Search

Code review as a typical type of user feedback has recently drawn increasing attentions for improving code quality. To carry out research on code review, sufficient review data is normally required. As a result, recent efforts commonly focus on analysis for projects with sufficient reviews (called “s-projects”), rather than projects with extremely few ones (called “f-projects”). Actually, through statistics on public platforms, the latter ones dominate open source software, in which novel approaches should be explored to improve their review-based code improvement. In this paper, we try to address the problem via building a review sharing channel where the informative review can be reasonably delivered from s-projects to the f-projects. To ensure the accuracy of shared reviews, we introduce a novel code clone detection model based on Convolutional Neural Network (CNN), and build suitable “s-projects, f-projects” pairs through the clone detection. Especially, to alleviate the dataset heterogeneity between the training and testing, an autoencoder-based semi-supervised learning strategy is employed. Furthermore, to improve the sharing experience, heuristic filtering tactics are applied to reduce the time cost. Meanwhile, the LDA (Latent Dirichlet Allocation)-based ranking algorithm is used for presenting diverse review themes. We have implemented the sharing channel as a prototype system RSharer+, which contains three representative modules: data preprocessing, code clone detection and review presentation. The collected datasets are first transformed into context-sensitive numerical vectors in the data proprecessing. Then in the clone detection, data vectors are trained and tested on the BigCloneBench and real code-review pairs. At last, the presentation module provides review classification and theme extraction for better sharing experience. Extensive comparative experiments on hundreds of real labelled code fragments demonstrate the precision of clone detection and the effectiveness of review sharing.

Due to the large number of work orders, strong randomness and mutual correlation between work orders, the operation and maintenance of distribution network was difficult. To solve these problems, a method based on association analysis and machine learning was proposed to analyze the work order law of distribution network and predict the problem area. Firstly, the work orders that affect the operation status of the distribution network are found out, and then the association rules between the initial work orders were obtained by the proposed association analysis method, and then the most likely problem area is found by the confidence degree of the index. Finally, the effectiveness of the proposed method was verified by the actual data of the business system. The example results showed that the new method can describe the association rules between indicators more systematically and comprehensively, and the proposed method performs well in hit rate and accuracy.

Modular multilevel converter-based solid-state transformers (MMC-SSTs) have flexible bidirectional power management capability and are proposed to integrate different types of distribution networks. Due to the large number of insulated gate bipolar transistors (IGBTs) adopted, reliability is one of the major challenges of these grid devices. In this paper, the submodule (SM) IGBT open-circuit fault characteristics of an MMC-SST are analyzed in detail. Unlike conventional MMC applications, the MMC-SST exhibits different fault behaviours due to its unique topology and complicated operation modes. Under certain operation modes, the capacitor voltage of the faulty SM may not be affected by the faults and the faulty unit is able to maintain its power transfer capability. Therefore, the existing fault diagnosis schemes in the literature cannot be directly applied to the MMC-SST to detect and locate all fault conditions. Simulated study in MATLAB/Simulink shows good agreement with the presented analysis. The analysis provides the ground for developing advanced fault diagnosis and fault-tolerant methods for the MMC-SST under SM IGBT open-circuit faults.

Distribution network work order is one of the important indicators to reflect the operation and management level of distribution network, which has important information value for providing auxiliary production decision-making of distribution network. In the face of massive distribution network work order data, need to solve the problem of how to transform the data into auxiliary decision-making information. In this paper, through machine learning, big data and other artificial intelligence technology and data mining technology, the distribution network operation index correlation analysis and intelligent prediction, found the inherent law between the operation index, realized accurate operation and maintenance and scientific decision-making. Firstly, the distribution network work order data was preprocessed to clean the error and abnormal data, and the fuzzy algorithm was used to match the corresponding station area according to the lack of station area. Secondly, the type and characteristics of each type of work order were analyzed, and the distribution network work order index mining was carried out through PrefixSpan algorithm. Finally, the effectiveness of the proposed algorithm was verified through the actual data, and the operation and maintenance of the distribution network were analyzed The paper put forward the preventive measures for the weak links in the service.

DC Solid State Transformer (DCSST) plays a significant role in medium/low voltage DC distribution grid, which functions as changing voltage level and managing power flow between the MVDC distribution grid and the LVDC micro- grid. The conventional DC transformers can only provide single-pole or pseudo-bipolar LVDC ports. Therefore, this paper proposes a bipolar-type DC transformer topology based on NPC (Neutral Point Clamped)-DAB (Dual Active Bridge) and its asymmetric operation control scheme. This topology has bipolar LVDC ports which can meet the requirements of low- voltage users for multiple voltage levels, power safety and reliable supply. The main work of this paper includes: a DC transformer topology with bipolar LVDC ports based on NPC - DAB are proposed; the steady model based on DC current injection are established; the mechanism of bipolar asymmetric operation principle and the method of eliminating DC bias flux in the transformer are analyzed based on the steady model; the dynamic model based on DC injection are established; the asymmetric operation control strategy and the control parameter design method are proposed based on the dynamic model. Finally, the feasibility and effectiveness of the proposed control strategy are verified by MATLAB/Simulink experiment.

The widespread utilization of distributed energy resources including distributed generations, energy storage and controllable loads has posed threats to distribution network (DN) in terms of voltage violation, feeder congestion and bidirectional power flow issues. To mitigate the aforementioned problems, flexible AC interconnector (FACI) can be exploited to interconnect feeders in the place of normally open points. FACI can provide functions of continuous active power control and reactive power compensation to enhance the control capability of DN. This paper proposes a novel embedded multi-port FACI (eM-FACI) based on static synchronous compensator, which has advantages of modularity, low cost, high power density, flexible scalability and is capable of decoupled regulation of power flow in multiple feeders. First, topological configuration and operating principles are analyzed. Then inner energy balance mechanism is established and control strategy is proposed. Finally, theoretical analysis and feasibility of eMFACI are validated by a 3.3 kVA protype.

The dual closed-loop PI control is used for three-phase PWM rectifier, which arises large overshoot voltage and overshoot current in the moment when the PWM rectifier starts running and load changes suddenly. The dual closed-loop PI control is presented based on the voltage-current model for the three-phase PWM rectifier, which illustrated that the DC voltage to the current is in linear relation. The reason why the large overshoot current and voltage occurs in traditional control method is analyzed in theory. On the basis of that the new power-current model is proposed which shows that the square of DC voltage to the current is in linear relation. An improved control strategy is proposed based on the new power-current model. The proposed control method adopted the DC voltage square as the feedback in voltage loop and used current predictive control in current loop. Also, the load current and the DC voltage current feed forward are also employed to improve the performance. Theory analysis show that the proposed control strategy can greatly improve the system dynamic performance and the overshoot voltage and current in starting time and load dynamic changing moment can be greatly decreased. The validity of the proposed methods is proved by the simulation results.

Ultra-wideband (UWB) radar imaging has been applied in various applications such as synthetic aperture radar (SAR) imaging, through-wall-radar imaging and terahertz imaging. High level sidelobes raised by imaging approaches have serious impact on the image quality. In this paper a sidelobe reduction method for UWB radar imaging is proposed. The spatial spectrum coherence factor (CF) based sidelobe reduction method reduces the sidelobes and keeps the mainlobes simultaneously. The simulated and real data results have validated the effective, stable and convenient method.

We consider efficient communications over the multiple-input multiple-output (MIMO) multiway distributed relay channel (MDRC) with full data exchange, where each user, equipped with multiple antennas and broadcasts messages to all the other users via the help of a number of distributive relays. We propose a physical-layer network coding (PNC)-based scheme involving linear precoding for channel alignment, nested lattice coding for PNC, and lattice-based precoding for interference mitigation. Unlike most existing approaches characterizing the degrees of freedom of MIMO multiway relay channels, we focus on analyzing the rate gap of the proposed scheme towards the fundamental channel capacity. In this regard, we establish a sufficient condition on the system configuration for that distributive relaying achieves the same sum rate as cooperative relaying does in the high SNR regime. We further show that the proposed scheme achieves the asymptotic sum capacity of the MIMO MDRC within a constant gap in the high SNR regime. Numerical results demonstrate that the proposed scheme considerably outperforms the existing schemes including decode-and-forward and amplify-and-forward.

Reliability and fault-tolerant operation are among the major concerns of the modular multilevel converter (MMC) based applications due to their use of a large number of power electronic devices. This paper presents a novel global redundancy scheme for medium-voltage (MV) MMC based solid-state transformers (SSTs) to deal with the submodule (SM) IGBT open-circuit failure. By sharing the same redundant SM in one phase via bidirectional switches, the proposed fault-tolerant scheme can significantly maintain the SST operation with minimum degradation and save the initial cost and volume of the system. The characteristics of the IGBT open-circuit fault with two different SM topologies are summarized in detail. The proposed global redundancy scheme can efficiently restore the system from SM failure without severe malfunction and damages. Its effectiveness is verified by simulations carried out in the MATLAB/Simulink software.

With the increasing penetration of renewable energy generators in power grid, many inertia emulation control strategies have been raised to solve the problem of reduced power system inertia. A co-simulation research platform is built based on RTDS and GH Bladed, detailed comparisons are made in terms of inertia response performance, weak grid operation stability, secondary frequency drop and wind tower load analysis. Research results point out that different control schemes are similar in inertia response while virtual synchronous generator control (VSG) and inertial synchronization control (ISynC) can improve weak grid operation stability, VSG strategy is also effective in reducing frequency secondary drop and wind tower fatigue load, the combination of VSG and ISynC can get the best control results.

This paper presents a comparative analysis of impedance characteristics of modular multilevel converter (MMC) with direct modulation and compensated modulation. The analytical derivations of the ac-side impedances of the MMC with these two modulation modes are provided, respectively, which are then validated by the measured impedances in the time-domain simulation model. Subsequently, a case study is carried out to show the impact of the two modulation modes on the stability of the MMC-based interconnected system using the impedance-based analysis method. The findings could provide a reference for controller design of MMC-based power electronic systems from a stability perspective.

In the future AC-DC hybrid distribution system, solid state transformer (SST) will take place of conventional transformer, as it could reduce the weight and volume of transformer, and achieve voltage and current regulation, power flow control, power quality enhancement. MMC-based SST is a novel SST topology, which could provide the MVDC port. However, the excessive bulky submodule capacitors limits this topology into practical application. This paper proposes a fluctuation power control strategy for MMC-based SST to reduce the submodule capacitor voltage oscillation, which could optimize the capacitance and reduce the volume and cost of capacitor. The feasibility of this proposed method is verified by MATLAB simulation.

This paper presents the grounding design and fault analysis of a modular multilevel converter (MMC) based multi-port flexible interconnection device (FID) for future distribution networks. Based on the requirements of the FID, zig-zag transformer, large reactance grounding, and mid-point grounding schemes are investigated. Characteristics of these grounding schemes are analyzed and compared under both single line-to-ground (SLG) fault on the AC side and single pole-to-ground (SPG) fault on the DC side. Based on the performances of the grounding schemes, zig-zag transformer grounding is proved to be more attractive on both medium and low voltage sides of the proposed FID system.

Evolving from the popular modular multilevel ac-dc converter, modular multilevel dc-dc converter (MMDC) is attractive for medium-and high-voltage applications. However, exploiting ac circulating power to balance the submodule energy, the existing MMDC topologies inevitably suffer from high circulating current through the arms, large filter inductor at the low voltage side and complex submodule voltage balancing algorithms. For overcoming these issues, this paper presents a new power transfer mechanism to balance the submodule energy automatically by reconstructing the half-bridge submodule into a circuit quasi-resonant. Based on this submodule structure, a novel MMDC topology for step-up applications is proposed. Compared with the prior-art MMDC, this topology offers the following merits: 1) common-mode circulating current through the lower and upighefficiencyper arms is avoided. 2) self-balancing of the capacitor voltages is ensured by the proposed modulation method to insert and bypass the adjacent submodules in a complementary manner. 3) soft-switching operation is achieved for the majority of switches to alleviate the switching losses and electromagnetic interference. 4) voltage stress across the low-voltage side inductor is limited to the submodule voltage, thus reducing the size of the inductor significantly. Experimental results verify the performance of proposed MMDC.

In ultra-wideband (UWB) radar imaging applications, high level sidelobes have serious impact on the image quality. In this paper, we analyzed the relationship between point spread function (PSF) and spatial spectrum supporting region, and proposed a method based on spatial spectrum segmentation to improve the image quality. The proposed method reduces the sidelobes and remains the main lobe simultaneously. The simulation and real data results validate the effective, robust and uncomplicated method.

In order to reduce the influence of complex measurement noise, an improved track association fusion method was proposed. The method took into account the nonzero system error of the position parameters, relaxed the threshold of the position component association test, increased the association test of the first difference of the position parameter, and formed a position-velocity association test. At the same time, considering the influence of the gross error caused by the complex electromagnetic environment, the method applied the robust statistics to the related processes, such as distributed filtering, track association and weighted fusion, and obtained the final track association fusion method. The test of typical measured data showed that the proposed method had strong practicability.

In this paper, we investigate the impact of channel sparsity on the fundamental limits of massive multiple-input multiple-output (MIMO) systems with K single-antenna transmit terminals and an N-antenna receiver, where ``massive'' means N >> K >> 1, and ``sparsity'' means that a large portion of the channel coefficients are zeros. We propose a novel blind detection scheme that simultaneously estimates the channel and data through factorizing the received signal matrix. We show that the proposed scheme can achieve a degrees of freedom (DoF) arbitrarily close to K(1-1/T) with T being the channel coherence time, provided that N is sufficiently large and the channel is sufficiently sparse. This achievable DoF has a fractional gap of only 1/T from the ideal DoF of K, which is a remarkable advance for understanding the performance limit of the massive MIMO system. Furthermore, we consider the algorithm design for the proposed blind detection scheme. Numerical results demonstrate that the proposed blind detection scheme significantly outperforms its counterpart schemes in the practical SNR regime.

This paper proposes a multi-bus flexible interconnection scheme to balance the power of multiple transformers in low-voltage distribution systems. Unlike feeder interconnection schemes that may lead to extra infrastructure cost and potential stability problems, the proposed scheme adopts transformer bus interconnection instead by using a multi-terminal voltage-source converter. With the integration of energy storage device, improved performance is achieved when the interconnection scheme is in effect. The basic configuration, operation principle and control strategy of the proposed scheme are presented. Finally, economic evaluation conducted on a practical case reveals the significant benefits obtainable from application of the proposed scheme.

With the development of offshore wind power generation, large-capacity medium-voltage wind turbine becomes the development trend. In this paper, the cascaded multi-cell topology with four-quadrant operation capacity is applied to medium-voltage wind power converters, which consist of a large number of switching devices, so it is essential to improve the converter's fault-tolerant ability. Once the same-level fault cell bypass method is adopted, the generator has a large reduction in the voltage magnitude, sacrificing part of the power generation capacity. In order to keep the output voltage balance, as well as achieve the maximum output power of the generator under fault condition, this paper proposed a control method combining the neutral point shifting and the optimized pitch control for the cascaded multi-cell medium-voltage wind power converter, a model has been established on Matlab/Simulink simulation platform to verify the validity and effectiveness of the proposed fault-tolerant control.

This paper proposes the dc bus current optimization control strategy in DFIG wind power systems with current source converter (CSC). In CSC-based systems, the converter losses are dependent on the dc bus current so that it is important to reduce it to minimum. The dc current is chosen as the larger value of rotor side converter (RSC) current and grid side converter (GSC) current. The RSC current is usually larger than GSC current because RSC will provide the generator's exciting current additionally. If the generator's stator can provide some reactive power to provide part of the exciting current, and the GSC compensate this part of reactive power to keep the grid working in unit power factor condition, the dc bus can be reduced further. The detailed control strategies for optimum dc bus converter are proposed. The simulation results has verified the validation of dc bus current optimization control strategy.

We consider efficient communications over the multiple-input multiple-output (MIMO) multiway distributed relay channel (MDRC) with full data exchange, where each user, equipped with multiple antennas, broadcasts its message to all the other users via the help of a number of distributive relays. We propose a physical-layer network coding (PNC) based scheme involving linear precoding for channel alignment, nested lattice coding for PNC, and lattice-based precoding for interference mitigation. We show that distributed relaying achieves the same sum-rate as cooperative relaying in the high SNR regime in most scenarios, which implies that the proposed scheme with distributed relays is more suitable for practical systems than the schemes with cooperative relays.

We consider training-based channel estimation for cloud radio access networks (CRANs), in which a large amount of remote radio heads (RRHs) and users are randomly scattered over a certain service area. In this model, assigning orthogonal training sequences to all users, if possible, will cause a substantial overhead to the overall network. Instead, we introduce the notion of local orthogonality, in which the training sequence of a user is required to be orthogonal to the training sequences of other users in its neighborhood. We model the design of locally orthogonal training sequences as a graph coloring problem. Then, based on the theory of random geometric graph, we show that the minimum training length scales in the order of ln K, where K is the number of users covered by the CRAN. Therefore, the proposed training design yields a scalable solution to sustain the need of large-scale cooperation in CRANs.

This paper presents a systematic step-by-step design procedure for the dc inductance of a current source converter (CSC). The design is based on the analysis of acceptable ripple level in the dc current of the CSC. A model of current source inverter (CSI) under the operation of space vector modulation is established to estimate the dc current ripples in the system. The influence of different switching sequences is analyzed in detail to provide more precise design results. The analysis and design of the dc inductance have been verified by simulations in MATLAB/Simulink. Compared to conventional methods, the proposed method can reduce the dc inductance remarkably in the same ripple conditions.