Search

This paper provides a comprehensive categorization of extracurricular clubs (e.g., robotics, STEM, and business clubs) in hundreds of American high schools and examines how socioeconomic factors—such as race, income level, pupil-teacher ratio, and school size—affect the quantity and variety of these clubs. We reveal that robotics clubs have become the most prominent extracurricular activity in promoting electrical and computer engineering, as well as STEM fields at large, outperforming even math clubs, with 38.9% of schools hosting them. Although schools across different socioeconomic backgrounds all manage to offer robotics clubs, school demographics do affect the total number of clubs a school provides. Nevertheless, successful schools are capable of expanding their extracurricular offerings despite these constraints. Specifically, within schools of similar demographics, the top 25% offer 8.8 times more clubs than the bottom 25%.

This paper delves into the realm of observer-centric tracking control for discrete-time nonlinear networked control systems, accounting for packet dropouts and addressing the issue of state variables that are not directly measurable. To tackle this, an innovative fuzzy observer independent of premise variables is devised. A key innovation lies in the introduction of a new fuzzy observer that compensates for unmeasurable states, utilizing the system’s dynamic output to achieve more precise estimation results. The observer employs both the error between the measurable output of the system and the observer’s estimate, as well as the error between the system’s measurable output and the reference model’s output, as compensation mechanisms. Furthermore, in harmony with the controller, the observer adopts a regulatory factor structure and utilizes the same membership functions as the controller, thereby achieving a less conservative design. Subsequently, a novel fuzzy tracking controller is meticulously proposed, leveraging the estimates provided by the observer. Moreover, sufficient criteria are formulated to ensure that the resulting closed-loop system attains stochastic stability. Finally, to underscore the practical effect of our approach, illustrative examples are presented.

In order to improve the stability of direct current (DC) bus voltage when it is disturbed, the virtual DC motor (VDCM) control is applied to the DC converters. However, the constant control parameters in traditional VDCM control cannot provide sufficient inertial and damping support for the system while achieving good dynamic performance. Aiming at this issue, the four-parameter adaptive VDCM control is introduced in this paper. Firstly, the small signal model of dual active full bridge (DAB) with VDCM control is built, then the influence of four different VDCM control parameters on the stability of DC system is studied by drawing the bode diagram and the step response curve. On this basis, the four-parameter adaptive VDCM control is proposed. Finally, the superiority of four-parameter adaptive VDCM control in suppressing DC bus voltage fluctuation and improving the system dynamic response speed is verified by simulation and experiment, the experimental results show that, compared with three-parameter adaptive VDCM control, the fluctuation peak value of DC bus voltage can be reduced about 33% under the four-parameter adaptive VDCM control, and the recovery time after receiving a disturbance of DC bus voltage can be reduced about 25% under the four-parameter adaptive VDCM control. The results show that the four-parameter adaptive VDCM control strategy studied in this paper can further suppress the fluctuation peak of DC bus voltage and improve the dynamic response speed of DC system, which is of great significance to improve the stable operation of DC microgrids.

This paper considers the event-triggered distributed fixed-time consensus control problem of leader-following multi-agent systems with nonlinear dynamics and inherent disturbances. First, an event-triggered distributed strategy, that is used to decide when to update states by agents, is proposed. Under the proposed strategy, each agent receives its neighbors’ state to make decisions only if the measurement error exceeds a certain threshold. Second, we design a nonsingular terminal sliding mode consensus protocol so that all followers reach the leader’s state in a fixed-time. Particularly, the gain in the consensus protocol is determined by the settling time, which makes devising and tuning the gain conveniently. Simulation examples are worked out to demonstrate the effectiveness of our theoretical results.

With the advantages of easy implement and few adjustment parameters, grey wolf optimization (GWO) performs well in solving global optimization problems. However, the original GWO can suffer from slow convergence speed and plunge into local optimums in dealing with complex problems. To address these issues, improvement of mixed-strategy based grey wolf optimization (MGWO) is proposed in this paper. Firstly, we present chaos strategy in combination with quasi-opposition learning strategy for generating more high-quality population to improve convergence speed of the algorithm. Then, we embed sine cosine mechanism into the original GWO to jump out of local optimums. We utilize the interaction of sine and cosine to better balance the exploration and exploitation capabilities of the algorithm. Finally, we use benchmark functions for testing the performance of proposed algorithm. Experimental results demonstrate that proposed algorithm can improve search accuracy and convergence speed compared with other state-of-the-art algorithms. Simultaneously, to verify further the performance of proposed algorithm, classification problems on gene expression datasets are also evaluated. Both continuous and binary versions of MGWO are respectively used for completing parameters optimization and feature selection. Experimental results show that proposed algorithm has better performance than others in terms of Sensitivity, Specificity, G-mean, Accuracy and F-measure.

In order to form a collaborative operation between village power grid and the end users, a two-stage management strategy is proposed where residential photovoltaic power systems have been linked to the distribution network in a rural area. At stage one, a regulation method considering active and reactive power is formed with a purpose of maintaining power grid within a security domain in operation; at stage two, to maximize the benefits of both the grid and its costumers, a “grid-user” interactive strategy is established so as to allow both the grid and users have opportunities and privileges equally. At the second stage, the power grid takes economics as an objective while delivering the extra benefits to the user as an incentive reward to improve the costumers income while avoiding extra resource investment to reach the same purpose. The test results show that the proposed strategy can effectively ensure the security of the network while improving the economy of the network operator and users.

This article under unreliable communication links considers the problem of dynamic compensation output-feedback control for non-linear discrete-time networked control systems (NCSs) with packet dropouts in the light of the Takagi-Sugeno (T-S) fuzzy method. Specifically, the packet dropouts are modeled by the Bernoulli random binary distribution in both the controller-to-actuator (C/A) and sensor-to-controller (S/C). In addition, to estimate the missing packets, attention is focused on the dynamic output-feedback controller, which is designed using a new improved compensation strategy. Here, sufficient conditions are obtained such that the closed-loop control system is stochastically stable with the prescribed $H_{\infty }$ performance by constructing fuzzy Lyapunov function. To reduce conservativeness and deal with the non-convex problem, a cone-complementarity linearisation (CCL) program is introduced. Finally, the effectiveness of the method is demonstrated by the simulation results of the two numerical examples.

With the development of 3D sensors, 3D point cloud data can now be obtained conveniently. Therefore, it is crucial to process point cloud data automatically. Region growing is a commonly used algorithm to segment point clouds, which greatly depends on the accuracy of points’ normals and requires tuning two thresholds; i.e., the increment of curvature ( $\sigma _{th}$ ) and normal angles ( $\theta _{th}$ ). In this paper, we improve the region growing algorithm in two ways: Accurate normal estimation and strengthening the region growing criteria. For the first aspect, principal component analysis (PCA) is utilized to estimate the initial normals of the point cloud. Then, the points are divided into regular points (RP) and sharp feature points (SFP), according to their initial normals. A robust estimator-based PCA is then applied to refine the SFP normals. For the latter aspect, non-connected and non-coplanar points are detected and ignored when region grows. Finally, the segmentation performance of the proposed method is evaluated using internal and external indices. The results indicate that the proposed method can accurately estimate the point normals within an acceptable time, and obtain a better result than the classic PCA-based region growing algorithm and advanced DetMM-based methods.

Aiming at the problem that traditional surface detection is easily affected by complex industrial environments and cannot extract effective features, a deep learning-based knowledge distillation anomaly detection model is proposed. Firstly, a pre-trained teacher network was used to transfer knowledge of normal samples to the student network in the training phase. In the testing phase, defect detection was achieved based on the feature differences in the output of the teacher-student network for abnormal samples. Secondly, the attention mechanism module and the feature fusion module were added to the teacher network, which enhanced the detection ability of various defects of different sizes and increased the difference in output features between the teacher network and the student network. Finally, the defect image was located at the pixel level. Compared with the advanced knowledge distillation methods, the experimental results of the proposed model on the detection on MVTecAD and the anomaly localization on MVTecAD showed that the method in this paper improved to 90.3% and 90.1% on AUROC respectively, which verified the effectiveness of the method.

The two-dimensional optimization capability of the space-time minimum variance distortionless response (STMVDR) can improve the spatial spectrum estimation performance of the broadband beamformer with limited samples, but its high sensitivity to frequency errors may lead to the suppression of some narrowband signals not at the grid point of the frequency spectrum calculation. In order to reduce the sensitivity of STMVDR to frequency errors, a broadband spatial spectrum estimation method based on STMVDR and frequency derivative constraints is proposed in this paper. First, based on the beam response of the STMVDR beamformer, an arbitrary order frequency derivative constraint is derived. Second, combining the gain constraint and frequency derivative constraint, the linear constrained minimum variance problem is established to solve the optimal weight of the beamformer. Finally, the calculation method of the broadband spatial spectrum is derived. Numerical simulations show that the proposed method not only solves the problem of frequency mismatch caused by the insufficient number of points for calculating frequency spectrum, but also does not cause the spatial spectrum peaks to deviate from the true direction due to the broadening of the main lobe of frequency spectrum. In addition, the proposed method can reduce the computational complexity of the algorithm by more than half by increasing the interval of the frequency calculation grid points, and does not significantly loss in the angular resolution capability and estimation accuracy of the algorithm.

The frequency-domain constrained broadband space-time beamformer has been widely used in many fields because of its excellent performance and no need for pre-steering delay. However, the beam response of the frequency-domain constrained space-time broadband beamformer depends heavily on the phase center location of the array. The wrong phase center location of the array will lead to undesirable wide main lobe and high side lobe. To choose the location of the phase center of the array correctly, this paper makes an in-depth theoretical discussion on this dependency based on the uniform linear array. First, the dependence of the quiescent beam response of the frequency-domain constrained space-time broadband beamformer on the phase center location of the array is theoretically derived, and it is proved that there is an optimal phase center location of the array which can make the beamformer obtain the best quiescent beam response. Then, a selection criterion of the optimal array phase center location is proposed, and based on this criterion, the closed-form solution of the optimal array phase center location is theoretically derived when the constraint matrix contains two pre-selected points. Finally, based on the closed-form solution, through correct extrapolation, the suboptimal solution for the optimal array phase center location is obtained when the constraint matrix contains any preselected points. Numerical simulations verify the correctness and effectiveness of the proposed methods.

During the production process of steel, there are often some defects on the surface of the product. Therefore, detecting defects is the key to produce high-quality products. At the same time, the defects of the steel have caused huge losses to the high-tech industry. A steel surface defect detection algorithm based on improved YOLO-V7 is proposed to address the problems of low detection speed and low detection accuracy of traditional steel surface defect detection methods. First, we use the de-weighted BiFPN structure to make full use of the feature information to strengthen feature fusion, reduce the loss of feature information during the convolution process, and improve the detection accuracy. Secondly, the ECA attention mechanism is combined in the backbone part to strengthen the important feature channels. Finally, the original bounding box loss function is replaced by the SIoU loss function, where the penalty term is redefined by taking the vector angle between the required regressions into account. The experimental results show that the improved model proposed in this paper has higher performance compared with other comparison models. Based on our experiments, the proposed method yields 80.2% mAP and 81.9% on the GC10-DET dataset and NEU-DET dataset with high speed, which is better than other existing models.

The decode-and-forward (DF) based free-space optical-radio frequency (FSO-RF) hybrid system combines the advantages of both FSO links and RF links, which makes the system easy to be deployed and enables the extended transmission coverage. In the case of multiple users, the unreasonable dynamic antenna selection algorithm will lead to multi-user interference (MUI) and reduce the spectral efficiency of the system. Therefore, we propose hybrid precoding based on manifold learning with the antenna partitioning algorithm for dual-hop hybrid FSO-RF systems. In the hybrid precoding of dynamic subarray structure, the high-dimensional channels are embedded into the low-dimensional manifolds by the optimized multidimensional scaling (MDS). The potential spatial correlations of the high-dimensional channel are preserved by the scaling by majorizing a complicated function (SMACOF) algorithm. Through proper user clustering, the hybrid precoding is investigated for the sum-rate maximization problem by manifold quasi-conjugate gradient methods. Meanwhile, an antenna subarray partitioning algorithm is proposed, so that each antenna unit can be assigned to an RF chain based on the increment of the user’s maximum signal to interference noise ratio (SINR). By calculating the simulated equivalent channel SINR for the selected users, the antenna division can greatly reduce the computational complexity and the size of the search space, and ensure fairness among users. Simulation results show that this method can obtain almost the best summation rate and higher spectral efficiency compared with the conventional method.

Clustered storage systems are dominant solutions for the era of data-intensive computing. Ceph represents a sustainable clustered storage solution, supporting object, block, and file storage capabilities with no single point of failure. Despite the strong management abilities, security remains a serious concern in the Ceph storage system. To date, authentication and access control are the only supported security protocols in the system. Data confidentiality will be undermined if a malicious insider or outside intruder accesses storage devices. This study proposes a lightweight cryptographic-based interface, CephArmor, for a Ceph storage system to ensure data confidentiality in storage. The proposed method has been integrated into the Ceph stable version, Pacific, and evaluated through 45Drives Storinator servers, a commercial hardware commodity for storage solutions in real-world scenarios. The experimental results denote a nuanced overhead regarding elapsed time, throughput, average operations per second, and latency on a write operation. In contrast, the read operations illustrated near-zero performance overhead for the same metrics.

Link prediction provides insight into the evolutionary mechanisms of complex networks by predicting missing edges. Existing research has proposed many similarity algorithms based on local information, and some link prediction algorithms typically perform better in different networks. It is generally believed that a megamerger is beneficial. In the perspective of link prediction, merging the good-performing algorithms brings higher prediction accuracy. And the more times the experiment is executed, the higher the accuracy of link prediction. Therefore, this research proposes a new link prediction algorithm based on the theory of megamerger in management and the concept of partnership, and uses ten actual complex networks for experiments to test the above two hypotheses. The experimental results show that megamerger is not applicable to the link prediction algorithm. In addition, there is no positive correlation between the increasing the quantity of experiments and improving the accuracy of the experiments, so the above two hypotheses are rejected. Hence, this research presumes that megamerger of the comprehensive information of the network, such as the resource flow between nodes, the degree of common neighbor nodes, and partnership of nodes, does not improve the accuracy of link prediction. For a refined network with a small number of nodes and a short average path length, it is recommended that the quantity of experiments be set to only ten can achieve the required accuracy of link prediction.

Millimeter Wave (mmWave) massive multiple-input multiple-output (MIMO) systems with hybrid analog-digital architectures can greatly increase system capacity and communicate with multiple users at the same time. Accurate channel estimation is crucial for multi-user communications, but its accuracy is limited as the number of antennas and users increases. In the matrix high-dimensional operation for multi-user channel estimation, not only is it computationally intensive, but also the estimation accuracy is low. It makes our work turn to channel estimation of the user group within a certain region to improve the accuracy of estimation. In this paper, we propose a tensor dictionary manifold learning method for channel estimation and interference elimination of the multi-user mmWave massive MIMO system. A multi-user digital-analog mixed received signal model is presented. The tensor dictionary manifold learning scheme is proposed to model the received signal as a third-order low-rank tensor to handle the high-dimensional user, antenna, and channel. After segmentation, clustering and manifold learning, multiple tensor dictionary manifold models containing a group of user signals are fitted. Tensor dictionary manifold learning can take advantage of the inherent multi-domain properties of signals in the frequency, time, code and spatial domains to maintain inter-user correlation within a user group while reducing the high-dimensional channels of the user group. Using the convex relaxation property of the tensor alternating direction method, we propose a strategy to eliminate interference from other groups. And with the help of the multi-signal classification method, the channel parameters of user groups are obtained to improve the accuracy of multi-user channel estimation. This method can perform channel estimation for multiple users with only a few pilots, and improve the performance of the system. Numerical results confirm the good performance of this method.

Temperature control is a critical aspect for the semiconductor material growth of the Molecular Beam Epitaxy system. The growth rate depends on the temperature of the beam source, and the quality of epi-layers heavily depends on the temperature of the substrate. In this paper, we reported an intelligent temperature control algorithm based on the BP neural network, which is specially optimized for the Molecular Beam Epitaxy system. After training the BP neural network by the collected temperature data of the Molecular Beam Epitaxy, the intelligent temperature control algorithm keeps the temperature error within 0.1°C. Moreover, this intelligent algorithm in the field of Molecular Beam Epitaxy achieves a better control effect, reduces the regulation time of the system and makes the system overshoot to zero.

The paper presents a wall-climbing robot featuring self-compliance for variable curvature façades. The high payload and maneuverability make it highly potential in heavy-duty operation of industrial applications. The robot consists of two traction modules and one link module. Each traction module is equipped with magnetic adhesion and crawler traction submodules. The two traction modules are connected by the link module with 4 degrees of freedom (DoF) of passive compliance. Variable curvature façade self-compliance is achieved by the passive compliant link module, which results in the attitude change decoupling of the two traction modules. The attitude can be adjusted under the effect of magnetic adhesion force to comply with the curvature variations. High payloads are achieved by the large contact area between the crawler and the surface. Omni-directional high maneuverability is enabled by the speed difference between the motors of two traction modules. The robot can maneuver in any direction on the surface with a minimum radius of 1 m and carry 36 kg payload on vertical surfaces.

This paper investigated the problem of multi-target tracking (MTT) over a vehicle sensor network. A novel adaptive square root cubature joint probabilistic data association (ASRCJPDA) was proposed. Motivated by enhancing the stability of joint probabilistic data association (JPDA) in practical application, the proposed methodology implemented a numerically stabled cubature Kalman filter for JPDA state estimate process. It improved numerical stability and acquired more accurate estimated results. Additionally, enlightened by enhancing the real time efficiency of JPDA, an adaptive tracking gate designed for the JPDA measurement associate process was proposed. It combined with the kinematics of vehicle to reduce the computational complexity of data association, which improved the robustness of MTT in complex scenarios. The virtual vehicle target tracking scenarios were built in PreScan software in order to better simulate the real traffic condition. Simulations of target tracking examples are presented to show great effectiveness and superiority of the proposed method.

In this paper, we propose TSFE-Net, two stream feature extraction networks for active stereo matching. First, we perform extra local contrast normalization (LCN) for dataset due to dependency between speckle intensity and distance. Second, we construct two stream feature extraction layers which consist of convolutional layers and deconvolutional layers in different scales to simultaneously learn the features of the original images and LCN images and aggregate context information to form the left and right features. Third, we convert the obtained depth map into disparity map in virtue of camera parameters to construct a supervised learning model. The TSFE-Net not only solves illumination effects between speckle intensity and distance but also reserves details of the original image. Our dataset are captured by RealSense D435 camera. We research extensive quantitative and qualitative evaluations based on a series of scenes, and achieve the end point error (EPE) accuracy of 0.335 on the TITAN XP platform only for valid pixel. The assessment results show that our network has the ability of real-time deep reconstruction for active pattern.

The minimum weight vertex cover problem (MWVCP) is a fundamental combinatorial optimization problem with various real-world applications. The MWVCP seeks a vertex cover of an undirected graph such that the sum of the weights of the selected vertices is as small as possible. In this paper, we present an effective algorithm to solve the MWVCP. First, a master-apprentice evolutionary algorithm based on two individuals is conducted to enhance the diversity of solutions. Second, a hybrid tabu search combined configuration checking and solution-based tabu search is introduced to intensify local search procedure. Harnessing the power of the evolutionary strategy and a novel variant of hybrid tabu search, Master-Apprentice Evolutionary Algorithm with Hybrid Tabu Search, MAE-HTS, is presented. Results of extensive computational experiments using standard benchmark instances and other large-scale instances demonstrate the efficacy of our algorithm in terms of solution quality, running time, and robustness compared to state-of-the-art heuristics from the literature and the commercial MIP solver Gurobi. We also systematically analyze the role of each individual component of the algorithm which when worked in unison produced superior outcomes. In particular, MAE-HTS produced improved solutions for 2 out of 126 public benchmark instances with better running time. In addition, our MAE-HTS outperforms other state-of-the-art algorithms DLSWCC and NuMWVC for 72 large scale MWVCP instances by obtaining the best results for 64 ones, while other two reference algorithms can only obtain 27 best results at most.

Permittivity sensing is commonly utilized in multiple fields for various applications. In this study, an enhanced RFID (Radio Frequency Identification) permittivity sensor system is designed. Additionally, we investigated the signal processing procedures in a stepwise manner to obtain the resultant graphs. For validation, a simplified system including a reader and a tag is tested. The reader in the system comprises a pulse generator and a pair of antennas connected to an oscilloscope, whereas the tag comprises an antenna and a delay line sensor. The pulse generator produces short pulses of less than 1 ns periodically, whereas the short-ended delay line on the tag acts as a sensor, wherein the delay time of the traversing wave is primarily associated with the relative permittivity of the material under test (MUT). We tested the system to derive an equation for the value of relative permittivity of the MUT using the measured signal. The system was calibrated by performing both unloaded and air tests to obtain measurement equations. Moreover, an additional test validated the water permittivity sensing, and the relative permittivity measured was 78.1. Considering the errors affecting the measurements, the measured relative permittivity of water concurs with the actual value.

This paper investigates the conducted electromagnetic interference (EMI) characteristics of an on-board multiplexing converter that utilizes paralleled silicon carbide (SiC) devices to achieve high efficiency and high-power density for the application in electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEVs). The multiplexing converter can operate as a three-phase interleaved DC/DC converter with a peak output power of 60 kW or an integrated two-stage non-isolated on-board charger (OBC) with an output power of 6.6 kW. The novelty of this paper is that it reveals that in the DC/DC mode the series resonance between the input negative lead parasitic inductance and the high frequency transmission line effects of the input power inductor will increase the conducted EMI levels, which can be mitigated by optimizing the winding methods of the power inductor or by adding a small capacitor, and reveals that in the non-isolated OBC mode the shielded cable on the battery-side will increase the grid-side EMI in the low frequency ranges, which can be suppressed by integrated installation of EMI filters without affecting the system power density. Firstly, parasitic parameters of the components of the multiplexing converter, such as the power inductors, the SiC MOSFETs and the shielded cables, are extracted and validated with impedance measurements, and the corresponding equivalent circuit models are established. Then, the conducted EMI simulation models of the multiplexing converter in different working modes are established. The influence of different interference sources and different parasitic parameters on the system EMI characteristics are analyzed, and the effective EMI suppression measures are given without affecting the system power density. Finally, the conducted EMI characteristics of the multiplexing converter and the effectiveness of the proposed EMI suppression measures are tested and verified through experiments.

Resource allocation has a direct and profound impact on the performance of resource-limited smart grids with diversified services that need to be timely processed. In this paper, we investigate a joint communication, computing, and caching resource allocation problem with distinct delay requirement of services in smart grids. This paper aims to optimize the long-term system utility based on reward and loss function. Considering the unknown dynamic environment as well as the huge state and action space in smart grids, a deep reinforcement learning algorithm based on the polling method is exploited to learn the policy by interacting with the environment. Specifically, the edge nodes (ENs) act as agents to enable the services to schedule resources appropriately. Then, the agents that are allocated based on the service requirements are queried according to the polling mechanism and the well-designed reward function is utilized to update the strategy. Extensive simulation results show that the proposed algorithm outperforms three known baseline schemes in terms of network performance with decision results. Besides, in the face of a large number of services in the smart grids, the proposed system still surpasses that of existing several baseline schemes, especially in the improvement of cache hit rate and the decrease of computing delay.

Graph embedding learns low-dimensional vector representations which capture and preserve information in original graphs. Common shallow neural networks and deep autoencoder only use adjacency matrix as input, and usually ignore node attributes and features. Shallow graph neural networks cannot spread the node characteristic information on a large scale. Many deep models suffer the problem of over-smoothing. Therefore, these methods can’t fully incorporate network information. In this paper, we propose a novel Semi-AttentionAE model to fully utilize node features, node labels, and network structure. More specifically, we integrate a supervised information extraction graph attention network to capture both node features and network structure, with an unsupervised feature extraction autoencoder to reduce dimension while preserving structure information. Finally, ensemble learning is introduced to jointly train the combined model to obtain final embedding. We conduct the node classification and visualization experiments on four real-world datasets, including two citation networks, one co-occurrence network, and one commodity network. The results suggest that the proposed Semi-AttentionAE model is capable of embedding both graph structure and node features. The integrated model has successfully exceeded or matched performance across four well-established baselines.

As the main driving equipment of modern industrial production activities, if a motor fails, it causes serious consequences. Bearings are the component with the highest motor failure frequency. It is of practical engineering significance to establish a high-precision algorithm diagnostic model for motor bearings. At present, in data-driven motor bearing fault diagnosis methods, the method of manually adjusting hyperparameters is usually adopted in complex network structure models with many hyperparameters. To realize the automatic optimization selection of hyperparameters, in this paper, a motor bearing fault diagnosis algorithm based on a convolutional long short-term memory network of Bayesian optimization (BO-CLSTM) is proposed. The algorithm combines the Bayesian optimization algorithm (BO), a long short-term memory network (LSTM) and the convolutional layer of a convolutional neural network (CNN). It saves the considerable workload of manually adjusting the hyperparameters, has good noise resistance, and realizes the true end-to-end motor bearing fault diagnosis. The proposed method is trained based on the original vibration signal of the bearing, and the accuracy of the final model reaches 100%. In addition, compared with other advanced fault diagnosis methods based on deep learning, the performance of the proposed method is significantly improved.

Aspect-level sentiment classification aims to solve the problem, which is to judge the sentiment tendency of each aspect in a sentence with multiple aspects. Previous works mainly employed Long Short-Term Memory (LSTM) and Attention mechanisms to fuse information between aspects and sentences, or to improve large language models such as BERT to adapt aspect-level sentiment classification tasks. The former methods either did not integrate the interactive information of related aspects and sentences, or ignored the feature extraction of sentences. This paper proposes a novel multi-grained attention representation with ALBERT (MGAR-ALBERT). It can learn the representation that contains the relevant information of the sentence and the aspect, while integrating it into the process of sentence modeling with multi granularity, and finally get a comprehensive sentence representation. In Masked LM (MLM) task, in order to avoid the influence of aspect words being masked in the initial stage of the pre-training, the noise linear cosine decay is introduced into $n-gram$ . We implemented a series of comparative experiments to verify the effectiveness of the method. The experimental results show that our model can achieve excellent results on Restaurant dataset with numerous number of parameters reduced, and it is not inferior to other models on Laptop dataset.

In sensor-cloud applications, a huge amount of time series are generated. Efficient similarity search approaches are necessary for processing these sensor time series data. Concerning of time series search, whole matching similarity search and subsequence similarity search are two main research focuses. In this paper, we study the whole matching similarity search problem. We propose EPLA-DSTree, which extends piecewise linear approximation on a dynamic segmentation tree index for whole matching on time series. Compared with DSTree, EPLA-DSTree improves data locality of nodes by a better time series representation. EPLA-DSTree has a tighter lower bound for nodes which leads to a better query performance. Experiments show that it has a less index building time and a better query performance. To meet the requirements of sensor-cloud applications, we present an parallel EPLA-DSTree on MapReduce, which is a popular cloud programming model.

Traditional model predictive current control (MPCC) method depends on motor model for predictive control, when the motor parameters change with the working conditions, the predictive performance of MPCC will be deteriorated. To improve the parameter robustness of MPCC, a model-free current predictive control method that combines ultra-local model and sliding mode observer is proposed. First, the prediction model of MPCC based on the mathematical model of surface-mounted permanent magnet synchronous motor (SPMSM) is replaced by the ultra-local model that does not use any motor parameters. Second, the sliding mode observer is adopted to observe the parameter of ultra-local model and compensate parameter disturbance. Finally, the stability of the sliding mode observer is proved by the Lyapunov stability criterion. The traditional MPCC method and the proposed model-free current predictive control method are comparatively analyzed, simulation and experimental results show that the proposed model-free current predictive control method can improve the parameter robustness of MPCC.

By considering the application requirements of automatic welding of labels on the heads of bundled rebars in the iron and steel industry, an automatic label welding robot system is developed based on machine vision and image processing. The system consists of six modules: pickup unit, image unit, control unit, welding stud unit, label unit, and rebar unit. It can automatically pick up welding studs, pick up labels, perform welding, detect dropped labels, and detect dropped welding studs. The development process of label welding mainly includes the following steps. First, the Cascade RCNN object detection algorithm is used to detect the rebar head, so that the number of rebars and the pixel coordinates of the rebar head centers can be obtained. The detection accuracy of the number of rebars can reach 100%, and the Cascade RCNN average score is 0.9801. Then, an algorithm to identify the center of bundled rebar heads based on the variable-scale method (Davidon–Fletcher–Powell formula (DFP)) is proposed. In addition, limiting conditions, and a process for selection of weldable points for double-label welding are provided. And the coordinate conversion equations of weldable points in different coordinate systems are derived. Finally, a serious of functional verification tests of the robot system were carried out on an actual rebar production line, which verified the accuracy of the weldable point image recognition process, and the usability of the robot system. The results indicate that the accuracy of weldable point recognition can reach 98.99%.

Two-tier massive multiple-input multiple-output (MIMO) systems achieve high sum spectral efficiency by simultaneously serving large numbers of users. However, as the number of service antennas and users tend to infinity, the performance is limited by directed inter-cell and intra-cell interferences. Handling these interferences are challenging due to the large channel dimensionality and the high complexity associated with implementing large precoding/combining matrices. Moreover, two-tier massive MIMO is computationally demanding, as the high antenna count results in high-dimensional matrix operations when conventional MIMO hybrid precoding is applied. In this paper, a manifold learning two-tier beamforming (MLTB) scheme is proposed to enable efficient and low-complexity operation in large scale dimensional MIMO systems. Users of multi-cell are clustered into several regions of user groups by manifold learning. Most of the high-dimensional channels are embedded in the low-dimensional subspace by manifold learning, while retaining the potential spatial correlation of the high-dimensional channels. The nonlinearity of high-dimensional channel is transformed into local linearity to achieve dimensionality reduction. Through proper user clustering, the beamformers are split into outer beamformers and inner beamformers. The outer beamformers can minimize inter-cell interference and the inner beamformers can minimize multi-user interference of intra-groups. The high signal to interference plus noise ratio (SINR) is achieved and the computational complexity is reduced by avoiding the conventional schemes to deal with high-dimensional channel parameters. Performance evaluations show that the proposed MLTB scheme can obtain near-optimal sum-rate and considerably higher energy efficiency than the conventional schemes.

Trailing edge flap (TEF) devices will change downstream wake development in wind farms comprising smart rotors, which in turn affect the performance of downstream wind turbines. To study the influence of TEFs on downstream wake development and power capture of wind turbines in wind farms, computational fluid dynamics (CFD) simulations using the three-dimensional rotor model are performed in this paper. The CFD software Fluent is adopted to simulate a wind farm with two tandem wind turbines with TEFs at rated and below rated turbulent wind conditions. Under the 11.4 m/s turbulent wind conditions, the results show that the deflection of the TEF increases the velocity deficit and reduces the wake width, making the wake more complicated. And the positive TEF angle has a greater influence on downstream wake than the negative TEF angle. Additionally, under the 9 m/s turbulent wind conditions, the total power of the two wind turbines increases by 6.5% when the TEF angles are 6°.

Event argument extraction, which aims to identify arguments of specific events and label their roles, is a challenging subtask of event extraction. Previous approaches solve this problem in a two-stage manner that first extracts named entities as argument candidates and then determines their roles. However, many nested entities may be missed or wrongly predicted during the argument candidate extraction procedure, which substantially affects the performance of the downstream classifier. In this paper, we propose a novel one-step question answering based framework, which performs argument candidate extraction and argument role classification simultaneously to mitigate the error propagation problem in conventional two-stage methods. Since the conventional question answering based framework cannot be applied directly to this task, we design a Question Answering based Sequence Labeling (QA-SL) model to tackle inexistent argument roles and multiple argument token spans. Moreover, considering the overwhelming number of parameters in question answering based neural network models and the relatively small size of event extraction corpus, we fine-tune the pre-trained model from BERT to mitigate the data scarcity problem. Extensive experiments demonstrate the benefits of the proposed method, leading to a competitive performance compared with state-of-the-art methods. To the best of our knowledge, this is the first work to cast event argument extraction as a question answering task.

In this paper, we have experimentally demonstrated a secure 100 Gb/s intensity-modulation and direct-detection transmission over 100 km standard single-mode fiber (SSMF) based on quantum noise stream cipher (QNSC) for the first time. The original 4-level pulse-amplitude modulation (PAM4) data is mapped to M-level data sequence randomly, and quantum noise such as the generated amplified spontaneous emission noise and the shot noise in the channel can mask these adjacent signal levels. The legitimate receiver needs to distinguish not M-level encrypted data but 4-level data with the shared key. In our scheme, a calculated detection failure probability of 98.72% is achieved. Apparently, more quantum noise contributes to higher security but the worse signal to noise ratio. To balance algorithm complexity, transmission performance and security performance, a sparse ℓ1 regularization based on recursive least square (RLS) algorithm is proposed and firstly used in Volterra equalizer. To eliminate the power fading induced by fiber dispersion, a dual-drive Mach-Zehnder modulator is used to achieve single-sideband modulation, and no dispersion compensation is required. By these means, 100 Gb/s PAM4-QNSC signal transmission over 100 km SSMF with the bit error rate (BER) below the 7% overhead hard-decision forward error correction threshold of 3.8 x 10-3 is achieved, and >59% complexity reduction of Volterra equalizer is realized. Moreover, the measured BER of 150 Gb/s PAM8-QNSC signal transmission over 50 km SSMF could go below the 20% overhead soft-decision forward error correction threshold of 2.0x10-2. The results validate that the proposed scheme is effective to realize low-cost, high-speed (>100 Gb/s), and secure optical fiber transmission in the data center.

Pellet quality including chemical composition, physical properties and metallurgical performance of three parts, its quality and mineral composition, properties and structure of the pellets has the close relation, studies show that the mineralogical micro structure and distribution of pellets had significant effects on the metallurgical properties, so the analysis and determination of pellets of mineral composition and micro structure is very important to improve the quality of pellets. Paper to pellets micro mineral as the research object, mainly studies the CNN and PCA two kinds of image processing algorithm, in the heart of the traditional model of CNN structure characteristics obtained by convolution PCA dimension reduction at a time, will be the main features of PCA to extract into the depth of the CNN learning, realize the mineral phase of shallow and deep information of the image to do effective fusion, in a larger extent, reflects the mineralogical characteristics, thus more intuitive response metallurgical properties. The location and alkalinity of the ore phase were identified by the extracted deep fusion feature, and the results were compared with those of the traditional CNN algorithm. It was found that the accuracy of location and alkalinity recognition of PCA and CNN coupling algorithm was 93.82% and 91.26%, respectively, which were higher than the accuracy of traditional CNN algorithm 92.73% and 88.93%, which verified the accuracy of PCA and CNN coupling model and its applicability in mineral phase recognition.

In this paper, the GaN-based MIS-HEMTs with Si3N4 single-layer passivation, Al2O3/SiNx bilayer passivation, and ZrO2/SiNx bilayer passivation are demonstrated. High-k dielectrics are adopted as the passivation layer on MIS-HEMTs to suppress the shallow traps on the GaN surface. Besides, high permittivity dielectrics passivated MIS-HEMTs also show an improved breakdown voltage characteristic, and that is explained by 2-D simulation analysis. The fabricated devices with high-k dielectrics/SiNx bilayer passivation exhibit higher power properties than the devices with plasma enhanced chemical vapor deposition-SiNx single layer passivation, including smaller current collapse and higher breakdown voltage. The Al2O3/SiNx passivated MIS-HEMTs exhibit a breakdown voltage of 1092 V, and the dynamic Ron is only 1.14 times the static Ron after off-state VDS stress of 150 V. On the other hand, the ZrO2/SiNx passivated MIS-HEMTs exhibit a higher breakdown voltage of 1207 V, and the dynamic Ron is 1.25 times the static Ron after off-state VDS stress of 150 V.

Interior permanent magnet synchronous motors (IPMSMs) have high power densities and speed control performance, and they are widely used in the industry. The problem of reducing the torque ripple of an IPMSM is one of the hot issues in the field of electrical machine design. In order to determine the optimal combination of the geometric parameters to reduce the torque ripple of an IPMSM, a range analysis was conducted on the data from the orthogonal experiments in this study, dividing the rotor geometric parameters into two categories (important and ordinary) based on their degree of impact on the torque ripples of the IPMSM. Thereafter, an optimization of the ordinary parameters was carried out based on the results of the range analysis, whereas the optimization of the important parameters was carried out through a method that combined a multi-island genetic algorithm (MIGA) and Radial Basis Function (RBF) neural networks. The torque ripple of the IPMSM was effectively reduced without materially affecting the output power. Finally, the results of this optimization process were verified using a finite element simulation. The optimization method used in this study divided the motor geometric parameters into two categories and applied a different method of optimization to each parameter type, so it was able to efficiently optimize multiple geometric parameters for the IPMSM.

This paper proposes a grid topology based passive optical interconnect (POI) architecture that is composed of multiple floors of arrayed waveguide grating routers (AWGRs) to offer high connectivity and scalability for datacenter networks. In the proposed POI signal only needs to pass one AWGR, and thus can avoid the crosstalk accumulation and cascaded filtering effects, which exist in many existing POI architectures based on cascaded AWGRs. Meanwhile, due to high connectivity, the proposed grid topology based POI also has the potential advantage of high reliability. Simulation results validate the network performance. With a proper node degree, the proposed grid topology can achieve acceptable blocking probability. Besides, steady performance is kept when the number of floors increases, indicating good scalability of the proposed POI.

Network embedding (NE) aims to represent network information appropriately by learning low-dimensional and dense vectors for the nodes and edges of information network. Actually, the real world is almost full of heterogeneous information networks, which stimulates the emergence of heterogeneous information networks (HINs) embedding models. However, parts of existing HIN embedding models like meta-path-based methods only capture limited and aggregated information of relations, whereas some models based on metric or distance learning are usually of high computational complexity and slow training speed. In this paper, we present a novel heterogeneous information network embedding model, which applies dynamic projection metrics and translation mechanisms to the complicated heterogeneous information networks including multiple nodes and different relations. In order to overcome the imbalance of the distribution of relations in HIN and optimize the training process, we introduce an adaptive loss function for model optimization. Further more, we propose a hybrid model with baseline method as the initialization of the model. Experiments have been implemented on some real-world HIN datasets. And empirical results show that our model significantly outperforms the state-of-the-art representation learning models.

Automatic extracting road intersection points is essential for applications such as data registration between vector data and remote sensing images, aircraft-assisted navigation. However, at a large scale, it is difficult to quickly and accurately extract road intersection points due to the problems caused by complex structures, geometric texture noise interference. In this context, taking OpenStreetMap (OSM) data as priori knowledge, we propose a method for automatic extraction of complex road intersection points based on fuzzy inference. First, OSM data are analyzed to obtain structural information of intersection points. Local search areas are built around the intersection points. Second, within the local search area, the candidate intersection point set are generated. Meanwhile the input image is segmented using multiresolution segmentation; then we establish a fuzzy rule to infer the road area from the segmentation result. The fuzzy indexes and rules are established for the candidate intersection point set to deduce the road intersection area. Finally, based on the results of the previous step, the road intersection points are extracted based on the line segment constraint, structure matching, and linkage equation. Three sets of high-resolution remote sensing images were used to verify the feasibility of the method. We demonstrate that the correctness and positioning accuracy of this method are superior to those of other methods through contrastive analysis.

Enormous computation and long data processing time of the online tolerance inspection procedure has been a bottleneck in reducing the production cycle time of automatic assembly line. To reduce time consumption while maintaining accuracy, a fast assembly tolerance inspection method is proposed in this paper using scale-reduced models. First, the general part surface, input by a laser-scanned dense point cloud, is segmented through feature-based clustering. Shape variation estimation are designed to determine reduction ratios between clusters adaptively. Then, assembly simulation method is further advanced to utilize coarse point models with imbalanced density. Finally, numerical experiments are designed to estimate both static and dynamic assembly tolerances as results of tolerance inspection. A case study on a motion component of a rail material transporter shows that the proposed method could reduce 81% of the data processing time compared to traditional practice. The precision loss has minor influence on the inspection results because spatial positioning error is no more than 6% of tolerance interval and 18% of scanning resolution. Moreover, the proposed method has a comprehensive advantage of time and accuracy than two conventional point cloud simplification techniques based on iterative and clustering methods.

As an interesting and challenging problem, generating image caption automatically has attracted increasingly attention in natural language processing and computer vision communities. In this paper, we propose an end-to-end deep learning approach for image caption generation. We leverage image feature information at specific location every moment and generate the corresponding caption description through a semantic attention model. The end-to-end framework allows us to introduce an independent recurrent structure as an attention module, derived by calculating the similarity between image feature sequence and semantic word sequence. Additionally, our model is designed to transfer the knowledge representation obtained from the English portion into the Chinese portion to achieve the cross-lingual image captioning. We evaluate the proposed model on the most popular benchmark datasets. We report an improvement of 3.9% over existing state-of-the-art approaches for cross-lingual image captioning on the Flickr8k CN dataset on CIDEr metric. The experimental results demonstrate the effectiveness of our attention model.

Interpretable multi-hop question answering requires step-by-step reasoning over multiple documents and finding scattered supporting facts to answer the question. Prior works have proposed the entity graph method to aggregate the entity information to improve the ability of reasoning. However, the entity graph loses some non-entity information that is also important to understand the semantics. Moreover, the entities distributed in the noisy sentences may mislead the reasoning process. In this paper, we propose the Coarse and Fine Granularity Graph Network (CFGGN), a novel interpretable model that combines both sentence information and entity information to answer the multi-hop questions. The CFGGN consists of a coarse-grain module to perform sentence-level reasoning and a fine-grain module to make entity-level reference. In sentence-level reasoning, the sentence graph is constructed to filter out the noisy sentences and capture the sentence features. In entity-level reference, a dynamic entity graph is used for the entity-level reasoning. We design a fusion module to integrate information of different granularity. To enhance the interpretability of the overall process, we calculate the reasoning score for each step and present the reasoning path from the multiple documents to the final answer. Evaluation on the HotpotQA dataset in the distractor setting shows that our method outperforms the published SOTA entity-based method in five out of six metrics.

With the rapid development of mobile networks and the proliferation of mobile devices, Spatial Crowdsourcing (SC) has attracted the interest of industry and research groups. In addition to considering the specific spatial constraints in the existing research spatial crowdsourcing, each task has an effective duration, operational complexity, number of workers required, and incentive budget constraints. In this scenario, we studied the MQC-TA (Maximum Quality and Minimum Cost Task Assignment) problem. Firstly, the worker incentive model is established. The MQC-GAC algorithm is designed according to the MQC-TA problem to maximize the task completion quality and minimize the incentive budget. The algorithm combined the fast convergence of Genetic Algorithm and the positive feedback mechanism of Ant Colony Optimization Algorithm. Finally, the effectiveness and efficiency of the proposed method are verified by a comprehensive experiment on the data set.

We not only collected 1102 samples form 3 different breast invasive carcinoma (BRCA) datasets, which included a TCGA dataset and 2 GEO datasets, but also collected 23 other cancer datasets, one of which included more than 100 samples. Using these datasets, we topologically inferred miRNA-mediated subpathway activity profiles, which integrated gene expression profiles, prior gene interaction information and target relations between miRNAs and genes, and topological information. Then we constructed the Global Directed Pathway Network (GDPN) with genes as nodes, and from 3 BRCA datasets and other 23 cancer datasets identified a set of miRNA-mediated subpathways that are survival-related risk markers. The results showed large activity values correlated with poor prognosis, such as hsa-miR-107 and hsa-miR-142-3p of BRCA datasets. We assessed the stability and robustness of the miRNA-mediated subpathway survival markers with 2 GEO datasets and 23 external independent datasets. The results showed that the proposed method can significantly reduce noise from sequencing errors and samples heterogeneity by integrating pathway topological information, and can break down the boundary of pathways and provide a new measure for detecting survival-related markers. The top miRNA-mediated subpathways are more reliable in stratifying high risk group and low risk group and selecting therapeutic strategies.

The high-dimensional and imbalanced data classification appears in many actual applications, but there are many problems encountered in practical operation. To overcome the disadvantages of kernel-based multilayer extreme learning machines (ML-KELM), enhanced kernel-based multilayer fuzzy weighted extreme learning machines (EML-KFWELM) has been proposed in this study. First, ML-KELM ignores imbalance learning, so we embed weighted strategy into ML-KELM to enhance the classification performance of the minority class. Meanwhile, we propose fuzzy membership to eliminate classification error caused by outlier and noise samples. Furthermore, we develop an enhanced grey wolf optimization (EGWO) method to perform the parameters optimization and improve the generalization performance of ML-KELM. In addition, the advantage of EML-KFWELM is that representation learning and classification can be integrated into a single learning process. Finally, computational comparisons with other state-of-the-art methods are performed on various real-world and gene expression data. Experimental results demonstrate that the proposed EML-KFWELM has good stability and can efficiently deal with the high-dimensional and imbalanced data.

Urban subway, because of its speed and punctuality, receives considerable popularity and injects great vitality into the city's economic development. However, overcrowding frequently observed in subway carriages during peak hours potentially leads to an increase in accident risk, while there are still some spaces available in buses in some situations. Such imbalance between the two public transits, which was examined based on the load ratios reported for Beijing public transits in a map App, motives us to develop a periodic stop-skipping strategy to shift the excessive demand from subway to bus to alleviate the overloaded situations while to reduce the unwanted influence on the passengers. To achieve such bi-objectives, we propose a non-linear model with consideration of the passengers' choice under the stop-skipping operation. To avoid to exhaustively examine all stop patterns, the stop patterns required to evaluate are identified theoretically. We develop a genetic algorithm to generate the optimal plan, with the triggering mechanism devised to give the priority to the safety objective. With the calibrated choice model, four sets of numerical experiments were performed to evaluate the proposed approach. The results indicate that our approach can effectively alleviate the overcrowding while the efficiency can be increased in some circumstances as the remaining bus capacity is utilized. The optimality and the efficiency of the algorithm and the effectiveness of the triggering mechanism were also verified. The proposed approach can be applied as an effective and economical way to rebalance the peak demand when the bus capacity is surplus.

Physical layer security is a cross research hotspot of information security and wireless communications. Recently, orbital angular momentum (OAM) technology has attracted widespread attention due to its potential of improving capacity and spectrum efficiency. However, there are few researches on system security based on OAM technology. In this work, we studied the security of the proposed OAM wiretap system by applying physical layer security theory to OAM communications. Firstly, we proposed a uniform circular array-based multi-mode OAM system and derived the secrecy capacity. Secondly, the phase profiles, channel capacity and received power of OAM beams are analyzed in terms of various system parameters in oblique scenarios at 30GHz. Thirdly, we discussed the security of the proposed OAM wiretap system. The results show that the system using vortex waves is superior to conventional communication systems using planar electromagnetic waves in terms of system security due to the inherent divergence and spiral phase structure of OAM beams. Moreover, we found that the distortion of OAM beams during transmission and the center shift at receiving end will occur and that system is more stable for high-mode OAM propagation in oblique scenarios. This work can provide an effective guidance for the research and application of OAM communications.

In the Industrial Internet of Things (IIoT), it is an urgent task to reduce power loss and enhance energy efficient through reasonable allocation of resources. Inspired by time slot channel frequency hopping, this paper puts forward a dynamic allocation model for time and power resources. Based on the proposed model, a dynamic resource allocation algorithm was designed to reduce energy consumption. In addition, a power and time allocation algorithm was developed to maximize the energy efficiency of the system. The workflows of the two algorithms were introduced in details. Simulation results show that both dynamic resource allocation algorithm could reduce the energy loss of the communication system, while ensuring the stability of the data queue. The research findings help to promote the performance of communication systems in different scenarios of the IIoT.

Vortex wave carrying orbital angular momentum (OAM), an important property of electromagnetic wave, has bridged a new way to achieve high-capacity radio-frequency communications using different OAM-modes. However, the system performance of OAM multiplexing with orthogonal-frequency-division multiplexing (OFDM) technologies has not been thoroughly investigated in multipath channel environment. In this paper, we first built up the uniform circular array (UCA)-based multi-mode OFDM-OAM communication system, where the reflected paths has been generated based on ideal specular reflector. Based on the multi-mode OFDM-OAM communication system model, we also derived the two-path OAM channel model with the direct and reflected paths. Then, we proposed a robust optimal power allocation algorithm by decomposing the whole optimization problem into inner-layer OAM-modes optimization and outer-layer OFDM subcarriers optimization in order to control the allocated power flexibly based on the channel quality differences between OAM-modes. The simulation results demonstrate that the proposed algorithm can significantly increase the channel capacity for multi-mode OFDM-OAM communications in multipath channel as compared with the traditional equal power allocation method and has a considerable low computational complexity than the binary search water-filling method when the number of OFDM subcarriers is large enough.