Search

Coal is an indispensable energy source for humans. As an important part of the mining industry, intelligent separation of coal and gangue will promote development. The traditional methods of recognition do not consider the interference created by a dynamic environment. There are many problems such as noise, complex backgrounds and occlusion, which lead to low accuracy and cannot satisfy real-time requirements in mining. Aiming at dynamic environments, a real-time multilevel fusion recognition system was built in this paper. First, we introduced a near-infrared camera into the field of separation, which was used to form a binocular system with a visible light camera. The SVM classifier was obtained by feature selection and fusion training of the binocular system, which overcomes the interference of environmental factors. Then, we proposed a new deep learning training method of two-sample fusion to improve the recognition network performance by expanding the number of samples and features. Finally, the SVM and deep learning algorithms were combined to establish a fast detection strategy. In addition, the length suppression algorithm was added to solve the occlusion problem. The accuracy of the fusion algorithm was 0.923 and the detection speed was increased to 26 fps. The experimental results indicated that the sorting system satisfied the requirements of real-time and robust of the coal industry.

International audience; This paper deals with unknown input interval observers for discrete-time linear switched systems. A new structure of interval observer is used to overcome the design difficulty of coordinate transformation often used in such context. The interval observer gains are computed by solving Linear Matrix Inequalities (LMI) derived from multiple quadratic Lyapunov functions (MQLF) under average dwell time switching signals.

This paper concerns interference estimation problem for device-to-device (D2D) communication networks. In the considered system, D2D users share common spectrum resources, such that the D2D links have interference with each other. To achieve effective interference management, it is necessary to have an accurate understanding of the interference relationship of the D2D devices, which is difficult since the locations and the number of mobile devices can vary over time. In this paper, we formulate an interference estimation problem for the D2D communication network where the D2D users change over time. Our objective is to get accurate estimations for the interference suffered by a generic D2D link and for the interference a D2D link introduces on other users. We propose a graph convolutional neural network (GCN) based estimation model which can estimate the interference of a D2D link based on the location information of the corresponding D2D pairs. Simulation results show the performance of our method.

In this paper, we propose a multi-agent deep reinforcement learning (DRL) based approach to solve the distributed Volt-VAR optimization (VVO) problem in distribution networks by considering the uncertainty of system load and renewable power generation. We formulate the distributed VVO problem as a non-cooperative Markov game. Specifically, we split the distribution network into multiple regions that are controlled by a group of networked agents. We consider the statuses/ratios of switchable capacitor banks (SCBs), the tap position of voltage regulators (VRs), and the reactive power of inverter-based distributed generators (DGs) as the control variables. The objective is to minimize the total system loss while maintaining bus voltages within a normal operating range. To solve the problem, a multi-agent trust region policy optimization (MATRPO) based approach is applied to learn a set of decentralized policies. Simulation results on a modified IEEE-34 test system show that the proposed approach can successfully learn a set of high-quality decentralized policies for the agents to collaboratively reduce the system loss and regulate the bus voltages. The simulation results also demonstrate the superiority of the proposed approach over independent learning and the MADDPG method.

This work proposes an event-triggered adaptive asymptotic tracking control scheme for flexible robotic manipulators. Firstly, by employing the command filtered backstepping technology, the “explosion of complexity” problem is overcame. Then, the event-triggered strategy is utilized which makes that the control input is updated aperiodically when the event-trigger occurs. The utilized event-triggered mechanism reduces the transmission frequency of computer and saves computer resources. Moreover, it can be proved that all the variables in the closed-loop system are bounded and the tracking error converges asymptotically to zero. Finally, the simulation studies are included to show the effectiveness of the proposed control scheme.

In this article, mean square consensus issue for linear discrete-time agent models in an undirected topology is explored. Assuming the neighbor's transmitted information is affected by constant delay and identically distributed packet dropout process, a distributed consensus protocol is constructed and the positive definite solution of a modified Riccati inequality is utilized to design control gain. After studying the joint influences of communication delay, agent models, network topology and packet dropout probability, sufficient criterion are produced to derive mean square consensus. Finally, a numerical simulation is given to confirm the feasibility of the acquired results.

Siamese network based trackers formulate the visual tracking task as a similarity matching problem. Almost all popular Siamese trackers realize the similarity learning via convolutional feature cross-correlation between a target branch and a search branch. However, since the size of target feature region needs to be pre-fixed, these cross-correlation base methods suffer from either reserving much adverse background information or missing a great deal of foreground information. Moreover, the global matching between the target and search region also largely neglects the target structure and part-level information. In this paper, to solve the above issues, we propose a simple target-aware Siamese graph attention network for general object tracking. We propose to establish part-to-part correspondence between the target and the search region with a complete bipartite graph, and apply the graph attention mechanism to propagate target information from the template feature to the search feature. Further, instead of using the pre-fixed region cropping for template-feature-area selection, we investigate a target-aware area selection mechanism to fit the size and aspect ratio variations of different objects. Experiments on challenging benchmarks including GOT-10k, UAV123, OTB-100 and LaSOT demonstrate that the proposed SiamGAT outperforms many state-of-the-art trackers and achieves leading performance. Code is available at: https://git.io/SiamGAT

This paper investigates the frequency regulation and active power sharing problem in AC microgrids with mobile power resource. Considering the impact of communication time-delays, a novel distributed finite-time secondary control method is proposed. The stability of the proposed control method is analyzed based on Lyapunov method. The upper bound of convergence time is derived and is only a function of the designed parameters, the topology of communication network, and communication time-delays. Numerical simulations are conducted to verify the performance of the proposed control method.

In this paper, fault detection is studied for discretetime switched systems subject to unknown but bounded disturbances and measurement noise. A fault detection observer with a new structure is investigated based on an L ∞ criterion to attenuate the effects of uncertainties. The design conditions are given in terms of Linear Matrix Inequalities using Multiple Quadratic Lyapunov Functions, under an Average Dwell Time switching signal. In addition, the FD results are obtained via an ellipsoidal analysis. Finally, a numerical example is performed to illustrate the effectiveness of the proposed approach.

In this paper, the consensus achievement for first-order multi-agent systems is researched over undirected graph. Suppose that the system is unstable and exactly knows the communication delay that affects the actually transmitted information, consensus gain in the protocol is designed with the delay information. Then, conditions are obtained to guarantee consensus for any large yet bounded communication delay. At the end of the article, an example is given to demonstrate the validity of the conclusion.

This paper studies the fault diagnosis of wind motors, which is an important way to improve the safety and reliability of wind motors. It is non-trivial to extract the fault features from the original vibration signals by the traditional methods. We propose a novel method to improve the fault diagnosis performances of wind motors. First, the Wigner-Ville distribution method is used to generate the time-frequency images of the vibration signals in different speed ranges of the motor, which is helpful for fault features extraction. Then, we use the convolutional neural network, an important tool in the field of deep learning, to extract the fault features from the time-frequency images. Finally, simulation results based on the measurement data of an actual wind motor are provided to demonstrate the effectiveness of the proposed method.

This paper presents a sensor fault diagnosis method for the satellite attitude determination system based on analytical redundancy and wavelet transform. Specifically, we design a full-order observer to achieve fault detection. After the fault is detected, wavelet transform is used to isolate the fault sensors. Moreover, the gyroscope faults are estimated by using the algebraic observability. The proposed fault diagnosis approach combines the analytical redundancy-based method and the signal processing technique, its major benefit is that the computational burden is effectively reduced by introducing the fault detection method based on the analytical redundancy. Simulation results illustrate the effectiveness of the proposed method.

HAZOP plays an important role in chemical safety, in the project of Chinese named entity recognition of HAZOP, In response to the problem of low accuracy of entity recognition and low efficiency of model training in the past methods, this paper proposes Albert-BiLSTM-CRF model. In the process of text pre-training, we use Albert model instead of word2vec, Bert, CNN and other traditional structures to train word vectors, and get word embedding at the character level related to context, which improves the ability of text feature extraction. The experiments show that: on the basis of the accuracy of entity recognition, the F1-score of this model on HAZOP test corpus reaches 91.09%, which is 2.27% and 2.36% higher than that of BiLSTM-CRF model and CNN-BiLSTM-CRF model, and is equivalent to Bert-BiLSTM-CRF model, with F1-score exceeding 90%. on the basis of the speed of model training, the average training time of each epoch is only 162 seconds, which is faster than the training speed of other models under the same configuration environment. The results reflect that the Albert-BiLSTM-CRF model can better solve the shortcomings of the previous models, not only to ensure high accuracy of entity recognition, but also to improve the training speed of the model. This model is more suitable for HAZOP in named entity recognition project.

UAV (Unmanned aerial vehicle) is widely used in high voltage transmission lines inspection presently. Aiming at the problem of lightning arrester counters automatic recognition in UAV inspection videos, this paper proposes an image recognition solution based on deep learning, and designs a lightning arrester counter automatic recognition system. The designed automatic recognition system detects and locates the region of the arrester counters in the UAV inspection video firstly, and then recognizes the meter reading in the located region. This paper focus on the target detection and location of arrester counter in UAV inspection videos. The results show that deep learning algorithm can accurately locate the regions of the arrester counters in the images, and lay a foundation for the subsequent meter reading recognition.

In this paper, we present Calibration Recurrent Convolutional Neural Network (CalibRCNN) to infer a 6 degrees of freedom (DOF) rigid body transformation between 3D LiDAR and 2D camera. Different from the existing methods, our 3D-2D CalibRCNN not only uses the LSTM network to extract the temporal features between 3D point clouds and RGB images of consecutive frames, but also uses the geometric loss and photometric loss obtained by the interframe constraint to refine the calibration accuracy of the predicted transformation parameters. The CalibRCNN aims at inferring the correspondence between projected depth image and RGB image to learn the underlying geometry of 2D-3D calibration. Thus, the proposed calibration model achieves a good generalization ability to adapt to unknown initial calibration error ranges, and other 3D LiDAR and 2D camera pairs with different intrinsic parameters from the training dataset. Extensive experiments have demonstrated that our CalibRCNN can achieve state-of-the-art accuracy by comparison with other CNN based methods.

In this paper, we investigate the economic dispatch problem in islanded microgrids and we propose a distributed finite-time algorithm to solve the economic dispatch problem. Different from the existing distributed algorithms which solve the economic dispatch problem asymptotically, the proposed algorithm can get the optimal solution in finite time and the bound of the convergence time only depends on the designed parameters in the proposed algorithm. The rapid convergence rate of the proposed algorithm makes it suitable for application in microgrids with penetration of large scale distributed generation resources. Numerical simulations are conducted to evaluate the performance of the proposed algorithm.

The smart grid is becoming increasingly reliant on communications systems for measurement and control data. A simple cyber-physical attack on the power grid is to introduce delays in the delivery of these measurement and control messages to the control centers of the grid. This paper investigates the consequences of delay using the IEEE 30 bus standard benchmark. The Power System Simulation software PSCAD is used to model the power system. A case study to introduce delays and line cuts and measure the line voltage is presented. The case study is designed to pinpoint the amount of delay that can be tolerated before the system becomes unstable.

By decomposing the visual tracking task into two subproblems as classification for pixel category and regression for object bounding box at this pixel, we propose a novel fully convolutional Siamese network to solve visual tracking end-to-end in a per-pixel manner. The proposed framework SiamCAR consists of two simple subnetworks: one Siamese subnetwork for feature extraction and one classification-regression subnetwork for bounding box prediction. Different from state-of-the-art trackers like Siamese-RPN, SiamRPN++ and SPM, which are based on region proposal, the proposed framework is both proposal and anchor free. Consequently, we are able to avoid the tricky hyper-parameter tuning of anchors and reduce human intervention. The proposed framework is simple, neat and effective. Extensive experiments and comparisons with state-of-the-art trackers are conducted on challenging benchmarks including GOT-10K, LaSOT, UAV123 and OTB-50. Without bells and whistles, our SiamCAR achieves the leading performance with a considerable real-time speed. The code is available at https://github.com/ohhhyeahhh/SiamCAR.

The weak adhesion of DLC films has always been a poser and a hot issuer in the previous researches of DLC films. In order to satisfy the cutting requirement of difficult-to-cut materials and adapt to the severe service conditions, the methods to improve adhesion must be more efficient and convenient. The Ti plasma etching treatment was carried out by mid-frequency magnetron sputtering method. A Ti-DLC film was deposited on the etching treated surface. The film thickness and surface roughness were measured by stylus profiler. The surface morphology was characterized by SEM. The composition of DLC films was characterized by micro Raman spectrometer. The adhesion strength was measured by liner-loading scratch tester. The etching treatment removed a micro layer of 407nm on the substrate and increased the surface roughness from 3.50nm to 18.26nm. The surface WC crystals were exposed and partly broken by Ti plasma. Compared with the DLC films on non-treated surface and Ti layer, the DLC films on etching treated surface had the highest adhesion strength of 34.14N. The DLC films had similar compositions, so the adhesion difference came from the surface treatments instead of the inherent properties of DLC films. The Ti plasma etching treatment removed a micro layer of substrate material and it brought higher chemical activity, which contributed to improve the adhesion. The increased roughness deferred the continuous spalling of DLC film in the scratch test. The changes worked together to improve the adhesion strength of DLC films.

The cyber-physical security of smart grid is of great importance since it directly concerns the normal operating of a system. Recently, researchers found that organized sequential attacks can incur large-scale cascading failure to the smart grid. In this paper, we focus on the line-switching sequential attack, where the attacker aims to trip transmission lines in a designed order to cause significant system failures. Our objective is to identify the critical line-switching attack sequence, which can be instructional for the protection of smart grid. For this purpose, we develop an evolutionary computation based vulnerability analysis framework, which employs particle swarm optimization to search the critical attack sequence. Simulation studies on two benchmark systems, i.e., IEEE 24 bus reliability test system and Washington 30 bus dynamic test system, are implemented to evaluate the performance of our proposed method. Simulation results show that our method can yield a better performance comparing with the reinforcement learning based approach proposed in other prior work.

This paper proposes a two-step interval estimation approach for discrete-time Takagi-Sugeno fuzzy systems with unknown but peak bounded uncertainties. A Takagi-Sugeno fuzzy observer is designed by using the L ∞ technique to attenuate the influence of the uncertainties and then improve the estimation accuracy. Based on the designed observer, the interval estimation results are determined via ellipsoidal analysis. By integrating the L ∞ observer design with the ellipsoidal analysis, the proposed method can obtain accurate interval estimation results with high computational efficiency. Finally, numerical simulations are presented to demonstrate the effectiveness of the proposed approach.

About a decade ago, people concerned about the risks of adopting cloud computing. It was an unproven new thing that raised more questions than it answered. Nowadays, we hear more about the risks of not adopting the cloud. Three of the leading cloud players, Amazon Web Services, Microsoft Azure, and Google Cloud Platform, and other participants have developed complex cloud platforms that are driving the cloud agenda and launching innovative new products to meet the needs of modern businesses. When looking at processors, core components of the cloud, there is a trend for hyperscale data centers is to move beyond the CPUs and turn to dedicated chips, such as graphics processing units, field programmable gating arrays, and application specific integrated circuits. We think it is an artificial intelligence (AI) realization process and provide a detailed survey about hardware server design in this process. After discussing and summarizing various disclosed techniques and platforms, we conceived a hybrid hardware structure for efficient AI applications.

OAPA In this paper, the robustness of model-based state observers including extended Kalman filter (EKF) and unscented Kalman filter (UKF) for state of charge (SOC) estimation of a lithium-ion battery against unknown initial SOC, current noise, and temperature effects is investigated. To more comprehensively evaluate the performance of EKF and UKF, two battery models including the first-order resistor-capacitor (RC) equivalent circuit and combined model are considered. A novel method is proposed to identify the parameters of the equivalent circuit model. The performance of SOC estimation is evaluated by employing measurement data from a commercial lithium-ion battery cell. The experiment results show that UKF generally outperforms EKF in terms of estimation accuracy and convergence rate for each battery model. However, the advantages of UKF over EKF with the combined model is not as significant as with the equivalent circuit model. Both EKF and UKF demonstrate strong robustness against current noise. The updates of model parameters corresponding to operational temperatures generally improve the estimation accuracy of EKF and UKF for both models.

In this paper, the load frequency control of multi-region power systems is studied. We propose a robust controller based on the frequency deviation between different regions. Using the non-smooth theory, we analyze the conditions for the feedback parameters of the proposed controller. Furthermore, the effectiveness of the proposed controller is evaluated on a four-region power system.

Treating graph structures of Markov random fields as unknown and estimating them jointly with labels have been shown to be useful for modeling human activity recognition and other related tasks. We propose two novel relaxations for solving this problem. The first is a linear programming (LP) relaxation, which is provably tighter than the existing LP relaxation. The second is a non-convex quadratic programming (QP) relaxation, which admits an efficient concave-convex procedure (CCCP). The CCCP algorithm is initialized by solving a convex QP relaxation of the problem, which is obtained by modifying the diagonal of the matrix that specifies the non-convex QP relaxation. We show that our convex QP relaxation is optimal in the sense that it minimizes the L1 norm of the diagonal modification vector. While the convex QP relaxation is not as tight as the existing and the new LP relaxations, when used in conjunction with the CCCP algorithm for the non-convex QP relaxation, it provides accurate solutions. We demonstrate the efficacy of our new relaxations for both synthetic data and human activity recognition.

A new on-line active fault detection method is proposed based on state set-membership ellipsoid in this paper. First, the system states are estimated based on $H_{\infty}$ observer method. To implement on-line fault detection, the sliding-window approach is used in this paper Then, an optimal auxiliary input signal is designed to guarantee the separation of normal state ellipsoids and faulty state ellipsoids. Finally, compared with the method of output set-membership ellipsoid, simulation results demonstrate the advantages of the proposed method.

In this paper, a fault analysis and prediction of aircraft equipment is proposed via the construction of association rules and the analysis of Weibull Distribution. Firstly, for the recorded data in the airborne detection system, the prefix path graph is introduced to improve the Apriori algorithm to further mine the association rules between the fault equipments with low and high supports. Then, the first failure time of the equipment with high support is recorded, and the reliability analysis model of the equipment is established according to the Weibull distribution. Finally, the results of reliability analysis model obtained from the association rules provide the theoretical basis for the maintenance program of aircraft. The experiments in aircraft airborne detection system verifies the effectiveness and practicability of the proposed method.

In this paper, the problem of sensor fault detection (FD) for linear discrete-time systems has been studied with the aid of multiple H-l H∞ observers approach. A limited number of fault modes can be obtained by considering the system only occurs one fault at a time. For each fault mode, an $H$ _ l H∞ fault detection observer (FDO) is designed with its residual is sensitive to fault and robust to disturbances and noises. Meanwhile, each FDO is optimized for the certain fault mode to improve the performance of fault detection. By assuming the disturbances and noises are unknown but bounded, zonotopic techniques are used to achieve residual evaluation based on the designed H-l H∞ FDQs. Finally, a second-order RC circuit simulation example is utilized to illustrate the effectiveness of the proposed method.

This paper analyses the activity of scientific and technological activities and the incidence of terrorist attacks in recent years, and studies the relationship between them, and puts forward corresponding countermeasures from three aspects: network domain prevention and control, airspace prevention and control, and regional prevention and control.