This paper presents a robust scheme to detect and isolate faults quickly to prevent significant damage to the DC microgrid. The proposed technique uses the joint framework of Hilbert–Huang transform and empirical mode decomposition for feature extraction and bagging tree classifier to accurately and swiftly identify DC faults, which is challenging due to the limited time available to interrupt them. The intermittency pertaining to PV source and outage of distributed energy resources (DERs) may further complicate the protection task. In this regard, this paper proposes an intelligent scheme for fast fault detection and classification in DC microgrid. The joint framework of Hilbert transform and empirical mode decomposition has been used to calculate discriminatory attributes for characterizing the fault behavior in the signal. The ensemble strategy of efficient bagging tree classifier has been exploited after extensive testing and comparison with other modern approaches in this framework. Compared to other methods, the proposed scheme is more precise and faster which ascertain its efficacy in providing resilient protection to the DC microgrid with immunity to stochastic behavior pertaining to weather intermittency and DER outage. The performance of developed protection technique has also been validated on OPAL-RT digital simulator for authenticating its applicability in field applications.

本文提出了一种快速检测和隔离故障的稳健方案，以防止对直流微电网造成重大损坏。所提出的技术使用希尔伯特-黄变换和经验模式分解的联合框架进行特征提取和装袋树分类器来准确、快速地识别直流故障，由于中断故障的时间有限，这是一项具有挑战性的技术。光伏电源的间歇性和分布式能源（DER）的中断可能会使保护任务进一步复杂化。对此，本文提出了一种直流微电网故障快速检测与分类的智能方案。希尔伯特变换和经验模态分解的联合框架已用于计算用于表征信号中的故障行为的判别属性。经过广泛测试并与该框架中的其他现代方法进行比较后，已经开发出高效装袋树分类器的集成策略。与其他方法相比，所提出的方案更精确、更快速，确定了其为直流微电网提供弹性保护的功效，并且不受与天气间歇性和分布式能源断电相关的随机行为的影响。所开发的保护技术的性能也在 OPAL-RT 数字模拟器上进行了验证，以验证其在现场应用中的适用性。