> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

Our main contribution to this manuscript is to serve as a Electric utilities reported that some major challenges on the
single reference for researchers, who will have been assessing power grid frequency’s behavior have taken their fastest
the performance of observers applied to modern power deviation in the time domain. It is due to the high renewable
systems under observability analysis. When the linear state- energy injected into the electric energy system. Even though
space equations belonging to the linear time-invariant (LTI) modern technologies can estimate the tracking of the grid
are set up herein. Besides, if this system is observable then frequency, the problem appears when the apparatus is down a
their key parameters must hold: failure. Figure 2 shows the tracking of either grid frequency
that is linked to a renewable energy system (RES) with 2%
ẋ ( t )= A ∙ x (t ) + B∙ u ( t ) , (the dashed blue line), 25 % (the dashed purple line), and 75
y ( t ) =C ∙ x ( t )+ D∙ u(t ) % (the red line). The simulation reveals that grid frequency
deviation is more unstable when justly a perturbation is
( 1) injected in a particular power grid as can be seen in the black
point, where starting the ROCOF analysis.
where x (t) and y (t ) are the state variables and the output
from the physical system in the time domain, respectively. A
and C are matrices where A ∈ R n× n, B∈ Rn × m, C ∈ R p ×n ,
and D is stated zero value because never its output is linked to
the input. Matrix A is also called the system matrix, which
shows the dynamical behavior of the system, and matrix C is
the output matrix, which gets the estimated values of its
sensor. n is the number of state variables of the system and p
is the size of the rows from the output matrix C or the number
of sensors installed inside a physical system.
Notice that, the size of matrices A , C can change their
Figure.2. Comparison of different frequency deviations (ROCOF) of an
values as can be seen in [8]; when the linearization technique electric utility with RES penetration (modified from [35]).
is applied to these in a fixed point of reference. The
Observability matrix always is set up across (2): If local or private utility control centers do not consider the
new challenges, eventually would bear with a sudden
T
[ C CA ⋯ C A n ] =M obs blackout or power outages on the power system. To overcome
these issues, many researchers are using tools related to
( 2) control system techniques under observability analysis, which
commonly are applied in control systems theory. It aims to
The remaining of this manuscript is divided into four parts.
estimate the major quantity of the electrical measurements
In the next section, the problem statement concerning the
(branch currents, node voltages, power system frequency, real
challenges of modern power networks was identified to deal
and reactive power, phase angles, etc.), which are considered
with in this manuscript. Section III presents the most relevant
the internal state variables of a power grid.
literature regarding observers applied to modern electric
It is only possible thanks to a state observer design. If the
power systems, meanwhile. Section IV is discussed in the
power system has a minimal number of onsite sensors
literature review, which has been significantly examined in the
providing actual measurement data in real-time operation.
preceding section, and Section V gets the conclusion of this
Thus, the key parameters can be estimated via a or more state
work.
observers in real-time performance. Moreover, tools or
I. PROBLEM OVERVIEW software belong transient simulations of the key parameters
from the electrical grid are verified to prove the effectiveness
of modern devices, such as hardware-in-the-loop (HIL). HIL
Given the transition from synchronous generators
provides an advanced monitor of transient simulation in real-
dominated by traditional power systems toward the power
time operation and examines the performance of these key
converter dominated across modern power systems will appear
parameters (e.g., OPAL-RT, software tools). In the same way,
a need of updating the electrical meters or transducers. If they
digital-Twins technologies provide several computational
are based on electronic inverters along with outdated
tools, which help to palliate the new challenges of modern
equipment from traditional power grids are prone to failures
power systems, which were made in [9] and [10].
inside the power grid. It is due to their deterioration, and lack
of onsite sensors by economic factors. Turns out they would
II.OBSERVERS IN MODERN ELECTRIC POWER SYSTEMS
have not provided accurate outcomes of measurement, which
are essential to the energy management system of a utility
Considering that a state observer is lineal, but the problems
control center.
on power systems widely are non-linear functions, hence
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

exists the need to design not only one observer but also two or u ∈ R is the control vector, such that ∆ f i is the ratio of
m

more that depend on the accuracy of outcomes in the state

deviation- frequency, ∆ Pi is the ratio of change from the
variables regarding the electric current, voltage, and grid
frequency. However, a non-linear function can be linearized power consumption, and ∆ X g is the ratio of change from the
by Euler and LaGrange method in specific fixed points on the governor valvular position, and by the last y is an output
tracking that causes itself. In that sense, this literature review vector.
will consider the most relevant investigations that describe the A set of disturbance signals had applied on a power system
performance of the observers in both conventional and non- model that is denoted by d (x , t , w). By using (3) on this
conventional power systems, which are using an or more model, one would leave in matrices:
observer design subject to requirements that imply the control
system approach such that observability analysis is used to
achieve the simulation results.

A. Observer-based sliding mode frequency control for power

systems along with renewable energy integration

Despite that, tracking of a grid frequency from a multi-

machine power system would have to obtain across modern
equipment. Nonetheless, the integration of renewable energy
sources into the power system cause a power grid frequency
deviation to be measured, and then a linearized observer based
on sliding mode frequency was set up in them [11]. Certainly,
when renewable energy sources deal whit photovoltaic
systems, the weather conditions constantly affect power
system stability, and its stochastic feature is of little benefit for
grid frequency. In consequence, the latter is susceptible to
unpredictably changing conditions, and its operational
capacity goes in decreasing. Despite having battery banks, it
only allows for the reduction of the peaks from swing Figure 2. Block diagram of the sliding mode control using disturbance
frequency. observer, (modified from [11]).
Automatic generation control (AGC) has been employed to
balance between supplied and consumed power across the unit
commitment method, maintaining the frequency in a set point
(e.g., 60 Hertz). When the stability and security of the no-
conventional power systems rely on the proper operation of
the AGC. Thus, a frequency correction has been estimated to
use the swinging changes on its grid frequency by way of a
disturbance observer.
A sliding mode control observer also was necessary to send
input variables toward the disturbance observer in the time
domain, which is designed to offset grid frequency
compensation by using the linear state-space equation.
The estimation can achieve by applying a linearization
technique to the modern power system's non-linear swing
equations. Figure 2 displays a block diagram of an observer
that justly is split into two schemes: a sliding surface
controller and a control input mode. Hence, the linearized
swing power equation was developed to attain the state-space
representation given by:

ẋ= Ax+ B1 u+B 2 d (x , t , w)

y=Cx
( 3)

where x=[ ∆ f i ∆ Pgi ∆ X gi ] ∈ R is the state vector,

n