Proceedings of 2021 International Top-Level Forum on Engineering Science and Technology Development Strategy

Inhaltsverzeichnis

1. Frontmatter

2. Transmission Line Fault Diagnosis Method Based on CNN Network and Multi-kernel Learning

   Zhang Heng, Rao Dan, Zhou Hua-liang, Li Lin-feng

   Abstract

   In order to adaptively fuse various electrical quantities generated during the operation of transmission lines, the complementary information of each electrical quantity is combined to improve the accuracy of fault diagnosis of transmission lines. This paper proposes a method of line fault diagnosis based on deep convolutional neural network and multi-kernel learning. First, use the convolutional neural network to automatically extract the features of various electrical quantities, and then map the extracted features to different feature spaces, and finally use the multi-kernel learning method to build a generalization model fusing multi-source feature data. The weight of different features we learned is used to clarify the different contributions of each characteristic data source to the fault diagnosis. The results showed that this method has higher classification accuracy for multi-source electrical quantities, and achieves an improvement in overall diagnostic performance.

3. A Negative Feedback Excitation Control Strategy for Suppressing Sub-synchronous Oscillation of Synchronous Generators

   Feng Ji, Lu Gao, Xiang Cui

   Abstract

   In this paper, a negative feedback excitation control strategy is proposed to suppress the sub-synchronous oscillation of synchronous generators. The development process of the vibration mode energy of the mechanical shafting, and the oscillating electromagnetic energy of the electrical circuit, are studied when the sub-synchronous oscillation occurred. This paper pointed out that most of the oscillation energy exists in the form of vibration mode energy of the mechanical shafting. Based on the root locus analysis, a negative feedback control strategy is proposed to suppress the vibration mode. The instantaneous electromagnetic torque could be controlled by the instantaneous field current, while the instantaneous field current could be controlled by the instantaneous excitation voltage. The electromagnetic transient simulation showed that the proposed strategy could effectively suppress the sub-synchronous oscillation, which caused by small disturbance or even large disturbance.

4. Distribution Network Differential Protection Method Based on Dynamic Time Warping Algorithm

   Wei Long, Shuyan Pan, Hai Wu

   Abstract

   As a new technology, 5G technology has lower cost than optical fiber and comparable transmission speed and carrying capacity, which can improve the performance of distribution network communication system. At the same time, differential protection based on two terminal data cooperation has a new life. However, 5G signal has delay jitter, and the transmission delay of channel may change, which increases the difficulty of synchronization and affects the reliability of differential protection. Therefore, this paper proposes an improved protection method based on dynamic time bending distance, introduces floating threshold as braking value, improves the reliability of protection, and has good adaptability to delay jitter and asynchronous error, which has certain engineering significance.

5. Power Supply System Modeling and Converter Control of Constant Voltage Subsea Observatory

   Lv Haoze, Xu Chu

   Abstract

   The cabled subsea observatory is the ‘pillars of a great power’ on marine resources development and marine defense construction. It is necessary to construct an accurate and reliable electromagnetic transient model of cabled subsea observatory power supply system. Based on the literature research of subsea observatory, the electromagnetic transient model of constant voltage subsea observatory power supply system is proposed, which with FBMMC as power feeding equipment and synchronous converter as junction box converter. Further, this paper proposes a simplified model for branch unit and an equivalent model for the optic-electric composite submarine cable. The constant voltage subsea observatory power supply system established in this paper is electromagnetic transient simulated by PSCAD/EMTDC, which can reflect the steady state, the transient state of line grounding fault, the transient state of voltage-stabilizing and current-limiting control. The simulation results and theoretical analysis verify the accuracy, effectiveness, and flexibility of the proposed electromagnetic transient model of constant voltage subsea observatory power supply system. This paper provides the necessary experimental basis for further research on the safety and stability analysis of subsea observatory power supply systems, protection schemes based on control and protection coordination.

6. A Leakage Oil Segmentation Network Based on Edge Information Fusion

   Huang Wenli, Weng Liangjie, Zhu Tao, Huang Gang, Mao Ji, Wang Taiping, Wang Chen

   Abstract

   The leakage of oil from the primary oil filling equipment in the substation will cause a great risk to the normal operation of the substation. Oil pollution is a typical non-rigid object. As the attachments change, the oil will have a variety of shapes. When the shadow is similar to the color of the oil and overlap, the edges of the oil become indistinguishable. In order to improve the detection rate of oil leakage targets and reduce the influence of shadow illumination on the detection model, this paper proposes an attention segmentation network based on edge fusion to solve the problem of oil leakage detection in substations. Firstly, in order to make full use of the spatial context information of oil pollution morphology, the self-attention mechanism is proposed to improve the detection rate of oil leakage. Secondly, in order to have better distinguishing performance between shadow and illumination, image edges are introduced into loss calculation. The edge information of the image can guide the network to learn the difference between shadow and oil. Finally, online hard example mining is adopted to reduce the false detection rate of oil pollution detection in the loss calculation. The experiment proves that the performance of the segmentation network designed in this paper is much better than that of the original segmentation network in detecting oil leakage, and it can be successfully implemented in the scene of substation operation.

7. Incipient Fault Identification Based Protection for a Medium Voltage DC Integration System

   Shi Zhiming, Jia Ke, Wu Wenqiang, Chen Miao, Chen Cong, Liu Bohan

   Abstract

   In a medium voltage DC integration system, the poor laying environment and mass cables collection lead to an increase in the line fault rate. Once a line-to-line short circuit occurs, the converters of the whole system will be quickly locked, and the system will be out of service for a long time, greatly reducing the system reliability. To address this issue, this paper takes a photovoltaic DC integration system as the research object, and proposes a protection that is based on the incipient fault identification. It can prevent the incipient fault from developing into a pole-to-pole short circuit. By analyzing the fault harmonic components generated by an incipient fault under the excitation of the inherent harmonics of the system, and then using the threshold method to identify the incipient fault, the problem of system outage caused by the short circuit fault can be solved. The simulation results in PSCAD/EMTDC show that the proposed method can accurately detect incipient faults, and is less affected by photovoltaic output changes and converter action. This provides a guarantee for the safe and reliable operation of the medium voltage DC integration system.

8. Design of Primary Frequency Control Online Test and AGC Control Performance Assessment System for Generator Unit

   Sun Baogong, Zhang Liudong, Fan Haifneg, Xu Rui, Pang Tao

   Abstract

   Unit’s ancillary services of frequency regulation should be evaluated rationally for just and fair in electric market. The problems in present assessment of primary frequency regulation (PFR) and automatic generation control (AGC) are analyzed firstly. And then the framework and design method of PFR online test and AGC control performance assessment system is proposed. According to requirement of data precision and timeliness, a PFR online test method based on remote terminal unit (RTU) and phasor measurement unit (PMU) data is proposed to promote accuracy of PFR performance evaluation. This method improves the calculation in time scale check of multiple data source and assessment index calculation. Meanwhile a control precision method is proposed basing on regulating rate distribution probability to establish linkage mechanism of control precision and regulating rate further to restrain and optimize unit’s control behavior. Validity of the proposed method is proved with an application case.

9. Parallel Decomposition and Coordination Algorithm for Security Constrained Optimal Power Flow

   Xia Xiaoqin, Xu Wei, Yuan Zhen, Xu Xiaotong

   Abstract

   In order to improve the online performance of power system control decision-making, this paper proposes a method to resolve the economic optimization problem considering security constraints into a base-case optimal power flow master problem with the objective of minimizing the generation cost and a series of security check sub problems relevant to various faults. A parallel decomposition and coordination algorithm based on adjustable space slices is proposed to avoid alternating iterations. First of all, the adjustable space is sliced to form multiple slicing schemes according to different proportions. Based on the parallel computing platform and the slicing schemes, the primal dual interior point method is used to solve the base-case optimal power flow master problem. Secondly, security check of faults is carried out by parallel computing for different optimized cases. Then, the solution with the minimum cost is chosen from the safe solutions. Finally, the effectiveness of the proposed method is verified by a case of a city power gird in Guizhou province.

10. Construction of SVSRB Based on DNN

    Hu Fan, Changhong Deng, Xuebin Wang, Haotian Zhang, Zhijun Long

    Abstract

    The increasing uncertainty of the power system makes the static voltage stability region (SVSR) an effective method to analyze the static voltage stability. The key is the construction of the static voltage stability region boundary (SVSRB). This paper proposes a SVSRB construction method based on Deep Neural Networks (DNN), which utilizes continuous power flow (CPF) as the basis to detect SVSRB boundary information as samples, and then puts the samples into DNN network to obtain SVSRB Solver. We takes IEEE9 and IEEE118 node systems as examples to verify the effectiveness of the DNN-based construction of SVSRB. The simulation results show that the method proposed in this paper can accurately fit the global boundary of SVSRB.

11. Research on the Behavior of Power Customer Responding to Electric Price Stimulation Based on Chameleon Mirror Effect

    Ji Bin, Chen Zhenghuan, Chang Li, Ding Qia, Cao Rongzhang

    Abstract

    With the steady advancement of power marketization on the distribution side, power customers (PC) participation in the power demand-side response has become an important part of the construction of the power market. Due to the randomness and profitability of PC, compared with the lack of effective control methods on the power generation side, it is difficult to show the effectiveness of flexible load resources on the demand side. Therefore, this paper proposes a collaborative modeling and analysis method based on the chameleon effect (CE) of customer’ power demand response. First, the social attribute factors of the CE are analyzed, and the chameleon mirror effect (CME) of PC is proposed and modeled. Then, CME and the evaluation benefit are modeled separately for the PC set of the distribution network system. Finally, quantitatively verify the model and evaluate the benefits of the system. The analysis of calculation examples shows that the establishment of the CME of PC in this paper conforms to the laws of the market. PC who follow CME can obtain more favorable power consumption benefits, which can stimulate PC to follow CME in reverse.

12. An Adaptive Static Voltage Stability Margin Estimation Approach Based on Incremental Learning-Enhanced XGBoost

    Han Gao, Guowei Cai, Deyou Yang, Lixin Wang, Bo Wang

    Abstract

    In order to adaptively estimate the static voltage stability margin (VSM) in the presence of topology changes in power systems, an incremental learning-enhanced XGBoost (ILeXGBoost) approach is proposed in this paper. This method exploits the gradient boost theory and a second-order Taylor expansion of the loss function to guarantee the accuracy of results; in addition, incremental learning technology is applied to incrementally adapt the model as the training data changes, continuously using previously learned knowledge and real-time updated phasor measurement unit (PMU) data to facilitate its learning procedure, when sudden changes of system topology and dynamic uncertainties occurred. The simulation results show that the scheme can effectively perceive the topology changes of systems to obtain estimation results adaptively, and the calculation speed is fast enough for online application.

13. Research on Optimizing Multi-energy Power Supply of Directed Energy Weapon Based on Automatic Demand Response Strategy

    Wang Liguo, Han Yakai, Qiao Jinxin

    Abstract

    This paper is focusing on the power demand of the directed energy weapon which has characteristics of high energy consumption, high efficiency and high power density. In order to safeguard the stability of DEW, push the limit of local power capacity, decrease the power consumption cost and increase the work efficiency, a multi-energy complementary power supply system consists of local power system, renewable energy and zinc-bromine flow battery energy storage system has been proposed. By combining Automatic Demand Response Strategy with modelling the inverter the multi-energy coordination planning strategy between the distribution module and energy storage module can be given based on predicting the power consumption of the DEW which is associated with the output voltage of the inverter. Experiments show that the lowest energy storage capacity and consumption cost can be determined by this work based on monitoring the electricity price. It can give a reference for establishing reliable multi-energy flow which may satisfy the power supply demand with characteristics of high performance price ratio of DEW.

14. A New Differential Protection Scheme of Distribution Network Line Based on Improved DTW Algorithm and 5G Communication

    Hongfei Pu, Tonghua Wu, Gang Yao, Feng Hong, Xiaojiang Zheng, Weiwei Wang, Xindong Li

    Abstract

    With the massive access of new energy, the differential protection will replace the traditional single-ended protection as the main protection of the distribution network line. Among the current differential protection channels, only 5G communication technology has both advantages of performance and price. However, using 5G communication as a communication channel needs to overcome the problems of delay jitter and data packet loss. This article first analyzes the principle of differential protection based on dynamic time warping (DTW) distance algorithm, and uses the nearest previous normal sampling point to replace the current abnormal one to deal with the problem of data packet loss, which is called alternative processing method. Secondly, the main starting criterion based on DTW distance of two-phase current and the auxiliary starting criterion based on DTW distance of the triple zero-sequence current are proposed. Thirdly, the calculation method of DTW algorithm under the short data window is proposed and the two operating criteria based on DTW algorithm are improved. And then, these 4 criteria are combined as a differential protection scheme for distribution network. Both theoretical calculation and simulation analysis show that the proposed scheme can operate well under different types of faults in different locations. With 5G communication delay being ignored, the scheme can operate in only 4–7 ms, and the data window of main operating criterion is only 3.75 ms.

15. Analysis of the Status and the Prospects of Pilot Protection in Distribution Network Based on 5G Communication

    Hongfei Pu, Tonghua Wu, Gang Yao, Feng Hong, Xiaojiang Zheng, Weiwei Wang, Xindong Li

    Abstract

    With the massive access of new energy resources, distribution network urgently needs to popularize the principle of pilot protection. As a suitable data transmission channel at present, 5G communication technology still has the problem of time delay and its jitter, which also directly affects the data synchronization method of pilot protection. On the basis of analyzing the composition of 5G time delay and reasons for its jitter, this paper first summarizes the current methods to alleviate this problem and proposes a new solution to deal with it, namely regional ad hoc network technology based on 5G CPEs. Secondly, after elaborating the advantages and disadvantages of data synchronization and self-synchronization technology based on 5G communication, in order to cooperate with regional ad hoc network technology based on 5G CPEs, the improved ping-pong time synchronization method is proposed. Thirdly, the principle of pilot protection based on waveform similarity is explained in detail, which has good performance of anti-synchronization error and anti-data packet loss. And then, the data transmission security and 5G data traffic problem are analyzed. Finally, the overall communication strategy of the regional distribution network based on 5G communication is proposed, and several suggestions for the development of the pilot protection of distribution network based on 5G are proposed.

16. Evaluation and Management of Operation State of Distribution Station Area Based on Incentive Demand Response

    Peng Jin, Hai Wu, Guo Hu, Qian Shen

    Abstract

    The popularization and promotion of smart home appliances and electric vehicles represents a challenge as well as an opportunity for distribution system operators. On the one hand, these new loads will significantly increase the operating burden of the distribution area; on the other hand, the controllability of these new devices also makes it possible to meet various technical and commercial purposes through flexible solutions such as incentive demand response. Based on the massive operation data collected from electricity meters, low-voltage fault indicators, leakage protection units and other devices in the distribution station area, a fuzzy comprehensive evaluation method is proposed to manage the operation status of low-voltage circuit in distribution station area. Then the evaluation results are applied to the incentive demand response, which realizes the comprehensive management of three-phase unbalance and under-voltage of low-voltage circuit in distribution station area through reducing load. The simulation results show that the proposed strategy can quantitatively evaluate the operation status of the distribution station area, and improve the load distribution and voltage level of distribution station area significantly through utilizing incentive demand response.

17. Hybrid HVDC Transmission Line Protection Method Based on Synchrosqueezing Wavelet Transform

    Xindong Li, Wei Dai, Tonghua Wu, Xianliang Teng, Kuncheng Zheng, Xiaofan Hou, Dan Wu

    Abstract

    The electromagnetic transient simulation model of hybrid HVDC transmission system is constructed based on the Baihetan—Southern Jiangsu Hybrid MTDC Transmission Project. The theoretical and simulation analysis targeting the transmission characteristics of boundary elements to line mode voltage fault components in Hybrid DC transmission system is conducted. Accordingly, a hybrid transmission line protection method based on fault voltage transient is proposed. The high and low-frequency energy ratio in line mode voltage fault components at both ends of the line is calculated to identify the internal and external faults of hybrid HVDC transmission line. Specifically, the extraction of transient component of line mode voltage and the characterization of its high-frequency energy are completed by synchrosqueezing wavelet transform. The calculation results show that, compared with the traditional continuous wavelet transform, the synchrosqueezing wavelet transform has better time–frequency concentration. According to the simulation results, such protection method can identify the internal and external faults of the line in a quickly and reliably manner, and can reliably act in case of high resistance grounding fault.

18. Island Division Method of Active Distribution Network Considering Operational Risk

    Zhang Jinfeng, Zhang Xueping, Zhang Tianzhong, Qian Chaojun, Fan Yuanzhu, Li Kunpeng

    Abstract

    The output of scenic resources in distribution network is uncertain, which brings great challenge to the island operation after fault recovery. In view of this, this paper proposes an active distribution network island division method that takes into account the operation risk, fully considers the islanding operation out-of-limit risk caused by the uncertainty of load and DG. Firstly, the risk index of island-bound fault in distribution network is established, and then the fault recovery model of distribution network is established with the objective of minimizing load loss and risk of island-bound fault, which is solved by improved bat algorithm, the optimal isolated island range is determined on the premise of economy and security. Finally, through the analysis of an example, it is verified that the proposed fault recovery method can effectively reduce the problem of island operation failure caused by wind, solar and load fluctuation, and ensure the safe and stable operation of the island in the whole operation period while reducing load shedding.