随着以分布式电源（Distributed generation，DG）和微电网（Microgrid，MG）为主体的多元电力系统的发展，使得电网控制和运行模式的灵活性大大提高。微电网需要在并网和孤岛运行模式之间无缝切换和稳定运行，因此需要配备快速、准确的孤岛检测方法。相较于交流系统，直流系统中可用于检测的特征量较少，基于频率、相位等适用于交流系统的孤岛检测方法无法移植到直流系统中，常用的被动和主动式孤岛检测方法均存在一定问题，因此亟需在噪声和干扰环境中能准确分辨孤岛和非孤岛扰动情况的直流系统孤岛检测新方法。本论文对两类传统方法分别进行改进，所做工作如下：

（1）研究并网和孤岛状态下直流微电网系统的拓扑结构和控制策略，搭建含多分布式电源的直流微电网仿真模型，对微电网中各种可能发生的孤岛和非孤岛运行状态进行仿真分析，研究孤岛发生前后暂态过程的故障差异。

（2）针对传统方法存在的不足，提出一种两阶段主被动相结合的混合式孤岛检测方法。首先在第一阶段中，利用小波阈值去噪方法对公共点电压信号滤噪，然后对降噪后的信号使用互补集合经验模态分解和希尔伯特变换（Complementary ensemble empirical mode decomposition-Hilbert transform，CEEMD-HT）来量测公共并网点电压信号的瞬时频率变化趋势，当瞬时频率发生突变时完成对疑似孤岛发生时刻的标定；在第二阶段，通过对直流电流参考值进行扰动实现孤岛状态下直流侧输出电压的振荡，根据并网点电压在选定的频率下振荡进而检测孤岛；最后使用PSCAD/EMTDC平台进行了仿真验证，结果表明所提孤岛检测方法能准确快速的检测孤岛现象。

（3）针对容易获得模拟数据或已有运行数据基础的直流系统，本论文提出一种基于特征提取和神经网络的智能式孤岛检测方法。首先使用变分模态分解（Variational mode decomposition，VMD）和多尺度精细复合标准差模糊熵（Multi-scale refined composite Standard Deviation Fuzzy entropy，MRC-SDFE）对信号进行信号预处理和特征提取，然后通过深度信念网络(Deep belief network ,DBN)对提取到的特征向量进行训练学习和测试，在训练过程中引入海马优化算法（Sea-horses optimization algorithm，SHO）自适应地调整网络参数。通过PSCAD/EMTDC平台对直流微电网系统中可能出现的各种孤岛和非孤岛状态进行了模拟，验证了所提孤岛检测方法的准确性和优势。

（4）将所提出的方法与已有孤岛检测方法进行比较，主被动混合式孤岛检测方法无需一直向变流器中注入扰动量，可在减少对电网电能质量影响的基础上准确快速的检测孤岛；基于深度信念网络的智能式孤岛检测方法本质上属于被动式方法，不易受直流微电网中多种非孤岛扰动情况的影响，具有较高的检测准确性。两种方法均适用于运行状态多样的微电网. 本文的研究成果为直流系统孤岛检测提供了新的思路。

With the development of diversified power systems dominated by distributed generation (DG) and microgrid (MG), the flexibility of grid control and operation modes has been greatly improved. Microgrids require seamless switching and stable operation between grid connected and islanding operation modes, thus requiring fast and accurate islanding detection methods. Compared with AC system, there are fewer feature quantities available for detection in DC systems. Islanding detection methods based on frequency, phase and other factors suitable for AC systems cannot be transplanted to DC systems. Moreover, common passive and active islanding detection methods have certain problems. Compared with AC system, DC system has less feature quantity for detection. Islanding detection methods applicable to AC system based on frequency and phase cannot be transplanted to DC system. In addition, both passive and active islanding detection methods commonly used have some problems. Therefore, a new method for islanding detection of DC systems which can accurately identify between islanding and non-islanding disturbance in noise and interference environment is urgently needed. In this thesis, two kinds of traditional methods are improved respectively. The work of this thesis is as follows:

(1) The topology and control strategy of grid-connected and islanding DC microgrid systems are studied. A DC microgrid simulation model with multiple distributed power sources is built. Various possible islanding and non-islanding operating states in the microgrid are simulated and analyzed and the fault differences in the transient process before and after the occurrence of islanding are studied.

(2) Aiming at the shortcomings of traditional methods, a hybrid islanding detection method with two phases combined active and passive is proposed. Firstly, in the first stage, the voltage signal is filtered by wavelet threshold denoising method. The complementary ensemble empirical Mode decomposition-Hilbert transform (CEEMD-HT) method was used to measure the trend of instantaneous frequency of voltage signal at common coupling points and complete the calibration of suspected island occurrence time when instantaneous frequency changes. In the second stage, the output voltage of the DC side in the island state is oscillated by disturbing the reference value of the DC current and then the island is detected by oscillating at the selected frequency according to the grid connection voltage. Finally, PSCAD/EMTDC platform is used for simulation verification, and the results show that the proposed method can detect the island phenomenon accurately and quickly.

(3) For DC systems that are easy to obtain analog data or existing operational data, this thesis proposes an intelligent islanding detection method based on feature extraction and neural network. Firstly, the variational mode decomposition (VMD) and multi-scale refined composite Standard Deviation Fuzzy entropy (MRC-SDFE) are used for preprocessing and feature extraction of signals. Then, the deep belief network (DBN) was used to train and test the extracted feature vectors, and the sea-horses optimization algorithm (SHO) was introduced to adaptively adjust the network parameters during the training. Various islanding and non-islanding states that may appear in DC microgrid system are simulated by PSCAD/EMTDC platform, and the accuracy and advantages of the proposed islanding detection method are verified.

(4) Comparing the proposed method with the existing islanding detection methods, the hybrid active-passive islanding detection method does not need to inject disturbance into the converter all the time. It can detect islanding accurately and quickly on the basis of reducing the impact on power quality of the grid. The intelligent islanding detection method based on deep belief network is essentially a passive method, which is not affected by many non-islanding disturbance situations in DC microgrid and has high detection accuracy. Both methods are suitable for microgrids with various operating states. The research results of this thesis provide a new idea for islanding detection of DC system.

[1] 张明锐, 王俊凯, 王佳莹, 等. 微网混合式孤岛检测及运行模式切换研究[J]. 电力系统保护与控制, 2020, 48(02): 1-8.

[2] 陈晓龙, 杨志颖, 李永丽, 等. 基于MPPT梯形电压扰动的直流微电网新型孤岛检测方法[J]. 电网技术, 2020, 44(12): 4799-4807.

[3] 刘健, 魏昊焜, 张志华, 等. 未来配电网的主要形态—基于储能的低压直流微电网[J]. 电力系统保护与控制, 2018, 46(18): 11-16.

[4] 王鹤. 含多种分布式电源的微电网运行控制研究[D]. 北京: 华北电力大学, 2014.

[5] 王成山, 李鹏. 分布式发电、微网与智能配电网的发展与挑战[J]. 电力系统自动化, 2010, 34(02): 10-14+23.

[6] S. Ramachandradurai, N. Krishnan and N. Prabaharan. Unintentional passive islanding detection and prevention method with reduced non-detection zones[J]. IEEE Trans. on Power Delivery, 2022, 15(9): 3038-3047.

[7] H. Khosravi, H. Samet and M. Tajdinian. Robust islanding detection in microgrids employing rate of change of kinetic energy over reactive power[J]. Energies, 2022,15(9): 3038-3051.

[8] S. Nikolovski, H. R. Baghaee and D. Mlakic. Islanding detection of synchronous generator-based DGs using rate of change of reactive power[J]. IEEE Trans. on Smart Grid,2020,13(4):4344-4354.

[9] M. El Azzaoui. Islanding detection method with load power factor improvement and high frequency transient suppressing[J]. IEEE Trans. on Smart Grid, 2021, 12(5): 4176-4184.

[10] 王阳光, 张孝军, 徐宇新, 等.电流控制型与功率控制型分布式电源并存场景下过欠电压与过欠频率的检测盲区分析[J]. 电力自动化设备, 2021, 41(04): 109-115.

[11] 王渝红, 叶葳, 李天泽, 等. 一种基于d轴参考电流的受端孤岛检测法[J]. 南方电网技术, 2019, 13(12): 34-40.

[12] 陈巧地, 张兴, 李明, 等. 基于阻抗辨识的下垂控制并网逆变器孤岛检测方法[J]. 电力系统自动化, 2020, 44(07): 123-129.

[13] 熊金龙. 基于小波分析的负序电流扰动孤岛检测方法研究[D]. 武汉:湖北工业大学, 2015.

[14] Afshin TK, Navid G. A novel phase let-based approach for islanding detection in inverter-based distributed generation systems[J]. Electric Power Systems Research,2020,182(7), 106226.

[15] DO THANH HIEU（杜成孝）. 基于小波包变换的光伏并网逆变器孤岛检测方法研究[D]. 合肥: 合肥工业大学, 2016.

[16] 王金龙. 基于EMD和SVM的光伏孤岛识别研究[D]. 哈尔滨: 哈尔滨理工大学, 2016.

[17] 王士波. 基于希尔伯特—黄变换算法的微电网孤岛检测方法研究[D]. 哈尔滨: 哈尔滨理工大学, 2014.

[18] 马圆圆. 分布式光伏发电多逆变器并网的孤岛检测方法研究[D]. 北京: 华北电力大学(北京), 2020.

[19] 谢兴. 含多类微源且考虑负荷特性的微电网孤岛检测方法研究[D]. 长沙: 湖南大学, 2020.

[20] 熊金龙. 基于小波分析的负序电流扰动孤岛检测方法研究[D]. 武汉: 湖北工业大学, 2015.

[21] 邵美阳, 张子墨, , 等. 一种无功功率扰动的孤岛检测新方法[J]. 电网与清洁能源, 2023, 39(06): 98-107+114.

[22] 邓聪. 分布式发电系统的孤岛检测方法[D]. 长沙:湖南大学, 2020.

[23] G. S. Seo, K. C. Lee, and B. H. Cho. A new DC anti-islanding technique of electrolytic capacitor-less photovoltaic interface in DC distribution systems[J]. IEEE Trans. Power Electron., 2013, 28(4): 1632–1641.

[24] A. M. I. Mohamad and Y. A. R. I. Mohamed. Assessment and performance comparison of positive feedback islanding detection methods in DC distribution systems[J]. IEEE Trans. Power Electron., 2017, 32(8): 6577–6594.

[25] T. L. Shi, Hongyi Chen. Detecting speed improvement and system stability enhancement for DC microgrids islanding detection based on impedance characteristic analysis [J] IEEE Trans. Power Electron. 2023, 38(3): 3785-3802.

[26] Q. H. Huang, Hongyi Chen. Islanding detection with positive feedback of selected frequency for DC microgrid systems[J]. IEEE Trans. on Power Electron.2021, 36(10): 11800 -11817.

[27] H. T. Do, X. Zhang, et al. Chu. Passive islanding detection method using wavelet packet transform in grid connected photovoltaic systems[J]. IEEE Trans. on power systems. 31(10): 6955-6967.

[28] 孙金. 基于智能算法的交直流配网的孤岛识别方法[D]. 北京: 华北电力大学(北京), 2022.

[29] M. Shaik, A. G. Shaik and S. Yadav. Hilbert-Huang transform and decision tree based islanding and fault recognition in renewable energy penetrated distribution system[J]. Sustainable Energy Grid & Networks, 2022,30(4).

[30] X. R. Kong, X. Y. Xu, et al. Deep learning hybrid method for islanding detection in distributed generation[J]. Applied Energy, 2018, 210: 776-785.

[31] A. K. Özcanlı and M. Baysal. Islanding detection in microgrid using deep learning based on 1D CNN and CNN-LSTM networks[J]. Sustainable energy grid & networks, 2022,32.

[32] 万庆祝, 吴开聪. 基于随机森林分类的直流微电网孤岛检测[J]. 太阳能学报, 2022, 43(01): 269-276.

[33] 张军, 张新慧, 彭克, 等. 基于零序电压正反馈控制的多分布式电源孤岛检测方法[J]. 高电压技术, 2023, 49(07): 3040-3050.

[34] 李卓雅, 曾成碧, 苗虹, 等. 基于双谐波注入的电网阻抗检测方法研究[J]. 中国设备工程, 2022, (15): 192-194.

[35] 曹文远. 交流微电网模式切换与孤岛运行控制策略研究[D]. 北京: 华北电力大学(北京), 2022.

[36] 宋伟伟. 柔性直流输/配电系统的孤岛检测及其切换控制研究[D]. 广州: 华南理工大学, 2019.

[37] IEEE Std 1547-2018. IEEE standard for interconnecting distributed resources with electric power systems[S].

[38] 分布式电源并网技术要求: GB/T 33593-2017[S]. 北京: 中国标准出版社, 2008.

[39] GB/T 35727-2017, 中低压直流配电电压导则[S]. 北京: 中国标准出版社, 2017.

[40] 黄庆辉. 基于特定频率电压正反馈的直流孤岛检测方法研究[D]. 杭州: 浙江大学, 2021.

[41] 贾科, 朱正轩, 方煜, 等. 基于谐波阻抗量测得多端口光伏直流配电网孤岛检测[J]. 电网技术, 2019, 44(4): 1229-1238．

[42] 邓聪, 江亚群, 黄纯. 基于不同次数谐波下品质因数一致性的孤岛检测方法[J]. 电力自动化设备, 2020, 40(12): 204-210+224.

[43] 邬蒙蒙, 周怀来, 林萍, 等. 改进的完备经验模态分解和广义S变换相结合的地震信号衰减分析[J]. 地球物理学进展, 2020, 35(5): 2001-2008.

[44] 汪一飞. 基于全矢CEEMD的滚动轴承故障诊断研究[D]. 郑州:郑州大学, 2019.

[45] 范新桥, 朱永利, 卢伟甫. 基于EMD-TEO的输电线路行波故障定位[J]. 电力系统保护与控制, 2012, 40(09): 8-12+17.

[46] 徐振曦. 混合三端直流输电线路故障测距方法研究[D]. 西安:西安科技大学, 2022.

[47] 李文娟. 基于经验模态分解与小波阈值降噪结合的柴油机故障诊断[D]. 太原: 中北大学, 2014.

[48] 张琳梅. 基于小波阈值处理的MEMS矢量水听器信号去噪方法[D]. 太原: 中北大学, 2020.

[49] J. Hu, K. Xu, et al. Analysis and enhanced control of hybrid-MMC-based HVDC systems during asymmetrical DC voltage faults[J]. IEEE Transactions on Power Delivery, 2017, 32(3): 1394-1403．

[50] 郑俊超, 文明浩, 秦瑜, 等. 具备故障选极能力的高压直流输电线路差动保护新原理[J]. 中国电机工程学报, 2018, 38(15): 4350-4358+4635．

[51] 宋国兵, 王婷, 张晨浩, 等. 利用健全极MMC注入特征信号的直流线路故障性质判别方法[J]. 电工技术学报, 2019, 34(5): 994-1003.

[52] 宋国兵, 王婷, 张晨浩, 等. 利用全桥MMC注入特征信号的直流自适应重合闸方法[J].电网技术, 2019, 43(1): 149-156.

[53] H. R. Baghaee, D. Mlakic, S. Nikolovski and T. Dragicevic．Anti-islanding protection of PV-based microgrids consisting of PHEVs using SVMs[J]． IEEE Trans. on Smart Grid, 2020, 11(1): 4830-500.

[54] D. Mlakic, H. R. Baghaee and S. Nikolovski. A novel ANFIS-based islanding detection for inverter–interfaced microgrids[J]． IEEE Trans. on Smart Grid, 2019, 10(4): 4411-4424.

[55] H. R. Baghaee, D. Mlakic, et al. Support vector machine-based islanding and grid fault detection in active distribution networks[J]． IEEE Journal of Emerging and Selected Topics in Power Electronics, 2020, 8(3): 2385-2403.

[56] X. W. Wang, J. Gao et al. High impedance fault detection method based on variational mode decomposition and teager-kaiser energy operators for distribution network[J]． IEEE transactions on smart grid, 2019, 10(6): 6041-6054.

[57] P. D. Achlerkar, S. Samantaray and M S. Manikandan．Variational mode decomposition and decision tree based detection and classification of power quality disturbances in grid-connected distributed generation system[J]. IEEE transactions on smart grid,2018,9(4): 3122-3132.

[58] 张金良, 刘子毅. 基于混合模型的超短期风速区间预测[J]. 电力系统保护与控制, 2022, 50(22): 49-58.

[59] F. Xu and P. W. Tse. A method combining refined composite multiscale fuzzy entropy with PSO-SVM for roller bearing fault diagnosis[J]. Journal of Central South University, 2019,26(9): 2404-2417.

[60] A. S. Minhas, G. Singh, et al. A novel method to classify bearing faults by integrating standard deviation to refined composite multi-scale fuzzy entropy[J]. Measurement, 2020,154: 107441.

[61] Y. B. Li, Y. Wei, et al. Fault diagnosis of rolling bearing under speed fluctuation condition based on Vold-Kalman filter and RCMFE[J]. IEEE Access, 2018, 6, 37349-37360.

[62] N. Balakrishnan, A. Rajendran, et al. Deep belief network enhanced intrusion detection system to prevent security breach in the internet of things[J]. Internet of things, 2021, 14(4), 100112.

[63] N. Liu, B. Fan, et al. Cable incipient fault identification with a sparse autoencoder and a deep belief network[J]. Energies,2019,12(18):3424.

[64] R. N. Mantegna. Fast, accurate algorithm for numerical simulation of levy stable stochastic processes[J]. Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics, 1994,49(5): 4677- 4683.

[65] S. J. Zhao, T. R. Zhang, et al. Sea-horse optimizer: a novel nature-inspired meta-heuristic for global optimization problems[J]. Applied intelligence, 2022， 243： 108457-.

[66] G. Liu and J. Guo. Bidirectional LSTM with attention mechanism and convolutional layer for text classification[J]. Neurocomputing, 2019, 337: 325-338.

[67] E. Shahryari, M. Nooshyar and B. Sobhani. Combination of neural network and wavelet transform for islanding detection of distributed generation in a small-scale network distributed generation in a small-scale network[J]. Int. J. Ambient Energy, 2019, 40.

[68] D. Kumar and P.S. Bhowmik. Artificial neural network and phasor databased islanding detection in smart grid[J]. IET Gener. Transm. Distribution. 2018, 12: 5843-5850.