Search

With the development of technology in modern society, more and more electrical and electronic instruments has been invented and used in factories and our daily life. Power supply is the basic and important need of any electrical and electronic appliance, but energy storage technologies for mobile use is far from satisfactory. For instance, one of the most important performance of Electric Vehicle(EV) is the range ability. Because of the constraints of battery capacity and charging consumes a long period of time, range anxiety is the fatal shortcoming of EV.Wireless power transfer(WPT) technology could alleviate the shortcoming of range anxiety to a large extent. With the WPT system’ spread, the charging process could be happening wherever and whenever the EV is parked.In the research area of WPT system, circuits design, coil design, compensation topology and control methods are the four most important parts. Improving the power transfer capacity(including power capacity and the range of transfer distance), system’s reliability and efficiency of the system are the main purposes of this research. By applying medium-frequency and optimized coil design as well as specially designed control strategy, a medium-frequency high-power WPT system is realized.

A novel single stage LLC resonant converter with high power factor is proposed in this paper. The basic principle of single stage LLC resonant converter with high power factor is analyzed. The design consideration of this converter is illustrated for ensuring the voltage gain of the converter always meeting the requirements of high power factor during half line cycle. Finally, a design example of 100W single stage LLC converter prototype is established to verify the theoretical analysis.

The problems of power supply voltage quality, system losses and power demand are increasing serious, load voltage characteristics need to be studied to solve these problems. The time-varying ZIP model can represent voltage dependence of loads and capture the load characteristics. This paper proposes the improved recursive least square (RLS) algorithm which based on cumulant and variable forgetting factor to estimate time-varying ZIP model parameters, the proposed method can suppress the influence of noise and track the change of parameters quickly, and the results of simulation show the method has good tracking performance and strong robustness.

AC/DC hybrid distribution network gains increasing attention due to the characteristics of high efficiency, high reliability and high quality. With the rapid development of power electronic technology, energy router is regarded as a flexible equipment with multiple ports, which is convenient to participate in power and voltage control. The topological structure of energy router is developed to build AC/DC hybrid distribution network flexibly. Then the control strategy for single energy router and multiple energy routers is proposed to address closed loop operation in short time scale in AC/DC hybrid distribution network, which aims to control energy routers actively to stabilize DC voltage and follow load demand. The effectiveness of the proposed control strategy is verified via simulation with multiple operation conditions.

In this paper, we study the time-domain neural network approach for speech bandwidth extension. We propose a network architecture, named multi-scale fusion neural network (MfNet), that gradually restores the low-frequency signal and predicts the high-frequency signal through the exchange of information across different scale representations. We propose a training scheme to optimize the network with a combination of perceptual loss and time-domain adversarial loss. Experiments show the proposed multi-scale fusion network consistently outperforms the competing methods in terms of perceptual evaluation of speech quality (PESQ), signal to distortion rate (SDR), signal to noise ratio (SNR), log-spectral distance (LSD) and word error rate (WER). More promisingly, the multi-scale fusion network requires only 10% of the parameters of the time-domain reference baseline.

Currently, low signal-to-noise ratio (SNR) and non-stationary noise cause severe performance degradation for most of speech enhancement models. For better speech enhancement at the above scenarios, this paper proposes a two-stage approach that consists of binary masking and spectrogram inpainting. In the binary masking stage, we first obtain binary mask by hardening soft mask and then use it to remove time-frequency points that are dominated by severe noise. In the spectrogram inpainting stage, we use a CNN with partial convolution to perform inpainting on the masked spectrogram from the previous stage. We compared our approach with two powerful baselines, including Wave-U-Net and CRN, on a low SNR dataset containing lots of non-stationary noises. The experimental results show that our approach outperformed the baselines and achieved the state-of-the-art performance.

Optical character recognition (OCR) has been widely studied in previous work. Except for the models used, the recognition accuracy depends most on the resolution of the image to be recognized. To enhance OCR performance, this paper proposes an approach based on a generative adversarial network to improve text image resolution. Our approach uses a perceptual loss function that consists of an adversarial loss, a content loss and an L1 loss. The adversarial loss and the L1 loss are used to ensure the generated super-resolved images are closer to the ground truth high-resolution images. Meanwhile, the content loss is used to ensure the generated super-resolved images and the input low-resolution images have similar features on the basis of perceptual instead of pixel similarity. To evaluate the proposed approach, we compare the recognition accuracies before and after improving the resolution of both English and Chinese text images. The results show that the recognition accuracies on the super-resolved text images obtained with our approach are significantly higher than those on the low-resolution images without processing.

The rise in popularity of long-lived applications (LLAs), such as deep learning and latency-sensitive online Web services, has brought new challenges for cluster schedulers in shared production environments. Scheduling LLAs needs to support complex placement constraints (e.g., to run multiple containers of an application on different machines) and larger degrees of parallelism to provide global optimization. But existing schedulers usually suffer severe constraint violations, high latency and low resource efficiency. This paper describes Aladdin, a novel cluster scheduler that can maximize resource efficiency while avoiding constraint violations: (i) it proposes a multidimensional and nonlinear capacity function to support constraint expressions; (ii) it applies an optimized maximum flow algorithm to improve resource efficiency. Experiments with an Alibaba workload trace from a 10,000-machine cluster show that Aladdin can reduce violated constraints by as mush as 20%. Meanwhile, it improves resource efficiency by 50% compared with state-of-the-art schedulers.

This paper proposes an attention-based neural network approach for single channel speech enhancement. Our work is inspired by the recent success of attention models in sequence-to-sequence learning. It is intuitive to use attention mechanism in speech enhancement as humans are able to focus on the important speech components in an audio stream with "high attention" while perceiving the unimportant region (e.g., noise or interference) in "low attention", and thus adjust the focal point over time. Specifically, taking noisy spectrum as input, our model is composed of an LSTM based encoder, an attention mechanism and a speech generator, resulting in enhanced spectrum. Experiments show that, as compared with OM-LSA and the LSTM baseline, the proposed attention approach can consistently achieve better performance in terms of speech quality (PESQ) and intelligibility (STOI). More promisingly, the attention-based approach has better generalization ability to unseen noise conditions.

A start-up scheme is presented for 10kV power electronics transformer composed of dual-active-bridge (DAB) and H-bridge. This paper firstly presents the 10kV power electronics transformer (PET) system control strategy as well as the controller structure including main controller and modular controller. Then, the start-up scheme is introduced both in systematic level and DAB converter level. This start-up scheme not only eliminates the soft-start equipment in traditional 10kV grid-connected converter but also maintain low inrush current for both DAB and H-bridge in the whole process. Finally, 10kV/1MW prototype is introduced and the experiment to verify the start-up scheme is presented.

This paper introduces a simple and reliable gate drive scheme which could effectively solve the negative voltage undershoot at the gate terminal of SiC device during the turnoff process of the Dual-Active Bridge (DAB) topology. Through inserting a diode and a magnetic bead into the classic gate drive circuit, the drive scheme could be formed. The operating principles of the proposed drive scheme could be divided into four stages and described in details, in consideration of the complementary switch in the phase-leg as well as the diode in the gate drive circuit. The effectiveness of the proposed gate drive circuit could be validated on a SiC-based double pulse tester under the working condition of DAB. The main contribution of this work is that the proposed scheme would largely improve the capability of the passive gate driver on the mitigation of the negative gate spike. Based on the simple structure and passive component's reliable performance, the proposed gate drive scheme could be directly adopted to DAB. The design guideline is also proposed, where the gate voltage spike could be controlled through the adjustment of magnetic bead's impedance.

Solid state transformer (SST) is able to interface PV panels and distributed grid, which could serve as a new structure for future PV plants. However, the wide operating voltage range of PV system reduces the efficiency of DAB based modular cascaded SST system. Thus, this paper analyzes the impact of DAB voltage deviation between input and output side under frequency component analysis. Besides adjusting primary and secondary side duty cycles, which is widely used in previous literature, this paper utilizes DAB output voltage as an extra control freedom in SST. Then, an adaptive voltage control (AVC) scheme is proposed, which changes DAB output voltage in accordance with input voltage. This paper presents the detailed implementation of AVC and thoroughly compares SST cell efficiency with previous modulation method. Finally, the proposed method is verified in 10kV/1MW SST system.

Thermal error is the most important source of error in the machine tool, and the thermal deformation caused by the temperature rise during the cutting process often leads to lower machining accuracy and poor quality. About 40%–70% machining error of machine tool is caused by thermal deformation. Considering that it is difficult to avoid the generation of heat during the operation or the cutting process of the machine tool, how to predict the temperature change of the machine tool and then design effective thermal management measures becomes an important task in the development of ultra-high precision machine tools. In this study, the thermal deformation of the machine tool structure due to the heat generated during operation is analyzed, and the embedded cooling channels is applied to exchange the heat generated during the operation to achieve the purpose of thermal error suppression. Then, the finite volume method is used to simulate the effect of cooling oil temperature on the thermal deformation and the thermal suppression is conducted on the feed system in combination of the cooler in order to improve the positioning accuracy of the machine tool. From the simulation results, it is observed that when the temperature of cooling oil is lower in the embedded cooling channels, more heat generated is taken away by the hydrostatic guideway and the average temperature of the hydrostatic guideway is reduced, thus decreasing the thermal deformation. Through the thermal suppression experiment of the embedded cooling channels, the positioning accuracy of the feed system can be improved by about 39.1% during the dynamic feeding process.

The modular multilevel converter (MMC) is a competitive candidate for VSC-HVDC systems. However, an account of various Sub-Modules bring serious difficulties for the appropriate operation principle design and performance analysis. This paper proposes the unified model of Half Bridge-MMC (HB-MMC) and MMCs with DC fault ride-through capability. Based on the operation essence of the cascaded rectifier and inverter, the capacitor voltage fluctuation with the influence of circulating currents and output currents is revealed. Furthermore, the optimal operation principles are specified for the capacitor voltage fluctuation reduction and the power loss reduction separately. Verification of the theory contribution proposed in this paper was implemented in a three-phase FB-MMC PSCAD/EMTDC simulation system with online loss simulation.

Low voltage ride through ability is necessary for HVDC links based on modular multilevel converter (MMC). However, there is not enough communication between inverter side MMC and rectifier MMC on the AC voltage fault information in some HVDC projects. This paper analyzes how the loss of communication could cause failure in low voltage ride through for VSC HVDC system. An active power control strategy based on burst mode operation is proposed. This control strategy is able to maintain the normal operation of MMC on both sides and eliminates the risk of over voltage both for MMC sub-modules and DC line. The mathematical calculations are presented and PSCAD/EMTDC simulation results verify the effectiveness of the proposed control strategy.

The three-phase solid-state transformers(SST) with DC bus have been widely studied in micro-grid systems. The double line frequency component of AC power on DC-link in SST is unavoidable because of the inherent single phase structure of the HV-side cascaded H-Bridge(CHB) converter. However, the AC power at double line frequency of three phases can be mutually eliminated by enabling DAB converters with the ability to process the power. This paper develops a closed multi-loops method of voltage control and fluctuation suppression for the three phase SST system with DC-bus in dual rotating reference frames. Theoretical analysis and modeling are provided. The performance of the proposed control strategy is verified through simulation results and the experiment is under way.

In general, the power systems with high penetration of renewable energies result in a system with a relatively low moment of inertia, and can be susceptible to significant frequency variations. However, the virtual inertia of wind power which imitates the kinetic inertia of synchronous generator can be used to improve the system dynamic behavior. This paper presents a novel hybrid adaptive fuzzy control scheme that utilizes both the rotating mass of the DFIG and a super-capacitor bank as the virtual inertia sources for DFIG-Based wind energy applications to provide short-term frequency regulation support. First, in order to compensate the intermittent nature of the wind, a super-capacitor is connected to the DC link of the back-to-back converters and an additional adaptive fuzzy controller is added to the supper-capacitor system controller to realize the short-term frequency support. In a ddition, to reduce the cost of the supper-capacitor, by utilizing the rotating mass of the DFIG, an adaptive fuzzy controller is also designed according to the deviation and derivative of the grid frequency for the active power regulation of the rotor-side converter to provide the optimal virtual inertia. Furthermore, a coordination optimizer is designed to obtain an optimal trade-off between the energy output from the DFIG rotating mass and the energy from the super-capacitor. Simulation and experimental results are presented to demonstrate the effectiveness of the proposed control strategy.

To reduce CPU, memory and bandwidth usage as cloud platform security consumption, and to prevent vulnerabilities that brought in by VM based antivirus software, we propose an agentless processes monitoring architecture for cloud platform. To demonstrate this architecture, we modified KVM kernel, programed OpenStack plugins, added security module on both management and compute node in a cloud platform. With this architecture, we are able to monitoring all processes running on every VMs agentless in one cloud platform, then locate and terminate suspicious processes. Analysis and test results showed that this architecture is able to reduce resource consumption for cloud platform security and increase its security level.

With the development of the renewable energy, grid-connected H-bridge inverter plays a more important role than ever. However, the dead-time and the zero crossing clamping (ZCC) effect caused by dead-time can cause THD problem. As a consequence, more and more researches have focused on this issue. Most of the researches use the harmonics to replace the effect of dead-time, which ignores ZCC effect. Some other researches use the average current to distinguish the ZCC effect, which is not accurate enough. This paper analyzes the whole changing process of inductor current ripples during power cycle and a complete discrete-time model of the H-bridge grid-connected inverter with dead-time is established. On the basis of the model, the analysis of dynamic behavior is carried out, the result of which shows that the length of dead-time can cause nonlinear phenomenon and reduces the stability region of the controller parameters. Besides, this model is also useful to the nonlinearity compensation of the system.

In the zinc oxide (ZnO) varistor characteristics testing, high tracking precision and stability are required both for the output voltage and load current, while the load is severely nonlinear. In this paper, a three-loop control strategy combining PI and Quasi-PR compensator is presented. Firstly, the inner-loop controls the inductor current using a PI controller to suppress the bias direct current (DC) flowing through the transformer. Secondly, the middle-loop controls the output voltage combining PI and Quasi-PR compensator to improve the tracking precision for sinusoidal waveform control. Finally, the outer-loop is a DC component of ZnO current feedback loop applying a PI compensator to achieve high control performance of the ZnO current. The proposed control strategy was verified in a 10kW testing power supply. Experimental results show that the tracking precision is nearly 1% and the THD is 0.8% under the proposed three-loop control strategy.

How to solve the recipe-based systematic problems of the control process and production management is one of the technical difficulties of the system shortest path planning (SSPP). The control recipe modeling method is proposed based on recipe engineering for the physical visualization experiment process of SSPP, and some control recipes are implementation in this paper. The result shows that the proposed modeling method can meet the requirement of the experiment process, and is applied successfully to the physical visualization experiment process.