Search

Edge computing expands the Internet of Things (IoT) by allowing partial computing tasks to be migrated to edge servers and alleviate the computing pressure of terminal devices. However, the edge servers will be attacked through malicious network traffic. A well-designed edge-based IDS plays a significant role in protecting from malicious attacks. In this paper, we employ a gated recurrent unit classifier to perform intrusion detection due to its characteristics on long-term memory of inputs and light structure. Furthermore, to reduce the cost of performing intrusion detection when facing a large volume of data, we propose an actor-critic network based on deep reinforcement learning for packets sampling to work on some packets by ignoring the others. The system we designed can achieve great classification performance through partial packets. We use dataset CIC-IDS-2017 to evaluate our model, and the accuracy of classification reaches 97% while the proportion of detection packets is 26% and below. Evaluations of our approach on UNSW-NB15 and CIC-IDS-2017 show that our model maintains under 26% and 18% of the selected packets, respectively.

Fault diagnosis of the subway plug door plays an essential role in the safe operation of the city subway. To improve the diagnosis accuracy of the subway plug door, fault diagnosis of the plug door based on Ensemble Empirical Mode Decomposition (EEMD) and adaptive feature extraction was presented in this paper. Firstly, EEMD was used for decomposition of raw data, and the intrinsic mode function (IMF) after decomposition was selected by correlation coefficient criteria. Then, the fault features in IMFs was extracted and the sensitive features among which was selected by the sensitive index. Finally, the faults were classfied by Gray Wolf optimized Support Vector Machine (GWO-SVM). The experiment with the measured data of a subway door shows that this fault diagnosis method can adaptively extract the relative optimal characteristic quantity, identify the normal state and four different fault states effectively with the recognition accuracy of 89.35%, which is valuable in the engineering application.

In view of the problem that the characteristic extraction of subway plug door fault diagnosis is too high, which leads to low diagnostic accuracy, a mixed feature selection method based on ReliefF algorithm and BGWO (binary grey wolf optimizer, BGWO) was proposed. Firstly, multiple domains feature extraction were carried out on the collected current signal of the subway plug door motor, and an original fault feature set describing the subway plug door fault was obtained. Afterwards, the ReliefF algorithm was used to evaluate the extracted original fault feature weights and screened out the less relevant features. Finally, the classification error rate of GWO (grey wolf optimizer, GWO)-SVM (support vector machine, SVM) is used as the fitness value, and BGWO is used as the feature selection algorithm to perform feature selection on the feature subset obtained by the ReilefF algorithm. The data collected in a Metro Depot in Jiangsu Province is used as the original data set for verification. The experimental results show that the method can screen out low dimensional fault feature sets with high correlation, low redundancy and high fault identification, it can effectively improve the accuracy of subway plug door fault diagnosis.

An IGBT reliability model based on Saber software which can reflect the aging process of IGBT is proposed in this research in order to better reflect the changes of multi-feature values in IGBT reliability model. At first, the coupling relationship among electrical, thermal and mechanical of IGBT is deeply analyzed, and then an IGBT reliability evaluation model based on Saber software is established according to the failure mechanism of IGBT bond line, and its static and dynamic characteristics are verified. The simulation results are in good agreement with the experimental results. The effects of collector current $I_{c}$ and gate driving voltage $U_{ge}$ on the saturation voltage drop of IGBT are verified by the built model. The conclusions are as follows: 1) the saturation voltage drop increases continuously with the deepening of the aging degree of IGBT. 2) the saturation voltage drop is affected by the collector current and driving voltage passing through IGBT. The study of this model has a certain guiding significance for realizing the reliability evaluation of IGBT and improving the reliability of power electronic power conversion devices.

The temperature field distribution of IGBT module will be inevitably affected by the unavoidable void of solder layer in IGBT module packaging, which will seriously affect the thermal stability. In order to reduce the influence of the void in the encapsulation of IGBT module, the IGBT failure process is analyzed based on ANSYS. In this paper, a three-dimensional model of the IGBT module is established. Then, the influence of solder layer void on the temperature distribution of IGBT module under different sizes and positions is studied based on ANSYS. The results show that under the same void ratio, the void on the top corner of the solder layer has the greatest impact on the junction temperature of the IGBT module. Besides, at the same position, the maximum junction temperature of the IGBT module increases linearly with the increase of the void ratio on the solder layer. The simulation results have guiding significance for improving the thermal stability of IGBT modules.

In the era of information explosion, how to effectively identify the named entities and the relationships between them from the massive judicial data is a key step to realize automatic trial, and it is also the focus of this paper. The traditional way is to use the pipeline method to identify named entities first, and then extract relationships, which will lead to the loss of related information between tasks and ignore the overlapping relationships between entities. For this reason, this paper designs a joint extraction algorithm for entities and relations in the judicial field. It uses the BiLSTM network model based on the BERT pre-training language algorithm and combines it with the attention mechanism, realizes joint learning through parameter sharing, and makes full use of the connections between tasks to optimize the results. In order to verify the effectiveness of this method, this paper constructs a small judicial data set to verify its generalization ability. Experimental results show that the model proposed in this paper can extract entities and relationships without manually defining complex features, and the average F value can reach 0.7.

Aiming at the problem of inaccurate estimation of state of charge (SOC) in lithium battery applications, this paper proposes a SOC estimation model based on Grey Wolf Optimizer (GWO) and BP neural network. This model takes the terminal voltage and discharge current of the lithium battery as the input terms of the neural network and the battery SOC as the output term. The traditional BP diagnostic model and the new model are trained and tested by using the test data at different discharge rates. Experiments with measured data indicate that compared with the traditional BP neural network, the estimation model proposed in this paper has higher accuracy of SOC estimation and smaller relative error.

The parameters of the LTO battery model are quite different between low temperature and room temperature conditions, which causes unacceptable errors in existing State of Charge (SOC) estimation model at low temperatures. In order to establish a precise LTO battery model, which can be applied to the State of Charge (SOC) estimation at both room temperature and low temperature, a method for identifying parameters of LTO battery model under low temperature is proposed. Firstly, the terminal voltage data during the discharge process of LTO battery at different temperatures are obtained through the test. Secondly, the data is imported into the Cuckoo Search with other optimization algorithms aiming at identifying the parameters of the second-order RC model at different temperatures, which are then applied to the model to obtain the estimated terminal voltage curves. Finally, the obtained estimated terminal voltage curves are compared with the actual values on curves, which shows that Cuckoo Search owns a higher accuracy and its identification results are applicable to the State of Charge (SOC) estimation of LTO battery within a wide range of temperature, especially at low temperatures.

The gradual improvement of global navigation satellite systems in recent decades and their combination greatly increase the visible satellite number for positioning and integrity monitoring process. This enhances the positioning accuracy and availability of the navigation system but at the cost of the large growth in computation load and function failure rate. Satellite selection is a useful way to relieve the side effect of superabundant measurements in use and has been developed for geometric dilution of precision (GDOP) optimization, but it has scarcely been studied in the integrity monitoring field. This paper proposes a slope-based fast satellite selection algorithm with the motivation to optimize the horizontal protection level (HPL) of subset while maintaining an acceptable positioning accuracy. The basis of this algorithm is a cost function which illustrates the relationship between satellite geometry and the slope effect. While the principle is to find the max slope satellite in the subset and introduce a new satellite which can significantly reduce the max slope by the cost function so the HPL decreases. Comparison simulations show the effectiveness of the proposed algorithm in HPL reduction, positioning accuracy maintenance and computation load mitigation, which means it is more suitable for real-time multi-constellation navigation applications.

Air passenger flow can show the spatial connection among cities. Based on the ticket data of major airports in Beijing-Tianjin-Hebei region, this paper systematically explored the temporal and spatial distribution characteristics of passenger flow in this region, drew the OD map among this region and other cities, and analyzed regional differences and connection characteristics among cities. Experimental results show that the passenger flow of airports in the Beijing-Tianjin-Hebei region is uneven in time and space. In terms of space, the regional distribution of air passenger flow is extremely unbalanced. The air transportation status in the east is extremely prominent, while the western part is relatively strong, and the central and northeast parts are relatively weak. In terms of time, the major airports in the Beijing-Tianjin-Hebei region have obvious peak and low peak periods, and the distribution of passenger flow direction in different time periods is also slightly different. The research results provide a basis for the analysis of population flow and economic relationship among urban agglomerations, and provide reasonable suggestions for traffic planning and passenger travel.

In allusion to residual chi-square test method has low sensitivity for soft fault detection and cannot isolate the faulty system in integrated navigation system, a redundant INS/ADS fault detection and isolation method based on Dempster-Shafer (D-S) evidence reasoning is proposed in this paper. In this method, three INS/ADS and three INS/GPS sub-filters are constructed respectively by introducing intact GPS in the three redundancy INS/ADS system, and the difference between the residuals of each two sub-filters is used as evidence, then the fault detection of sub-filters is completed by using D-S evidence reasoning. On this basis, the fault isolation and system reconstruction methods are given, and the fault isolation of the INS/ADS integrated navigation system is realized. The simulation results show that the proposed method can not only improve the sensitivity of soft fault detection, but also isolate the fault system in integrated navigation system.

For the problem that INS/GPS/ADS integrated navigation cannot accurately isolate faults, this paper establishes the chi-square detection models of INS/GPS and INS/ADS and the least squares detection model of GPS/ADS. A new isolation method is designed using the standard residuals component, which solves the position drift problem caused by isolating faulty GPS. By analyzing the diagnostic features of each detection model, the combined precise fault isolation method is proposed. Simulation experiments which take into account faults of different sizes and different duration show that the proposed method can effectively achieve the precise fault isolation.

New studies on grid-connected voltage-source inverters (VSIs) show that sequence-admittance-based stability analysis model cannot be simply simplified as decoupled positive-sequence and negative-sequence impedance ratio because of sequence-couplings resulting from the outer asymmetric power control loops, such as the phase locked loop (PLL) and direct voltage control. Additionally, compared with single grid-connected VSI system, parallel-connected VSIs have stronger circuit-couplings via grid impedance. To get accurate stability conclusions in wider frequency range, both couplings are considered in this paper, and this paper firstly derives a sequence-admittance-based stability analysis model to capture resonances in parallel-connected VSIs. The proposed model is described as open-loop model and closed-loop model, to which Nyquist stability criterion is effectively applied instead of the complicated generalized Nyquist stability criterion. Take the PLL-induced low-frequency resonance for example, stability analysis of the studied cases is successfully conducted. Finally, time-domain simulations verify the accuracy and simplicity of the proposed model.

This paper proposes a multi-ring BPSK/QNSC by expanding influence of amplitude noise on signal to enhance security. The simulation shows that SER of running key increases. Besides, the SER of legal receiver decreases by 2dB.

An auxiliary graph is constructed to describe the nodes adjacency in different levels influenced by distance, based on which a routing, wavelength and time-slot assignment algorithm is proposed and evaluated in metro quantum optical networks.

In the microgrid, the virtual synchronous generator (VSG) technology realizes the friendly access of the distributed power supply and improves the stability of the power system by simulating the operating mechanism of the synchronous generator (SG). However, the VSG inverter has a large impact current at the moment when it is switching from off-grid to on-grid mode. For this reason, this paper proposes a pre-synchronization control strategy to solve this problem. Firstly, the principle of VSG inverter is introduced, the switching process between off-grid and on-grid mode is analyzed, and the cause of the impact current when connected to the grid is also discussed. Then, the pre-synchronization control strategy based on PLL including frequency compensation and amplitude compensation is designed to realize the seamless and smooth switching of VSG inverter from off-grid to on-grid mode. Finally, simulation verification is conducted through MATLAB/Simulink.

In recent years, the automobile industry has developed rapidly and the drivers’ unregulated driving behaviors have caused a lot of traffic safety problems. By using the topic model, this paper can find the relationship between driving pattern with driving behavior and provide technical support for traffic safety. Taking the shortage of the current mainstream topic models pLSA and LDA (Latent Dirichlet Allocation) into account, this paper proposes an improved LDA model with time labels, namely the T-LDA model to identify driving patterns. The experimental results show that the improved model can effectively extract a series of continuous driving behavior characteristic and improve the accuracy of driving pattern recognition.

Electric Vehicles (EVs) have developed greatly under the government supporting to relieve energy crisis and pollution issues in recent years. However, there still a problem to restrict further development of EVs is that how to efficient and fast charging for them. So, a kind of owns high-output-power and high-efficiency charging module may be a useful key to solve the problem. However, the conventional charging modules with using Silicon (Si) devices couldn’t own high efficiency, because there are a lot of power losses caused by power semiconductors and magnetic devices. Hence, reducing the amount of power semiconductors and magnetic devices to decrease power loss is a useful way to realize high efficiency. Therefore, according to the idea, this paper plans to design and implement a 15kWDC/DC converter charging for EVs with LLC resonant topology and Silicon-Carbide (SiC) devices to realize high efficiency, the proposed converter could reach to about 98.4% which calculated at rated full-load condition. And some detailed comparison between the conventional converter and proposed converter about conversion efficiency and other aspects shown at the section-II in this paper.

Eelectric Vehicles $(EV_{s})$ have developed fastly supported by governments to relieve pollution issues and energy crisis in recent years. However, with the continuous devlopment of $EV_{s}$, efficient and rapid charging will restrict its further development. So a kind of charging module which owns high-output power and high-efficiency is a feasible key to solve the charging problem. However, conventional charging modules with using Silicon (Si) devices couldn’t realize very high efficiency, as a lot of power losses caused by power semiconductors and magnetic devices. So reducing power loss of these devices is a useful way to realize very high efficiency. Hence, this paper plans to design a 15kW DC/DC converter with LLC topology and Silicon-Carbide (SiC) devices to reduce power loss and verify its high efficiency. The proposed converter could reach about 98.4% and higher about 3% than conventional converter at rated full-load condition. Power density and volume of this proposed converter increases 27.6% and decreases 21.9% than conventional converter, respectively. Other detailed comparison information about this two converters shown as Part-D of section-II in this paper. In the end, relaive experimental waveforms used to verify the correctness and rationality of design.

Aiming at the measurement requirement of fast and accurate measuring precision for Individual Soldier Fire-control Equipment in laboratory, an echo laser simulator based on Phase Locked Loop technology is presented in this paper. The simulator adopts Positive Input Negative to receive the narrow pulse-width laser emitted by the fire-control equipment to complete the laser acquisition and photoelectric transformation. The simulator uses PLL module in FPGA to carry out 15 times frequency operation and high precision delay of electric signal. The laser radiation module uses MOSFET’s switch nature to produce the narrow pulse signal, the function of the radiation module is to modulate laser and simulate the echo laser in the actual measurement. The simulation distance of the simulator is calibrated by high frequency oscilloscope. The results of multiple measurements show that the simulator has high stability and the range simulation accuracy can reach 0. 2 m, the repeatability error is ±. 015 m.

This paper deals with the synchronization problem for complex networks under the passivity framework. In particular, the randomly occurring coupling delays are taken into account to describe more complicated dynamics in the real world. By means of stochastic analysis, delay-dependent sufficient synchronization criteria can be given for ensuring the desired passivity performance by matrix techniques. With the established conditions being satisfied, the designing procedure for the synchronization controller is proposed. Finally, we give an illustrative example for illustrating the efficiency of our synchronization strategy.

LLC resonant converters have been widely used in DC-DC conversion applications due to the advantages of high efficiency and high power density. There are many design methods of LLC resonant converters in former research. However, none of them take transformer parasitic capacitance into consideration. This paper proposes an optimal design method considering transformer parasitic capacitance. Firstly, a brief summary of design process is introduced. Next, a wire-wound transformer is designed and the transformer parasitic capacitance is measured. After that, the resonant parameters are selected aiming at reducing device conduction loss and satisfying gain requirement. Finally, experiments are conducted on a 400V-12V 200W GaN-based LLC resonant converter. Experimental results show that the optimal design method has good performance and the efficiency of the prototype is up to 94.2%.

The pedestrian navigation system (PNS)based on pedestrian dead reckoning (PDR) is a kind of navigation and positioning system which relies on pedestrians themselves. It can determine the pedestrian's current position real-timely. In this paper, we consider pedestrian navigation system in devices equipped with the foot-mounted IMU, where we designa pedestrian dead reckoning algorithm based on step size estimation. The peak detection algorithm is used for the gait recognition and step frequency detection. PDR contains gait detection algorithm, step size estimation algorithm and the heading angle provided by magnetometer and Inertial navigation system (INS). In order to reduce the error, we consider the Kalman filter fusion to achieve the heading information. Since pedestrian movement is often a combination of walking and running gait, we propose a navigation scheme of walking / running combination, and prove the feasibility of the scheme by walking / running combination experiment.

Conduction loss, which is greatly affected by transformer stray capacitance, has a huge influence on the efficiency of LLC resonant converters. The reason is that different transformer stray capacitance means different winding charge or discharge during the dead time, and then the magnetizing current and the dead time are changed accordingly. In order to investigate the impact of transformer stray capacitance on device conduction loss in LLC resonant converters, this passage is arranged as follows. Firstly, the transformer stray capacitance model is established and several methods of calculating the transformer stray capacitance are introduced. Next, the relationship between transformer stray capacitance and device conduction loss in LLC resonant converters is derived. Finally, transformers with different stray capacitances are applied in a 1.1MHz 400V–12V 200W GaN-based LLC resonant converter. The loss analysis shows that lower transformer stray capacitance contributes to smaller dead time and higher efficiency.

In low-voltage ride-through or unbalance gird voltage condition, the fundamental fluctuation in the neutral-point (NP) voltage of three-level voltage source converter will appear under imbalanced grid voltage situation and has not been researched in detail. A new NP current model under imbalanced grid is developed to investigate the limitations of conventional nearest-vector modulation and to give a more intuitive insight to solve this problem.

This paper presents an integrated solution of flexible DC-DC converter by embedding a flexible polyimide printed circuit (FPC) board and an inductor made of flexible ferrite-polymer composite in a wire. The cooper in the wire is utilized as the winding to make the embedded inductor with flexible ferrite-polymer sheets. Different wire shapes of the inductor are simulated and compared to optimize the design. A step-up prototype is fabricated and corresponding tests are executed to verify the performance. The flexible wire-embedded converter can output 5V-1W with variational input, and achieve 70% efficiency.

As an important bearer communication network for smart power distribution and utilization sever, the terminal communication access network should satisfy the various quality of service (QoS) requirements. And the queue scheduling is one of the key technology in multi-service QoS support. In this paper, a queue scheduling algorithm based on buffer size is proposed to minimize the waiting time of emergency service. Simulation results show that the proposed algorithm could reduce the delay and packet loss significantly.

Two sequence impedance models of LCL-type grid-connected three-phase inverters with different current sampling schemes are presented using the harmonic linearization method, respectively. The proposed positive- and negative-sequence impedance models are then applied to analyze the stability and resonance between the converter and the grid. An impedance-based analysis approach is used to characterize such resonance caused by impedance interactions under grid-side current (GC) feedback and converter-side current (CC) feedback, respectively. Detailed circuit simulation is built to validate the accuracy of the impedance models as well as their effectiveness applied to the comparative analysis of stability margin of different current sampling schemes quantitatively.

In order to meet the demands of high efficiency and a wide adaptable range of input voltage, a DualOperation Mode half-bridge LLC converter which can operate in two different modes according to the level of input voltage is obtained by adjusting the driving control strategy of 3-level LLC resonant converter, and the adaptable range of input voltage is doubled with the same parameters of resonant tank. The limiting factors on adaptable range of input voltage are defined by analyzing the voltage gain characteristic and power loss of LLC converter. After analyzing the Dual-Operation Mode of 3-level LLC converter and its influence on the adaptable range of input voltage, the driving control strategy is defined. The characteristic about wide input adaptability and high efficiency of the Dual-Operation Mode half-bridge LLC converter are verified through simulation and experiments.

This paper systematically studies a modified Lotka-Volterra differential algebraic predator-prey system, which is combined with nonlinear harvesting in prey and gestation delayed for predator. According to the stability and bifurcation theorem, through considering gestation delayed as a bifurcation parameter, the occurrence of Hopf bifurcation of proposed system is shown. Moreover, the nonlinear state feedback controller is designed to control the Hopf bifurcation, and the population of prey and predator can be driven to steady states by adjusting harvesting costs and the economic profit. Lastly, numerical simulations illustrate the effectiveness of the results obtained here.

With the rapid development of social media, research about its application and organization has attracted considerable attention. Many social media sharing websites allow users to tag the uploaded images with tags. However, the tags are usually uncontrolled, ambiguous, and overly personalized. One of the most challenging issues is how to match the textual tags with the visual image to rank tags accurately. Because a great deal of social images are stored and transmitted in a compressed format, an optimized tag ranking based on visual vocabulary in compressed domain is proposed in order to improve the accuracy of tag ranking and further reduce the ranking time. First, the low-resolution social images are constructed from the compressed image data. Then a visual vocabulary is created after the SIFT descriptor is extracted. Finally, the neighbor voting model is utilized to rank the image tags. Experimental results show that the proposed optimization method can significantly reduce the time of image tag ranking under ensuring the ranking accuracy of social image tags.

For the network environment with the limited transmission capacity, a multi-nodes image retrieval method based on visual words is proposed. Firstly, the visual words of query image are built by using the K-means clustering method after the color features and SIFT features of query image are extracted. Then the visual-words histogram of the query image is carried by the mobile Agent. The image similarity in each node is measured with the Euclidean distance. The results of image retrieval are returned to the user. Finally the weight-based relevance feedback method is utilized to optimize the retrieved results. Experimental results show that the proposed method can improve the speed and the accuracy of the image retrieval.

A warehouse management system(WMS) based on QR Code is designed to improve the efficiency of warehouse management. In this paper, the design of WMS is given, including the working principle and the development environment. After the introduction of the structure of the system, several problems have been discussed, then the solutions are given and the relevant code is post and analyzed.

This paper considers the benefit of joint beamforming across multiple cells in a MIMO (multiple-input multiple-output) system, where multiple base-stations may jointly optimize their respective beamformers to improve the overall system performance. This paper focuses on the criteria of minimizing the total weighted transmitted power across the base-stations subject to signal-to- interference-and-noise-ratio (SINR) constraints at the remote users. The proposed algorithm is based on uplink-downlink duality to the multicell setting. Simulations reveal that our algorithm provides appreciable performance improvements as compared to the conventional per-cell optimized one.

Different from previous work, the study reported in this paper attempts to simulate a more real and complex approach for region of interest (ROI) detection and quantitatively analyze the correlation between human visual system (HVS) and ROI. In this paper, an approach of ROI detection based on visual attention and gaze tracking is proposed. The works include pre-ROI estimation using visual attention model, gaze data collection and ROI detection. Pre-ROIs are segmented by the visual attention model. Since eye feature extraction is critical to the accuracy and performance of gaze tracking, adaptive eye template and neural network is employed to predict gaze points. By computing the density of the gaze points, ROIs are ranked. Experimental results show that the accuracy of our ROI detection method can be raised as high as 97% and our approach can efficiently adapt to users' interests and match the objective ROI.

Ray casting algorithm is one important method to render CT volumes with 2D CT slices, and it can be accelerated using graphics hardware. In this paper, we developed an efficient volume rendering system based on ray casting using C++ and VTK, and designed common shading and classification transfer functions to highlight different organs/tissues of a CT volume. With this system we experimented CT volumes using CPU ray casting and GPU ray casting respectively, and then we did the same experiments on different graphics cards. The results showed that we can highlight the organs/tissues in which we are interested by setting proper transfer functions and that the rendering efficiency can be accelerated greatly using graphics cards with nVidia GPU GT220 or higher almost with the same visualization quality as that of CPU ray casting.

In this paper, we study minimal power transmission for uplink systems of LTE with constraints on mean queuing delay. Power consumption is important concern for multimedia services of mobile station of LTE .The power minimizing scheduler adapts power and rate of transmission based on the queue and channel state. Our extensive simulations show that small increases in average delay can lead to substantial savings in transmission power, thereby providing another avenue for mobile devices to save on battery power. We propose a low-complexity scheduler that has near-optimal performance.

A low power reconfigurable DCT architecture is proposed, which can be run at three transform precision levels for different demands. Using the character of energy distribution of the DCT matrix after 2D DCT operation, we selected the best DCT bases which achieve considerable power reduction in DCT operation with minimum image quality degradation. The reconfigurable architecture can achieve power saving ranging from 22% to 41% for different precision levels and very useful for image video compression and transmission applications.