IEEE Electrification - December 2020 - 105

be predicted. Therefore, a system with high reliability may
not be remarkably resilient.
As a potential cyberattack might lead to a power blackout and cause severe social impacts, the MG must have a
self-healing ability to continue operations in the presence
of attacks. However, different from existing methods, the
restoration procedure should be completely automated
and independent of the demand from network operators.
Therefore, the MG system needs to present some autonomous characteristics to achieve system restoration. This
asks the MG to have a sense of situational awareness that
includes preparedness, perception, and responsiveness,
as shown in Figure 5.

Preparedness
The MG should also be well prepared to face security
threats, continue working, and prevent the system from
destruction after an attack. First, the MG can employ a reinforced power infrastructure, such as fortified points of common coupling, cables, or power devices, to withstand severe
events. Second, the system could have a certain resilience
with a redundant topology design such that it can guarantee the fault ride-through operation. For example, a loopbased topology can enhance MG resilience, as it provides
multiple paths for delivering power. Third, the converter
controller can be developed with multiple protection
schemes to allow the unfaulty switches to continue running and bypass the faulty switches by adjusting the modulation method. Fourth, the implementation of an adaptive
topology or extra hardware is also helpful in improving MG
resilience. The use of parallel redundant converters is a typical example of an industrial application. It is worth pointing out that such a method requires high precision to
determine the time for switching the redundant hardware.
Fifth, the implementation of MG clusters can increase the
stability of the entire system. Compared with only one MG,
the MG cluster has more capability in the prevention of
power blackouts. Finally, the MG in some communities can
be equipped with backup power sources or hybrid energy
sources in case of device or system failures.

Perception
The MG system needs to have the ability to continuously
monitor the system states in real time, such that the system can prepare operational measures to limit the impact
of potential damage in advance. The monitoring scheme
should include distributed attack detection and islanding
detection. Attack detection is the first step toward securing the system. The converter must be equipped with a
robust attack detection scheme to deal with various types
of attacks. Furthermore, unintended islanding may be
triggered by an attack, which must be detected as fast as
possible. Because the islanding detection methods usually rely on the analysis of voltage and frequency deviation,
voltage unbalances, and harmonic distortion, a trustworthy phase-locked loop technique is of crucial importance.

Responsiveness
The MG is expected to have responsiveness, meaning it
can respond to a severe situation in a reliable manner and
provide some restoration capabilities to the entire system
after the occurrence of a potential attack. At the converter
level, the converter should possess a self-awareness that
can react to the system with local information and
resources. However, at the MG level, the converter should
keep cooperation awareness so that it can collaborate
with other converters in the system.

Converter Level
Typically, the MG operates as a grid feeding in a grid-connected mode. To maintain the functionality of the power
system, the converter can shift automatically to a gridforming mode, which helps to restore the power system
without enduring significant transients. However, this
requires a reliable plug and play control design and stability analysis for the converters.

MG Level
Responsiveness in the MG level consists of reconfiguration
and cooperation awareness. The reconfiguration scheme
can achieve a self-healing awareness to reconfigure the
network and utilize local energy sources. If the main grid
or neighboring MG is corrupted by the attacks, the MGs
can perform an islanding option, operate autonomously,
and schedule optimally, thus mitigating the impact of an
attack within the MG. In this case, the local energy storage
system and backup generation sources should sustain the
power supply correctively. A possible load curtailment
may be a necessity under severe conditions. Additionally,
the islanded MGs should be reconnected to the main grid
seamlessly when the main grid has ridden through the
extreme conditions. This method requires powerful
islanding and synchronization technologies.
Cooperation awareness allows the various energy
sources to work in a coordinated way to sustain the MG. A
typical problem for this perspective is power sharing,
which can be performed by a standardized droop-based
hierarchical control approach. However, it may fail when
dealing with the reactive power and harmonic current
sharing problem, especially in severe conditions. How to
share and compensate the reactive power and harmonic
current between each distributed generation unit is
becoming a new challenge in MG design. Modified droop
and virtual impedance methods are two alternative solutions to solve this problem.

Discussion
The resilience of the MG can be improved by introducing
situational awareness to the system. With sufficient preparedness, the MGs can withstand severe disruptions and
maintain functionality during attacks. The perception of
the MG monitors the system performance and trigger an
alarm in the presence of attacks. The responsiveness of

IEEE Elec trific ation Magazine / D EC EM BE R 2 0 2 0

105

IEEE Electrification - December 2020

Table of Contents for the Digital Edition of IEEE Electrification - December 2020