maps to handle the problem of unseen categories. These two attention networks can be exploited to highlight the target regions. Besides, to integrate the features extracted by these two attention networks, an adaptive feature fusion module is designed to obtain discriminative features of the target position. The PA-GNN conducts measurement by constructing multiple layers of GNN and utilizes residual learning to alleviate oversmoothing. Furthermore, we introduce a fused loss to enhance the separability between different categories and the similarity within the same category. Experiments on OpenSARShip and MSTAR datasets demonstrate that the proposed PA-GNN can achieve comparable classification results compared with other FSL algorithms.

Microwave electric field precision measurement based on Rydberg atomic quantum coherence effect has rapidly advanced in recent years, capitalizing on its high sensitivity, broad bandwidth, and traceability. The utilization of distinct Rydberg quantum states facilitates achieving an ultra-wideband measurement response across various microwave frequency bands, offering extensive applications in wireless communication, radar positioning, astronomical observation, and related domains. In this work, leveraging the Rydberg atomic antenna, we achieve simultaneous multi-frequency reception in the C-band (7.91 GHz, 5.106 GHz) and S-band (3.78 GHz) through amplitude-shift keying modulation. Furthermore, we encode RGB information from color image pixels onto 3 carrier frequencies in the C-band and S-band, while optimizing the multi-channel signal amplitude. The pixel information of the color image obtained by atomic demodulation realizes the high-fidelity digital color image transmission, with a root mean square error of 4.806, a mean square signal-to-noise ratio of 1.627, a peak signal-to-noise ratio of 29.998 and a structural similarity index of 0.9589. This study establishes an experimental groundwork for the utilization of Rydberg atomic antennas in ultra-wideband wireless communication and measurement.

The utilization of polarimetric synthetic aperture radar (PolSAR) enables the preservation of more comprehensive target scattering characteristics compared to conventional SAR. Despite the significant advancements in deep learning (DL) technology for single polarimetric SAR ship detection, there remains a scarcity of DL research specifically focused on PolSAR ship detection. This article proposed a lightweight theory-driven network (LT-Net) for combining domain knowledge and DL techniques, which is suitable to detect small targets and reduce the computational complexity. The LT-Net incorporates convolutional block attention module, and employs differential downsampling during the process of high-dimensional feature extraction. The integration of domain knowledge lends each convolution step a distinct physical significance. Due to the limited availability of public datasets, this article presents a public fully polarized ship detection (FPSD) dataset for the first time, which contains 853 Pauli pseudocolor maps (in JPG format) and multilook complex data (in TIF format), with a total of 1714 ship targets from AIRSAR, UAVSAR, and RadarSAT-2. The FPSD encompasses a range of scenarios, with the majority of ship targets being small targets. Validation on FPSD shows that LT-Net exhibits significantly lower time complexity and space complexity compared to four popular target detection networks, Meanwhile, LT-Net achieves the best detection performance in the Pauli pseudocolor maps.

The topology of the distribution network and direction of the power flow will change when distributed generators (DGs) are connected to it, making it difficult to locate faults using conventional techniques like the impedance approach. Aiming at the two-phase short-circuit grounding faults of active distribution networks, this paper proposes a fault location method based on the time difference of the traveling wave modulus. The first step is the proposal of a zero-mode time-of-arrival calibration method for the ideal frequency band through the analysis of the attenuation of zero-mode traveling wave transmission. Next, define the relative wave velocity, research the quantitative relationship between the modulus transmission time difference and the zero-mode and aerial-mode wave velocities, and establish equation constraints between the modulus transmission time difference, relative wave velocity, and transmission distance. Then, time bounds and dynamic inequality constraints that establish relative wave velocities by fitting. Finally, combined with the abnormal data processing strategy, with the goal of minimizing the weighted deviation of the modulus time difference, the particle swarm optimization (PSO) algorithm is used to solve the fault distance. The PSCAD simulation result demonstrates that the method proposed in this paper has the advantages of high accuracy, strong error tolerance, and strong adaptability, and can quickly and accurately locate faults.

Deep learning has achieved satisfactory results in polarimetric synthetic aperture radar (PolSAR) image classification, which requires a large number of labeled samples for training. However, in practice, labeling work is time-consuming and laborious. As a result, an insufficient number of labeled samples will lead to a limited ability of the network to recognize different terrains. To alleviate this problem, we take advantage of labeled and unlabeled samples simultaneously to train the deep learning model and, thus, propose a semisupervised complex network with spatial statistics fusion (SCN-SSF) for PolSAR image classification. First, the semisupervised complex network (SCN) continuously updates the pseudolabels of unlabeled samples during the training of complex-valued CNN, and their errors constitute the regularization term of the objective function, which improves the generalization of the network. As a result, SCN can recognize different terrains more accurately, and the classification has a higher belief. Then, a parameter-free spatial statistics module is built to model neighborhood label interactions based on the product of experts (POEs), thus enhancing contextual smoothness and correcting some misclassifications. Finally, based on the Dempster–Shafer theory, the contextual label information of POE and pixel-level information obtained by SCN are integrated to preserve image structure. Overall, with only a small number of labeled samples, SCN-SSF can accurately identify each terrain and obtain smooth classification while preserving edge information. The effectiveness of SCN-SSF is demonstrated by classifying PolSAR images with a small number of labeled samples.

A beam-steerable wideband reflectarray antenna (WRA) for $C$ -band high-power microwave (HPM) application is presented in this paper. The $C$ -band WRA comprises a multimode conical horn antenna with high power handling capacity (PHC), $25\times25$ special designed antenna elements, a metal frame and a dielectric plate used as a radome. In our previous work, it has been verified that a wideband prototype using this antenna element has a relative bandwidth of 50% (3.3-5.5 GHz) and its beam scanning range reaches ±60° in low power experiments. In this paper, combined with the equivalent circuit model, the wideband and high PHC property of the WRA element is further explained. To demonstrate the excellent performance of the proposed element in HPM applications, we combined the $C$ -band WRA (which was redesigned from the previous wideband prototype) with a well-designed narrowband HPM source operating at 4.3 GHz, and constructed a matching HPM experimental platform. The $C$ -band WRA works at 4.1-4.5 GHz, and the results of low power experiments indicate that it can maintain good performance during beam scanning. Its sidelobe level and cross-polarization level are always less than −15 dB and −22 dB, respectively, while its maximum aperture efficiency reached 68.99% (the corresponding antenna gain is 28.07 dB at 4.3 GHz). On the other hand, the PHC of the WRA element reaches 3.82 GW/ $\text{m}^{2}$ and the $C$ -band WRA reaches 1.24 GW in simulation. HPM verification experiments have been carried out, and the experimental results show that the PHC of the fabricated prototype exceeded the expectation, which proves that the WRA has broad application prospects in HPM fields.

Two-dimensional phase unwrapping (2-D PU) is the process of converting the measured phase into the real phase in interferometric signal processing. Reliable unwrapping results are critical for digital elevation model generation using interferometric synthetic aperture radar (InSAR) and interferometric synthetic aperture sonar (InSAS). The majority of previous research has concentrated on accuracy, whereas the computational efficiency must be taken into account for the interferometric measurement system that requires real-time processing. This article proposes a low-time-consuming algorithm that can accomplish high-precision 2-D PU for this application scenario. The neural network and a new path-based 2-D PU algorithm make up this algorithm. First, the incorrect region in the gradient field is predicted and corrected using the neural network. The output channelwise variance is then calculated and used to generate the quality maps. Finally, to achieve phase reconstruction, the path-based algorithm performs path planning and flooding integral according to quality maps and compensated gradient. This article also provides a recommended data structure implementation to ensure the algorithm's high efficiency. Experimental results using InSAR and InSAS data show that the proposed algorithm is highly efficient and accurate.

Fusing features from different feature descriptors or different convolutional layers can improve the understanding of scene and enhance the classification accuracy. In this article, we propose a hierarchical deep texture feature fusion network, abbreviated as HDTFF-Net, aiming to improve the classification accuracy of high-resolution remote sensing scene classification. The proposed HDTFF-Net can effectively combine the shallow texture information from manual features and the deep texture information by convolutional neural networks (CNNs). First, for deeply excavating the multiscale and multidirectional shallow texture features in images, an improved Wavelet feature extraction module and a Gabor feature extraction module are designed by fully fusing the structural features into the backbone neural network. Then, to make the output texture features more discriminative and interpretative, we incorporate the above texture feature extraction modules into traditional CNNs (Tra-CNNs), and design two improved deep networks, namely Wave-CNN and Gabor-CNN. Finally, according to the Dempster-Shafer evidence theory, the designed Wave-CNN and Gabor-CNN are fused with the Tra-CNN by a decision-level fusion strategy, which can effectively capture the deep texture features by different feature descriptors and improve the classification performance. Experiments on high-resolution remote sensing images demonstrate the effectiveness of the proposed HDTFF-Net, and verify that it can greatly improve the classification performance.

Using remote sensing video to monitor aircraft dynamics is significant for military applications, airport management, and aircraft rescue. The aircraft has a fixed size and obvious characteristics, so it is suitable for correlation filtering. Correlation filtering algorithms can extract features from input data to predict motion trajectories, and the calculation speed of correlation filterings is fast. Hence, such algorithms are advantageous for tracking targets in remote sensing images. In this article, an antidrift multifilter tracker based on a correlation filter and the Kalman filter is proposed for this purpose. This article proposes a temporal consistency-constrained background-aware correlation filter algorithm based on temporal regularization that resists the model drift caused by clouds by using motion information to correct it. Experimental results show that our proposed method shows improved antidrift performance compared with other advanced tracking methods in cases of cloud occlusion and stable performance in other complex conditions. We believe that our model will be helpful for researchers who are interested in object tracking in satellite video, especially for processing satellite video data with cloud occlusion.

The reliability of the transformer differential protection is under threat from multiple interferences, including the inrush condition, the current transformer saturation condition, and the arc fault condition. For improvement, a new method is proposed to discriminate fault conditions and other non-fault conditions. Firstly, the proposed method identifies the unidirectional inrush via unidirectional index, since only it exhibits unidirectional characteristic. Then, the remaining signals are classified by the quartering-based similarity index. The quartering-based similarity index is obtained by quartering segmented fitting to overcome the drawback of overall fitting. Simulation results indicate that the proposed method is able to effectively avoid misjudgments caused by aforementioned interferences. Actual experimental tests, field data tests, and comparison analyses demonstrate its engineering adaptability and superiority.

As resource-constrained Internet-of-Things (IoT) devices become popular targets of various malicious attacks, frequent updates to keep their software up to date are essential to their security. However, state-of-the-art software delivery and payment systems incorporate multiple services in a client-server structure requiring multiple transits of information between client and server, while also creating a wide attack surface. We propose a blockchain-based end-to-end secure software update delivery framework for Internet of Things (IoT) devices, which aims to ensure confidentiality, integrity, availability, efficiency, and audit-ability for verified software delivery, while offloading the cryptographic computation from resource-constrained IoT devices to a decentralized blockchain system. In particular, we leverage Ciphertext-Policy Attribute-Based Encryption (CP-ABE) and design a customized authorization policy to not only ensure that software updates can only be decrypted and installed on authorized IoT devices but also significantly reduce the computational overhead for key generation and key delivery on the manufacturer side. Furthermore, secure and atomic software delivery and payments between IoT devices and the manufacturer are assured through smart contracts. The authenticity of the delivered software is guaranteed by offloading the computation-based signature validation to smart contracts. Compliance audits are satisfied through immutable records on the blockchain’s public ledger, and the smart contracts efficiently guarantee the delivery of software updates in exchange for payment. Security analysis and experiments are performed to compare the proposed framework with state-of-the-art studies and validate its effectiveness.

Versatile Video Coding (VVC) introduces many advanced video coding techniques. These advanced video coding techniques not only improve video compression efficiency and video quality, but also greatly increase coding time and computational complexity. Therefore, this paper uses a fast CU partitioning decision algorithm based on texture complexity and convolutional neural networks (CNN). First, we performed statistics and analysis of CU segmentation patterns for videos in the standard training set, then processed large blocks of CU based on texture complexity. To realize the purpose of fast partitioning, we design different CNN models for CU with different segmentation modes. In the design of CNN model, we use the symmetric convolutional kernel and asymmetric convolutional kernel to extract features in different directions effectively. In the loss function, we used the cross-entropy function to train the CNN model to improve the accuracy of the model. Finally, a double threshold is set in the candidate list to achieve a compromise between coding performance and coding complexity. Experimental results show that, compared to the VTM10.0 anchoring algorithm, our fast scheme, in terms of encoding time, decreases by 55.90% and BDBR increases by 1.79%; our moderate scheme, in terms of encoding time, decreases by 47.90%. BDBR increases by only 1.29%.

In this paper, a pipelined frequency-modulated continuous-wave (FMCW) radar baseband processor applied to real-time applications is proposed and implemented in 40-nm CMOS technology. The FMCW radar signal processing time is analyzed according to the system specifications. On the basis of the theoretical analysis and systematic considerations, a pipelined baseband architecture with internal single-port static random access memory (SRAM) is employed. The baseband processor is mainly composed of two-dimensional fast Fourier transform (2D-FFT), two-dimensional constant false alarm rate (2D-CFAR), digital beam-forming (DBF), and memory control modules. The 2D-FFT module is structured with a pipelined scheme and avoids the waste of data transferring time between modules. The 2D-CFAR module is programmable for different applications. The designed address control is proposed to depose the edge cells. The processor occupies a core chip area of 3.353 mm $\times 3.353$ mm and has been tested on the personal computer (PC) and field programmable gate array (FPGA) platform. The power consumption and processing time are also analyzed and compared with other works. The processor consumes 55.65 mW, including SRAMs. The processing time is 12.67 ms with the maximum window size and 256 targets when operating at 125 MHz. This time is estimated based on the assumption that each chirp lasts for 0.04096 ms, and data input takes 10.48 ms. Within this period, the range FFT is completed. The Doppler FFT, 2D-CFAR with the maximum window size, and DBF with 256 targets require 0.80 ms, 1.16 ms, and 0.23 ms respectively.

Fine-grained ship detection is an important task in high-resolution satellite remote sensing applications. However, large aspect ratios and severe category imbalance make fine-grained ship detection a challenging problem. Current methods usually extract square-like features that do not work well to detect ships with large aspect ratios, and the misalignments in feature representation will severely degrade the performance of ship localization and classification. To tackle this, we propose a shape-aware feature learning method to mitigate the misalignments during feature extraction. Furthermore, for the issue of category imbalance, we design a shape-aware instance switching to balance the quantity distribution of ships in different categories, which can greatly improve the network's learning ability for rare instances. To verify the effectiveness of the proposed method, we contribute a multicategory ship detection dataset (MCSD) that contains 4000 images carefully labeled with oriented bounding boxes, including 16 types of ship objects and nearly 18 000 instances. We conduct experiments on our MCSD and ShipRSImageNet, and extensive experimental results demonstrate the superiority of the proposed method over several state-of-the-art methods.

The FY3C/GNOS launched in 2013 can only detect the scintillation of GPS navigation signals, while the FY3D satellite launched in 2017 supports the scintillation observation of BeiDou navigation satellite system (BDS) signals, thus enabling simultaneous detection of GPS and BDS ionospheric scintillation. This work presents a preliminary accuracy evaluation of BDS ionospheric scintillation observed by the global navigation satellite system (GNSS) occultation sounder (GNOS) onboard the FY3D satellite, to support long-term ionospheric scientific applications based on the BDS system. First, the F-layer worst-case (maximum) amplitude scintillation index ($S4_{\text{max}}^F$) of FY3D-BDS, FY3D-GPS, and COSMIC-GPS are, respectively, quality-controlled, and the spatial-temporal matching of $S4_{\text{max}}^F$ between FY3D-BDS and FY3D-GPS/COSMIC-GPS is performed. Then, based on the statistical deviation (std) of the $S4_{\text{max}}^F$ data pairs, the data accuracy of FY3D-BDS $S4_{\text{max}}^F$ relative to FY3D-GPS and COSMIC-GPS is obtained. The results show that the std of the $S4_{\text{max}}^F$ differences in data pairs between FY3D-BDS and FY3D-GPS/COSMIC-GPS is smaller than 0.1, which proves the high precision of BDS ionospheric scintillation detection of FY3D/GNOS. Meanwhile, the std of the $S4_{\text{max}}^F$ differences between FY3D-BDS and FY3D-GPS/COSMIC-GPS at nighttime is higher than that at daytime, and the std in the mid-latitude region is lower than that in the low-latitude and high-latitude regions.

Networked microgrids (MGs) with inverter-based and droop-controlled distributed energy resources (DERs) require operational optimization with guaranteed stability performance to ensure the stable energy supply with minimum cost, yet it remains an open challenge. Additionally, the discrete nature of MGs leads to convergence issues to existing optimization methods thereby leading to difficulties obtaining feasible solutions for large-scale networks. This article develops a paradigm for discrete droop control to improve microgrids’ controllability in managing voltage and frequency fluctuations. With the emergence of Internet of Things, the computational tasks are distributed among local resources. The utilized Distributed and Asynchronous Surrogate Lagrangian Relaxation (DA-SLR) method distributes the optimization tasks among the MGs and efficiently coordinates the distributed subsystems. A small-signal model of the operational optimization is then developed to verify the system’s stability. Numerous case studies have proven the DA-SLR’s efficacy in comparison to various variations of the alternating direction of multipliers method (ADMM).

A fiber-optic accelerometer based on extrinsic Fabry-Perot interferometer (EFPI) is reported. The sensor adopts flywheel diaphragm structure as a sensitive element, where the mass block is bonded in the center, while the beam of the flywheel acts as a spring. The sensitive element forms single degree of freedom spring-mass-damper system. The parameters of the accelerometer are designed based on finite element simulation (FEM), and the transfer function of the sensor is established by numerical simulation. The experiments indicate that the sensor has a wide detection frequency band with the detectable ability for micro-vibration low as 0.01 Hz, and the sensitivity is 7.02 rad/g @20 Hz, and reach an average sensitivity of 6.39 rad/g in 5–20 Hz. The research shows that the accelerometer has a good development prospect in the field of low frequency micro-vibration detection.

In this paper, with the help of an intelligent reflecting surface (IRS), the source (S) and destination (D) exchange information through the two-way decode-and-forward relay (TW-DFR). We focus on the phase optimization of IRS to improve the system sum rate performance. Firstly, a maximizing receive power sum (Max-RPS) method is proposed via eigenvalue decomposition (EVD) with an appreciable system sum rate enhancement, which is called Max-RPS-EVD. To further achieve a higher system sum rate, a method of maximizing minimum rate (Max-Min-R) is proposed with high complexity. To reduce its complexity, a low-complexity method of maximizing the sum rate (Max-SR) via general power iterative (GPI) is proposed, which is called Max-SR-GPI. Simulation results show that the proposed three methods outperform the random phase method, especially the proposed Max-SR-GPI method is the best one achieving at least 20% system sum rate gain over random phase. Additionally, it is also proved the optimal system sum rate can be achieved when TW-DFR and IRS are located in the middle of S and D.

Quantum key distribution (QKD) has been considered a potent and mature approach to secure communication in the quantum era. However, there exist resilience gaps that hinder the practical usefulness of QKD. This article improves the resilience of a QKD-based microgrid by leveraging software-defined networking (SDN). Specifically, the following contributions have been made. First, a novel QKD- and SDN-based communication framework is devised for microgrids. Second, an enhanced QKD simulator is developed, which can simulate both the number of generated keys and the amount of classical data required, providing a more accurate QKD study in microgrids. Third, an SDN-based resilience enhancement strategy, which incorporates three practical SDN applications, is developed to improve the system's resilience. Finally, a QKD- and SDN-enabled cyberphysical microgrid prototype is built in an RTDS environment using real whitebox SDN switches. Extensive case studies validate the effectiveness and excellent performance of SDN to improve the resilience of QKD-based microgrids in various situations.

Rolling bearing fault diagnosis is one of crucial tasks in mechanical equipment fault diagnosis. Currently, artificial intelligence and machine learning-driven fault diagnosis methods are extensively utilized for rolling bearing. When compared to traditional techniques, the diagnostic accuracy has significantly improved. These methods, however, need a substantial amount of labelled training data, which is difficult to obtain in actual failures. In order to resolve this problem, Transfer Learning (TL) was created to learn in the target domain by accessing knowledge from the pertinent labelled source domain. Inspired by Maximum Mean Discrepancy, this paper puts forward a Convolutional Neural Network (CNN) based Two-layer Transfer Learning (CTTL) method for fault diagnosis. In the first layer, the fault features are automatically extracted by CNN and a term called Feature Weighted Maximum Mean Discrepancy (WMMD) is considered to minimise the difference between source and target domains. In the second layer, the Third Dataset, which is based on the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) method was designed on the basis of the predicted labels from the first layer. The Calinski-Harabaz (CH) index of the Target Dataset controls the iteration times of CTTL. CTTL change the process of Transfer Learning method from learning the distribution of domains to learning the distribution of fault types in more detail, which will get higher accuracy. Proposed CTTL is tested by the bearing datasets of Case Western Reserve University (CWRU) and XJTU-SY. The experimental findings reveal that CTTL is capable of achieving a high diagnosis accuracy across different load domains. In the majority of experiments, CTTL outperformed other algorithms, including Deep Neural Network, Support Vector Machine, and several other methods.

Precise orientation information facilitates target recognition of sonar images. Traditional orientation estimation methods, including the Hough transform method, have poor antinoise abilities, so they cannot achieve high estimation precision for sonar images, which usually have low signal-to-noise ratios. The convolutional neural network (CNN) regression method is sensitive to image affine transformations and requires extensive computation, so it cannot achieve strong robustness or high speed. To achieve high-precision, high-speed, and robust orientation estimation of sonar image targets, we present a novel orientation estimation method via the wavelet subimage energy ratio (WSER). The WSER varies with the rotation angles of images and has the highest value when the long axes of targets are vertical. It is translational and scale invariant and does not need supervised training. Therefore, we propose estimating orientations of sonar image targets by finding the max of the WSER to achieve high precision, high speed, and strong robustness to the translation and scale transformations of images. The results of experiments on a self-made sonar image dataset show that the mean absolute error (MAE) of the proposed method is 7.9°, while the MAEs of the CNN regression method and traditional methods are 10.1 and 22.9°, respectively. In addition, the proposed method is six times faster than the CNN regression method. The proposed orientation estimation method has also been applied to align the orientation of sonar images and CNN can achieve state-of-the-art performance for rotated target recognition using the aligned images. Data and codes are publicly available.

Two different Global Navigation Satellite System (GNSS) Reflectometry wind products are to be developed for the Global Navigation Satellite System Occultation Sounder II (GNOS-II) payload on the FengYun-3E (FY-3E) satellite: A global wind product optimized for global low-to-medium wind speeds and a cyclone wind product optimized for high wind speeds in the tropical and extratropical cyclones. This article presents a comprehensive assessment of the GNOS-II global wind product for both global positioning system (GPS) and Beidou (BDS) product at wind speeds below 25 m/s using its data of first six months. Three datasets, including the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis, HY-2B scatterometer and Soil Moisture Active Passive (SMAP) winds, are used for the assessment. The assessment of the Level 1 observables shows that the instrument's performance and calibration are generally stable. For the assessment of retrieved winds, the general performances of GPS and BDS products are similar. The overall wind speed root-mean-square error is 1.54 m/s for GPS-R and 1.44 m/s for BDS-R when compared with ECMWF, 1.48 m/s for GPS-R and 1.36 m/s for BDS-R when compared with HY-2B, and 2.08 m/s for GPS-R and 2.04 m/s for BDS-R when compared with SMAP. The overall wind speed biases for GPS-R and BDS-R are about zeros when compared with each dataset. The assessment is also made by the dependence on time, signal-to-noise ratio, observable, GNSS block, wind speed, incidence angle, and geography. Overall, the FY-3E GNOS-II global wind product is consistent over time and space with an accuracy meeting the requirement of operational use.

Accurate, timely, and reliable traffic flow prediction is essential for an intelligent transportation system due to the complex spatio-temporal correlation of traffic flow. The prediction model based on graph convolution neural network (GCN) has become mainstream in recent years. However, most of the prediction models based on GCN only use an adjacency matrix to characterize the spatial correlation of traffic flow, ignoring the complex and dynamic relationship between road network adjacent nodes and missing the hidden connection between the global nodes of the road network. This paper proposes a SAX-STGCN network for traffic flow prediction to solve the above problems. The SAX-STGCN model uses symbolic approximation (sax) to obtain the similarity of the historical data of the predicted node in the previous period, including adjacent nodes and non-adjacent nodes, forming a similarity matrix to replace the original adjacency matrix composed of 0 and 1, which is defined as a global sax-correlation matrix to characterize the correlation between nodes in the road network and capture the implicit spatial relationships in the road network. Then, based on the dynamic global sax-correlation matrix, GCN is used to capture the spatial correlation of traffic flow, and gated recurrent unit (GRU) is used to capture the temporal correlation of traffic flow. The prediction accuracy is better than the baseline and has long-term prediction ability.

The monocular simultaneous localization and mapping (SLAM) system is one of the most important among all visual or visual-inertial SLAM (VSLAM or VISLAM) systems due to its low cost, easy calibration and identification. Initialization is always crucial to bootstrap the monocular SLAM system. With the rapid growth of some quick-start required SLAM applications, e.g., augmented reality (AR) and unmanned aerial vehicles (UAVs), devising faster and more accurate initialization has become a central problem. Traditional initialization uses the first two frames to create landmarks and then computes camera poses for the subsequential frames, using a 3D-to-2D perspective-n-points (PnP) mechanism. In this paper, we propose a novel three-frame-only (3FO) initialization approach for the monocular SLAM system, which consists of two steps. In the first step, we use the first two frames to preinitialize poses and landmarks, and in the second step, we use the second and third frames to improve the preinitialization by using the scale consistency of the landmarks generated in the first step to filter out outliers and using inliers to generate more robust landmarks. In both steps, we use the pretrained multilayer perceptron (MLP) combined with homotopy continuation to solve the essential matrices. Finally, we perform a global bundle adjustment (BA) to refine the three camera poses and all the created landmarks. The proposed 3FO initialization approach is evaluated experimentally on the EuRoC benchmark data set with the initialization time and trajectory metrics. The results show that, compared to the traditional ORB-SLAM2 initialization, the 3FO approach reduces the initialization time by 5 times and improves the accuracy by 36.7% on average.

Aircraft detection in synthetic aperture radar (SAR) images is still a challenging research task because of the insufficient public data, the difficulty of multiscale target detection, and the complexity of background interference. In this article, we construct a public SAR aircraft detection dataset (SADD) with complex background and interference objects to facilitate the research in SAR aircraft detection. Then, we propose the scale expansion and feature enhancement pyramid network as the SADD baseline. It uses a four-scale fusion method to combine the shallow position information with the deep semantic information, effectively adapting to the multiscale target detection in SAR images, significantly improving the detection effect of small targets. The feature enhancement pyramid structure is connected behind the backbone network to weaken the background texture and highlight the target to achieve feature enhancement, improving the ability to extract target features in complex backgrounds. Finally, to further improve the detection performance of the small-scale SAR aircraft dataset, we propose a domain adaptive transfer learning method. Experiments on SADD show that this method can significantly improve the recall rate and F1 score. At the same time, we find that the transfer effect of using homologous but different types of targets as the source domain is better than those of heterologous but same types of targets in SAR aircraft detection, which is instructive for future research.

Although face recognition has achieved great success due to deep learning, many factors may affect the quality of faces in the wild, such as pose changes, age variations, and light changes, which can seriously affect the performance of face recognition. In this work, an effective approach called Efficient Lightweight Attention Networks (ELANet) is proposed to address the challenge brought by the impacts of poses and ages on face recognition performance. First, similar local patches are particularly important when the geometry and appearance of a face change drastically. To alleviate this challenge, spatial attention is used to capture important locally similar patches and channel attention is employed to focus on features with different levels of importance. Furthermore, Efficient Fusion Attention (EFA) module is designed to achieve better performance, which can alleviate the computational effort required by fusing spatial and channel attention. Second, multi-scale features learning is necessary because pose or large expression changes can cause similar recognition regions to appear at different scales. For this purpose, pyramid multi-scale module is presented, which constructs a series of features at different scales via pooling operations. Third, to unite low-level local detail information with high-level semantic information, the features of different layers are fused by Adaptively Spatial Feature Fusion (ASFF) instead of simply utilizing addition or concatenation. Compared to recent lightweight networks, the ELANet improved performance by 1.83% and 2.17% on the CPLFW and VGG2_FP datasets, respectively, and by 0.92% on the CALFW dataset. The ELANet addresses the challenge regarding the impacts of poses and ages on face recognition performance with few parameters and computational effort and is suitable for embedded and mobile devices.

Change detection methods play an indispensable role in remote sensing. Some change detection methods have obtained a fairly good performance by introducing attention mechanism on the basis of the convolutional neural network (CNN), but identifying intricate changes remains difficult. In response to these problems, this article proposes a new model for detecting changes in remote sensing, namely, MTCNet, which combines the advantages of multiscale transformer with the convolutional block attention module (CBAM) to improve the detection quality of different remote sensing images. On the basis of traditional convolutions, the transformer module is introduced to extract bitemporal image features by modeling contextual information. Based on the transformer module, a multiscale module is designed to form a multiscale transformer, which can obtain features at different scales in bitemporal images, thereby identifying the changes we are interested in. Based on the multiscale transformer module, the CBAM is introduced. The CBAM is split into a spatial attention module and a channel attention module, which are applied to the front and back ends of the multiscale transformer, respectively. Spatial information and channel information of feature maps are modeled separately. In this article, the validity and efficiency of the method are verified by a large number of experiments on the LEVIR-CD dataset and the WHU-CD dataset.

Remote sensing image superresolution has made significant progress in recent years, aiming to restore natural and realistic high-resolution images from low-resolution images. However, most image superresolution remote sensing methods are improved only by deepening their network and expanding the network size, consuming substantial computing resources and imposing a bottleneck in development. Here, we propose an end-to-end image superresolution network called texture transfer transformer for remote sensing image superresolution (T$^{3}$SR). For the first time, T$^{3}$SR introduces image texture transfer into remote sensing, which achieves the most advanced results. Specifically, T$^{3}$SR divides image superresolution into two stages: texture transfer and feature fusion. First, to solve the problems of missing textures, artifacts, and blurring in a single image superresolution approach, we design a texture transfer module to serve the shallow texture transfer. Second, to further reduce the dependence of the model on the reference image, we propose a U-Transformer-based feature fusion scheme to reduce the dependence on the reference image. Finally, we conduct numerous experiments on standard public datasets to fully evaluate our approach. In addition to verifying the method's superiority based on the reference image paradigm, we also test the performance without the reference image. All results show that our method yields an abundant texture and finish with better visual results. Moreover, the best score is also obtained in the quantitative parameters of PSNR and SSIM. Compared with the best available approach, T$^{3}$SR has an improved performance by 0.79 dB and 0.33 dB in the datasets of WHU-RS19 and RSSCN7, respectively.

In this paper, an environment-robust polarization-based phase-shift dynamic demodulation method for optical fiber acoustic sensor is proposed. Through optical lever effect of polarization low-coherence interference and amplitude scaling differential cross multiplication (AS-DCM) algorithm, acoustic-source induced interference phase of the acoustic sensor can be demodulated in real-time with high precision. The maximum relative phase demodulation error was only 0.694% when the cavity length had an offset ∼4 μm. The maximum cavity length demodulation error was keeping less than 3.988 nm under the optical power reduced ten times, which demonstrated an ultra-stable demodulation for the largest permitted optical power attenuation range, to our best knowledge. The proposed method has a capability of tolerating large optical power attenuation and large cavity length random offset, providing an environment-robust way for optical fiber acoustic sensor working in extreme environments.

Q-switching in a mode-locked laser not only makes the amplitude of every single output pulse unequal, but also limits the time width and peak power of output pulses. This paper investigates the Q-switching in a self-mode-locked semiconductor disk laser numerically and experimentally. By using the delay differential equations for passively mode-locking, conditions of Q-switching in a self-mode-locked semiconductor disk laser are numerically analyzed for the first time. Meanwhile, based on the experimental results, the causes of Q-switching tendency including the change of nonlinear refractive index and the change of soft aperture, are also discussed. Some possible measures to suppress Q-switching instability, i.e., to obtain stable continuous-wave mode-locking in a self-mode-locked semiconductor disk laser are proposed.

With the development of vision and haptic sensor technologies, robots have become increasingly capable of perceiving their external environment. Although machine vision and haptics have surpassed humans in some aspects of perception, it is difficult for robots to describe objects from multiple viewpoints using a combination of visual and haptic modalities. In this study, we use convolutional neural networks to separately extract visual and haptic features and then fuse these two types of features. Then, multitask learning is combined with multilabel classification to form a multitask-multilabel classification method. The developed method is used to identify the color, shape, material attributes, and class of an object from the visual-haptic fused feature vector. To verify the effectiveness of the proposed object description method, experiments are conducted on the PHAC-2 dataset and the collected VHAC dataset. The experimental results show that the proposed method produces the most accurate object descriptions with the smallest number of parameters.

In the process of acute resistance exercise, repeated variation in motion intensity can lead to muscle fatigue and heart failure. Therefore, acquiring the interval of motion intensity variation in time the training pattern and effect can be improved by acquiring the interval of motion intensity variation. In order to achieve this goal, an improved Inception-V3 model is proposed for motion intensity variation interval estimation. The MIVIE(Motion Intensity Variation Interval Estimation) dataset consisting of Strong, Moderate, Weak groups achieve centralized and uninterrupted collection. Then, the multi-modal fusion vectors of time-frequency eigenvalues are stacked up to $227\times227$ grayscale images fed into improved inception-CNN. Finally, the manipulator’s trajectory optimization is completed under the guidance of ATO-DQN (Adaptive Trajectory Optimization-Deep Q Network) algorithm based on the motion intensity interval estimation. This work can improve the non-stationary effect of motor speed caused by changes in motion intensity during rehabilitation, which can better guarantee the safety of patients.

A differential pressure fiber optic infrasound sensor based on a mechanical filter was developed, for the detection of infrasound within the frequency range of 0.01--1 Hz. The sensing unit employs a miniature polyphenylene sulfide (PPS) diaphragm which is fixed onto a glass ferrule. An extrinsic Fabry-Perot (F-P) interferometer was fabricated between the optical fiber end face and the PPS diaphragm, and a mechanical filter structure introduced to improve the minimum detectable frequency down to 0.005 Hz. Experimental results indicate that this infrasound sensor shows a favorable frequency response for infrasound signals between 0.01 and 1 Hz, and it exhibits an average minimum detectable pressure (MDP) of 88.85 mPa·Hz−1/2. Our platform provides a promising solution for sensors in infrasound signal detection applications.

In recent years, video satellite technology has greatly developed, and satellite video is captured using this technology. Because video satellites in motion are consistently aimed at certain areas, satellite video is usually composed of a series of frame images of the same area with different viewing angles; these satellite video images have different base-height ratios with each other, which makes multiview stereo mapping possible. This paper presents a method of multiview stereo mapping based on satellite video images to generate a digital surface model (DSM), digital elevation model (DEM), and digital ortho map (DOM) and analyzes the influence of different numbers and different viewing angles of satellite video images involved in multiview stereo mapping on the quality of the results. Satellite video images of Urumqi, China, captured by the “Jilin-1” video-03 satellite, were used as experimental data for the verification of the proposed method, and the results of the experiment show that: a high-quality DSM, DEM, and DOM can be generated by multiview stereo mapping based on two or more satellite video images, and the elevation precision and plane precision of the DEM results are 2.751 m and 2.419 m, respectively; such as in traditional stereo mapping, the larger the base-height ratio of the images involved in the calculation is, the higher the elevation precision of the DSM results will be within a certain extent. When the images participating in the calculation comprise images viewed from the nadir direction, forward direction and backward direction at the same time, DSM results with better building textures can be generated with high precision, and as the number of satellite video images increases, the building textures in the DSM results are optimized.

Light-pen coordinate measuring machines are widely used due to their portability; however, the measurement accuracy is affected by camera calibration, light-pen probe calibration, control points, etc. To study the influence of control point dispersion on measurement accuracy, a light-pen coordinate measuring system is constructed in this paper. In this system, a new type of control point distribution is designed. Based on this distribution, five light-pens with different control point dispersions are created and an algorithm is proposed to match control points with their images. The experimental repeatability results for the five light-pen probes indicate satisfactory repeatability, which ensures the effectiveness of the measurement results. The experimental measurement results for five light-pens demonstrate that the measurement error decreases with the decline of the control point dispersion at first, but begins to increase when the control point dispersion reaches a certain value.

The facial expression is one of the most common ways to reflect human emotions. And understand different classes of facial expressions is an important method in analyzing human perceived and affective states. In the past few decades, facial expression analysis (FEA) has been extensively studied. It illustrates few of the facial expressions are exactly individual of the predefined affective states but are blends of several basic expressions. Some researchers have realized that facial expression recognition can be treated as a multi-label task, but they are still troubled by the inaccurate recognition of multi-label expressions. To overcome this challenge, a novel multi-feature joint learning ensemble framework, called MF-JLE framework, is proposed. The proposed framework combines global features with several different local key features to consider the multiple labels of expressions embodied in many facial action units. The ensemble learning is introduced into the framework, combines the global module and the local module on the loss, and carries out the joint iterative optimization. The ensemble of the whole framework improves the accuracy of multi-label recognition of different modules as weak classifiers. In addition, the traditional multi-classifier cross-entropy loss has been replaced by the binary cross-entropy loss for a better ensemble. The proposed framework is evaluated on the real-world affective faces (RAF-ML) dataset. The experimental results show that the proposed model is better than other methods in both quantitative and qualitative aspects, whether compared with traditional shallow learning methods or recent deep learning methods.

Composite manufacturing processes such as automated tape placement (ATP) and filament winding process (FW) put forward specific requirements for the geodesic curvature along the layup paths of the composite tows and tapes. In this paper, a non-uniform cubic b-spline element method is proposed for solving the boundary value problem of curves with prescribed geodesic curvature. The differential equation system of the target curve is discretized through the point collocation method, and a quasi-Newton iteration scheme is adopted to approach the real solution from an initial approximation. The proposed method is proved to have third order accuracy, which shows more superiorities comparing with existing numerical methods. Simulations and experiments on a series of parametric surfaces are performed to investigate the performance of the proposed approach and the results verify the high efficiency. The proposed method could cope with the BVP for curves no matter their geodesic curvature vanishes or not. At the same time, the computed curves are natural and smooth such that interpolation technique is unnecessary to ensure the continuity of target curves. One potential application of this method is trajectory optimization for automated tape placement process.

Buildings’ achievable energy flexibility refers to the real load reduction amount in an incentive-based demand response (DR) event, which presents dynamic, subjective, and uncertain characteristics. It is different from the buildings’ theoretical energy flexibility, which refers to its physical load reduction potential during a certain time period and is static and certain. The former serves as a foundation in a DR aggregator’s market transaction strategy formulation process and thereby calls on the necessity of its accurate forecasting. However, most of the existing literature focuses on the latter one, which is not necessarily equal to the real achievable flexibility. Therefore, to help DR aggregators bid accurately in the ancillary service market, this paper proposes an achievable energy flexibility forecasting method for a building equipped with an integrated energy management system based on the decision tree model. The impact of gas-fired equipment on buildings’ achievable-energy-flexibility is also taken into account in the proposed method. The case study indicates that the proposed method exhibits promising performance in forecasting the achievable energy flexibility of a building.

A joint multiple-input multiple-output (MIMO) processing framework is proposed to exploit spatial diversity and moving cues for the enhancement of target detection in a shallow ocean environment. Orthogonal signals are transmitted to illuminate different aspects of a target and beamforming operation is carried out over the received data for estimating target bearing. The target range is achieved by a replica correlation integrator where the beamformer outputs are matched with transmitted signals. After meanshift clustering algorithm is carried out over the bearing-range spectrums to generate the clutter centers, the potential trajectories of targets can be tracked by an improved Bernoulli filter with inputs of these centers. The at-sea experimental results have shown the effectiveness of the joint processing framework in MIMO detection of moving targets.

Change detection methods for optical remote sensing images play an important role in environmental resource management. Although recent methods based on deep learning demonstrate incredible ability by constructing networks, first, extracting bitemporal features in a separate manner; second, fusing bitemporal images before forwarding them into the single-level network. Both severely neglect the effect of spatial-temporal feature correlation between bitemporal images. In addition, most existing methods represent multiscale feature pairs in a layer-wise manner like ResNet, failing to consider the inner multilevel structure. In this work, we propose a new siamese change detection feature encoder backbone named cross-siamese Res2Net (CSRes2Net), by establishing crossed and hierarchical residual-like connections within one single residual block. The CSRes2Net represents dual features in a fine-grained manner and fully leads to the flow of bitemporal features. In addition, recent learning-based methods designed some spatial-temporal relation modules to capture the pixel-level pairwise relationship and channel dependency based on self-attention mechanism, but they only consider spatial and channel dimension corrections separately with excessive parameters. So we propose a lightweight cross spatial-channel triplet attention module to capture cross-dimensional long-range relationship between triplet combinations: channel with height, channel with width, channel with channel. Finally, we propose a hierarchical-split block for generating multiscale feature representations in a coarse-to-fine fashion. The experiments results on LEVIR-CD and season-varying change detection dataset outperform most state-of-the-art models.

Few-shot object detection (FSOD) eliminates the dependence on tremendous instances with manual annotations in conventional object detection. We deem that the scarcity of positive samples is the main reason that restricts the performance of FSOD detectors. In this paper, a novel FSOD model via sample processing, namely, FSSP, is proposed to detect objects accurately with only a few annotated samples, which is based on the structural design of the Siamese network and uses YOLOv3-SPP as the baseline. Central to FSSP are our designed self-attention (SAM) and positive-sample augmentation (PSA) modules. The former attempts to better extract the representative features of hard samples, and the latter expands the number and enriches the scale distribution of positive samples, inhibiting the growth of negative samples. For the fine-tuning phase, we modify the classification loss function to increase the punishment for hard samples. Experiments conducted on the PASCAL VOC and MS COCO datasets confirm that the proposed FSSP achieves competitive detection performance compared with state-of-the-art detectors.

The condition of the transformer can be accurately assessed based on dissolved gas online monitoring and analysis technology. With the widespread use of Dissolved Gas Analysis (DGA) technology, a large amount of dissolved gas time series data is obtained and forms DGA big data, which brings challenges to the data selection process of traditional condition assessment methods. Inaccurate results might be generated by selecting excessive or insufficient data from a long-span dissolved gas time series. The shorter dissolved gas time series fails to fully characterize the operation law of transformers, and the longer dissolved gas time series contains redundant information that results in inaccurate analysis models and excessive calculation consumption. This paper attempts to mine the optimal data selection rule of long-span dissolved gas time series for more accurate and efficient analysis. In this paper, C-C method was employed to calculate the phase space parameters of the time series. By analyzing the convergence of the correlation integral, the boundary conditions that maintain the time series phase space stability were revealed, and so was the relationship between the optimal length and the embedding dimension. The real cases show that the optimal data set provides the same accuracy and significantly improves the computation speed by 113.54 times. The optimal data selection rule eliminates redundant information in the long-span dissolved gas time series and retains the “optimal data”, which not only preserves the characteristics of the original data, but also improves the computation speed of transformer condition assessment.