A DC grid primary protection algorithm based on current measurements
A DC grid primary protection algorithm based on current measurements
复制标题
DOI：
10.1109/epe.2015.7309268
复制
发表时间：
2015-09
期刊：
影响因子：
--
通讯作者：
S. Azad;W. Leterme;D. Van Hertem
中科院分区：
文献类型：
--
作者：
S. Azad;W. Leterme;D. Van Hertem
提出了一种基于数据的高压直流输电系统一次故障检测与识别算法。建议的FDI算法仅利用电流测量来识别发生直流故障的直流电网的保护区。与文献中先前研究的基于数据的方法相比，本文提出的算法使用从电流而不是电压测量获得的特征。基于电压测量的FDI算法还需要从电流传感器获得的测量来区分DC传输线路和母线端子上的故障。本文还提出了一个详细的光学电流传感器（OCT）的模型，并研究了所提出的FDI算法的精度和速度如何受到i）测量噪声和ii）采样频率的影响。该算法被应用到一个四端HVDC电网。提供给继电器的电流测量值是从精心建模的OCT获得的。研究结果表明，该算法准确地检测故障的直流输电线路和换流器母线端子，并确定保护区发生故障（母线端子或输电线路）。FDI算法的性能在各种测量噪声的情况下不会降低，即，光学、电子单元和量化。此外，采样频率的范围内，该算法可以使用而不损失精度的影响。
This paper proposes a data-based primary fault detection and identification (FDI) algorithm for HVDC grids. The proposed FDI algorithm utilizes only current measurements to identify the protection zone of the DC grid where the DC fault occurs. In contrast to the previously studied data-based methods in literature, the proposed algorithm of this paper uses features obtained from current rather than voltage measurements. The FDI algorithms which are based on voltage measurements also require the measurements obtained from current sensors to distinguish between faults on DC transmission lines and bus terminals. This paper also presents a detailed model of an optical current transducer (OCT) and investigates how the accuracy and speed of the proposed FDI algorithm is impacted by i) measurement noise and ii) sampling frequency. The proposed algorithm is applied to a four-terminal HVDC grid. The current measurements provided to the relays are obtained from elaborately modeled OCTs. Study results show that the proposed algorithm accurately detects faults at DC transmission lines and converter bus terminals and identifies the protection zone where the fault occurs (bus terminal or transmission line). The performance of the FDI algorithm does not degrade in case of various measurement noises, i.e., optical, electronics unit, and quantization. Furthermore, the range of sampling frequency for which the proposed algorithm can be used without losing accuracy is investigated.