Journal Pre-proofs

Advanced centralized protection system for AC microgrids using TW-DWT

techniques

Asmaa M. Aboelezz, Magda I. El-Afifi, Bishoy E. Sedhom, Abdelfattah A.

Eladl, Padmanaban Sanjeevikumar, Ahmed N. Sheta

PII: S2590-1745(24)00330-1
DOI: https://doi.org/10.1016/j.ecmx.2024.100852
Reference: ECMX 100852

To appear in: Energy Conversion and Management: X

Received Date: 20 September 2024

Revised Date: 11 December 2024
Accepted Date: 23 December 2024

Please cite this article as: A.M. Aboelezz, M.I. El-Afifi, B.E. Sedhom, A.A. Eladl, P. Sanjeevikumar, A.N. Sheta,
Advanced centralized protection system for AC microgrids using TW-DWT techniques, Energy Conversion and
Management: X (2024), doi: https://doi.org/10.1016/j.ecmx.2024.100852

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2024 Published by Elsevier Ltd.

 Advanced Centralized Protection System for AC
Microgrids Using TW-DWT Techniques
Asmaa M. Aboelezz1,2, Magda I. El-Afifi1,3, Bishoy E. Sedhom1, Abdelfattah A. Eladl1,
Padmanaban Sanjeevikumar4, and Ahmed N. Sheta1
1Dept. of Electrical Engineering, Faculty of Engineering, Mansoura University, Mansoura 35516, Egypt.
2Faculty of Engineering, Mansoura National University, Egypt.
3Nile Higher Institute for Engineering and Technology, Mansoura, Egypt.
4Department of Electrical Engineering, IT and Cybernetics, University of South-Eastern Norway, Porsgrunn, Norway

Abstract
Microgrid protection has made significant strides in recent years, yet distribution system operators face substantial
challenges. The advent of microgrid operation has revealed the limitations of existing protection devices,
particularly in recognizing, classifying, and locating various types of faults when in islanded mode. The dynamic
nature of microgrids complicates fault management. This paper presents a methodology for a central protection
unit that includes fault detection and classification, fault location estimation, and faulty line isolation. The proposed
method utilizes discrete wavelet transform and traveling wave techniques with a central protection unit. The central
protection unit depends on the LoRaWAN communication technique to transmit data between the central
protection unit and the system. A test system developed entirely in MATLAB/Simulink is used to demonstrate the
accuracy of each feature of the protection algorithm applied in microgrids. The proposed method is tested and
simulated on two systems: one is a ring system, and the other is an IEEE 13 bus system. Different types of faults at
various locations have been tested and simulated. The results show the method's effectiveness, indicating its
potential to enhance the current protection system and pave the way for implementing a modern microgrid station
protection system. The fault clearance time is within 11.5ms for the IEEE system and 7ms for the ring system.

Keywords: Microgrids, traveling wave, discrete wavelet transform, centralized protection.

1. Introduction
Renewable energy has emerged as the leading source of electricity in power grids. Several factors contribute to this
trend. First, fossil fuels have significant negative environmental impacts. Additionally, these fuels are depleted. Power
quality is another concern, as existing power systems are aging. Moreover, the global energy demand continues to increase
[1], [2]. Consequently, microgrids (MGs) have evolved to accommodate the extensive integration of renewable energy
sources (RESs). MGs include diverse forms of distributed generation, such as diesel engines, microturbines, fuel cells, and
renewable sources, to meet local energy demands. They can operate either in conjunction with the main grid or in an
islanded mode and can be configured in radial or loop designs [3].
Microgrids harness solar, wind, geothermal, biomass, ocean wave energy, and energy storage systems (ESSs) to
produce clean power. This method decreases dependence on conventional energy sources, reduces pollution, and provides
a sustainable and reliable energy supply. [3]–[5]. There are three types of microgrids: AC-MG, DC-MG, and hybrid
AC/DC-MG. AC MGs are of particular interest because they are compatible with existing AC power infrastructure and
the market predominantly uses AC loads [8]. As green energy usage increases and semiconductor switches improve, AC-
MGs have gained popularity. These smaller power systems address challenges faced by traditional power networks and
enhance smart grids with features like distributed energy resources (DERs), and digital and pilot communications. MGs
are capable of self-monitoring and resolving issues independently. Additionally, they can be remotely monitored and easily
adjusted as necessary [7]–[9]. As the penetration of DERs increases, traditional protection methods in existing distribution
networks become inadequate, necessitating advanced protection strategies for the reliable integration of renewable energy
sources in microgrids [10]–[12].
Integrating MGs into distribution networks (DS) significantly affects traditional protection strategies. Consequently,
ongoing research focuses on adapting conventional protection controls and measures to effectively tackle the emerging
challenges [13], [14]. During faults, MGs exhibit complex behavior due to bidirectional power flow and the interfaces
between inverters and various resources. These resources differ in their contribution to short-circuit currents, making fault
detection and MG protection crucial challenges to address. Besides, the presence of distributed power sources and the
extensive use of power electronic components like inverters raise the challenges [15]. Currently, most literature on
microgrids suggests using the characteristics of short-circuit faults, such as short-circuit current, for fault detection. For
instance, reference [16] utilizes the phase angle or amplitude difference of the sequence components of fault current as the
characteristic quantity to detect short-circuit faults. Although relying directly on short-circuit current characteristics is
simpler, it presents challenges in promptly detecting and accurately pinpointing short-circuit faults due to the necessity of

-1-
waiting for fault progression. In recent research [17], communication technology has been utilized to collect real-time data,
including output current, position, and control strategies of DGs. Intelligent algorithms are then applied for detecting and
localizing faults. These approaches share similarities with integrated protection schemes due to their high accuracy, but
they necessitate extensive information processing, leading to extended fault detection and localization times. Studies on
short-circuit fault detection and localization via the decomposition and analysis of transient electrical signals using
mathematical techniques such as wavelet transform have also demonstrated significant progress. In [18], the wavelet
transform is utilized to extract specific frequency band details from high-frequency transient signals during faults.
Detection of faults involves comparing the polarity of signals at both ends of a line with those at the bus's outgoing lines.
Ref. [19] introduces a method aimed at pinpointing high-impedance faults within distribution systems. This approach relies
on weighted least squares estimation and parameter error processing algorithms to accurately estimate fault distances. In
contrast, Ref. [20] utilizes wavelet transform to classify lightning strikes occurring on transmission lines. It suggests that
combining principal component analysis with wavelet transform enhances the extraction of fault features, thereby
improving fault identification effectiveness.
Traveling waves are utilized for fault detection in AC microgrids [21], providing fast detection speeds. However, they
face challenges like managing reflected waves, demanding precise time accuracy and data synchronization, and having
limited suitability for low-voltage microgrids with short lines. One emerging research trend is the utilization of
communication technology and advanced algorithms to analyze node information throughout the network domain, to
efficiently detect and pinpointing short-circuit faults. Besides, Ref. [23] suggests a traveling-wave-based protection method
aimed at accelerating the removal of faulty lines. However, this approach lacks the ability to precisely identify the faulted
section. When a fault occurs on a feeder, the entire feeder is disconnected, resulting in a broader blackout area. This
limitation makes it less suitable for multi-section distribution lines. Additionally, it fails to isolate faults from both ends,
which is critical for double- or multi-end distribution systems that include Distributed Generation (DG) sources. In
summary, most literature emphasizes the use of steady-state parameters following faults or employs a combination of
communication technologies and advanced algorithms to analyze large datasets for short-circuit fault detection in
microgrids. Limited research exists on early detection of short-circuit faults in microgrids, indicating significant potential
for improving the speed and accuracy of these detection methods. Wavelet transform and traveling waves have been
extensively used in traditional power grids, but their application for early detection and localization in microgrids remains
underexplored. Table 1 shows a comparison between the recently published papers in the field of fault detection and localization
techniques. According to this study, specific research gaps are identified as follows;
▪ Inability to recognize and classify faults in islanded mode: Existing protection devices have shown limited effectiveness
in identifying and classifying various fault types when microgrids operate in islanded mode, where there is no grid
support. This limitation can lead to delayed fault clearance and system instability.
▪ Challenges in fault location estimation: Traditional fault location techniques struggle to provide accurate fault
positioning in microgrids due to frequent changes in power flow and bidirectional energy exchange, making them
unsuitable for dynamic microgrid operations.
▪ Lack of robust protection adapted to microgrid dynamics: The dynamic nature of microgrids—characterized by frequent
switching between grid-connected and islanded modes—poses a challenge that traditional protection systems are not
equipped to handle, leading to misoperations and protection failure.
▪ Limitations in communication technologies for real-time data transfer: Current microgrid protection systems often rely
on communication networks that are either slow, expensive, or power-hungry for real-time fault detection and isolation.
These constraints limit the ability of the protection system to react promptly.
▪ Slow faulty line isolation and risk of cascading failures: Traditional protection schemes tend to isolate faulty lines too
slowly, increasing the risk of cascading outages. This makes it imperative to develop a faster and more responsive
protection mechanism tailored for microgrids.
The importance of Traveling Waves (TW) and Discrete Wavelet Transform (DWT) in the proposed method lies in their ability
to enhance fault detection, classification, and location estimation. Traveling Waves (TW) provide a fast and effective means of
identifying faults as they propagate through the electrical system. By analyzing the characteristics of TWs, the proposed method
can rapidly detect changes in current and voltage that indicate fault conditions. This leads to quicker fault recognition and
contributes to minimizing the risk of cascading failures. Discrete Wavelet Transform (DWT) is utilized to decompose the fault
signals into various frequency components, allowing for the extraction of critical features associated with fault events. DWT is
particularly effective in handling non-stationary signals, making it well-suited for the dynamic environments of microgrids. By
applying DWT, the method enhances the accuracy of fault classification, even in complex scenarios. The proposed method
addresses these challenges by implementing DWT and TW techniques within a central protection unit (CPU), ensuring precise
fault detection, classification, and location estimation. It uses the LoRaWAN communication technique to transmit data between
the system and the central protection unit. The proposed method was tested and simulated on two systems: the first is a ring
system and the other is an IEEE 13 bus system. The primary contributions of the paper include:
▪ Introduction of a novel centralized protection scheme specifically tailored for AC Microgrid (AC-MG) systems.
▪ Utilization of traveling wave (TW) techniques combined with discrete wavelet transform (DWT) to enhance the
detection, classification, and localization of faults.
▪ Integration of a central protection unit that leverages TW and DWT for comprehensive fault management.
▪ Implementation of LoRaWAN communication technology to enable efficient and reliable data transmission between
the system and the central protection unit.

-2-
 ▪ Demonstration of the method's applicability to different system configurations, showcasing its versatility and
robustness.
The structure of this paper is as follows: Section 2 describes the system configurations used to validate the proposed
scheme. Section 3 describes the fundamental principles, protection configuration, and criteria of the proposed protection
scheme. Section 4 presents the simulation results. Finally, Section 5 provides the concluding remarks for the paper.

Table 1: Comparison between the recently published papers for fault detection.

Ref Methodology Advantages limitations

Limited to certain fault types and
Convolutional neural High accuracy (93.1%) in fault
[24] conditions; requires high-quality
network classification; robust against noise.
data.
Simultaneous fault detection and May struggle with
[25] Decision tree algorithm classification for AC and DC lines; electromagnetic interference in
fast processing. hybrid systems.
Effective fault location using Complex implementation and
Parameter estimation
[26] multiple data sources; addresses reliance on accurate data inputs
technique
uncertainty in fault detection. for Bayesian networks.
Fast fault clearing times (1.32-1.8 Challenges with dynamic
Distance protection
[27] ms.); main and backup protection characteristics and fault current
scheme
capability. behavior.
- High flexibility, sensitivity,
and reliability; rapid fault detection
and isolation.
- Improve fault detection
across diverse conditions.
- Focuses on microgrid
Central Protection Unit adaptability, enhancing accuracy in
Needs to be tested under noise
Our paper utilizing TW and DWT real-world applications.
conditions.
techniques. - Combines data sources for
more reliable fault detection without
needing extensive historical data.
- Utilizes LoRaWAN for
communication, ensuring effective
coverage and reducing dependency
on traditional methods.

2. 2. System Description
The proposed method is tested and evaluated on two systems: the first represents a ring system, and the second
represents an IEEE 33 bus model as explained below.
2.1. Ring System
A general description of the medium voltage AC microgrid (MVAC-MG) configuration that was utilized to validate
the suggested protection strategy is given in this section. Figure. 1 illustrates the arrangement of the MVAC-MG
configuration. The network comprises four cables in its configuration, each with specific lengths for the respective links:
L1–3, L1–4, L2–3, and L2–4, with a length of 100 km for all of them. To aid in fault clearance, circuit breaker (CBs)
devices are placed at each link end. The MVAC-MG grid is designed with a symmetrical monopole configuration and
operates at a ± 11 kV voltage for point-to-point. The parameters of the MVAC-MG network are shown in Table 2.
Parameters of the equivalent π model for network cables are shown in Table 3. The communication technique is described
in section 3.

-3-
 Central Control Unit
Data Processing Data Storage Data Analysis

Actuation
Pervasive Sensing Commands
Measurements
High speed
communication network
LORAWAN

Bus 1 Bus 4
F1 F2 F3
R1 R2

R8 R3
PV Power Plant
Loads

R7

R4

R6 R5

Bus 3
Bus 2

Wind Power Plant Loads

Fig. 1. Configuration of the MVAC-MG.

Table 2: Parameters of MVAC-MG network.

ITEM ACTIVE POWER (MW)
AC Bus 1: Microgrid 1 Configuration 1.6
AC Bus 2: Microgrid 2 Configuration 2.3
AC Bus 3: Load 0.135
AC Bus 4: Load 0.93

Table 3: Parameters of the equivalent π model for network cables.

LINE R (Ω) L (MH) C (µF)
Line 12 1.723 0.09337 1.274
Line 13 1.273 0.09337 1.274
Line 24 1.723 0.09337 1.274
Line 34 1.723 0.09337 1.274

2.2. 2.2 IEEE 33 bus model

The topology of the proposed IEEE 33-bus distribution test system is illustrated in Fig. 2. Operating at 12.66 kV, the
system features a feeder substation, four distributed generation (DG) units, two reactive power compensation systems, 33
buses, and three looping branches with switchable tie lines. The test benchmark is available in unbalanced three-phase
systems as described in [28].

-4-
 Central Control
Data
Unit Data
Data Storage
Processing Analysis

Actuation
Pervasive
Commands
Sensing
High speed
Measurements
communication
network
LORAWAN

BES 4 PV 2

19 20 21 22

PV 1 BES 3 Unit 1
WT 1
BES 1
BES 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

MV Substation
BES 6 WT 2
BES 5 Unit 2

26 27 28 29 30 31 32 33

23 24 25

Fig. 2. IEEE 33 bus configuration.

3. 3 Proposed central protection approach for MVAC-MG

Microgrids, with their complex operating conditions and dynamic structures, require sophisticated protection
strategies. This proposed system integrates all relays and distributed generators within the microgrid through constant
communication with a Central Protection Unit (CPU). This centralized approach utilizes Traveling Waves (TWs) and
Discrete Wavelet Transform (DWT) techniques for effective fault detection and location. The CPU leverages TWs and
DWT for data acquisition, extracting critical features from traveling waves to aid in fault detection. Communication
protocols facilitate the exchange of wavelet-transformed data, ensuring coordinated protection and control, thereby
enhancing grid reliability. The CPU employs an interrupt-based algorithm, updating relays with new fault currents upon
detecting interruptions. Relays then operate independently to isolate faults by opening connections and setting fault
detection bits when currents exceed the threshold as described in [29].
3.1. Fault Detection and Location
TWs are electromagnetic waves that traverse cables and lines within the electrical system, generated by switching events,
faults, or lightning. When an incident TW reaches a power grid junction or terminal where circuit parameters change, it splits
into reflected and refracted components as shown in Fig. 3. The reflection rate is the ratio of reflected to incident TW amplitudes,
while the refraction rate is the ratio of refracted to incident TW amplitudes. A simplified transmission line model helps derive
fault detection equations from transmission line and wave propagation principles [30], [31]. Wavelet Transform (WT) is used to
extract data from fault current traveling waves, detecting internal faults through the maximum coefficient from wave analysis
via DWT. When the wave exceeds a predefined range, the breaker isolates the fault. DWT, a type of wavelet transform, selects
wavelets at specific intervals, capturing both frequency and location information, unlike the Fourier transform [32]. After
simulating various operating and fault scenarios, unique signals for each scenario are generated.
Wavelet transform of a signal 𝑓(𝑡) 𝜖 𝐿2(𝑅), where 𝐿2 is the lebesgue vector space, is defined by the inner product between
𝛹𝑎𝑏(𝑡) 𝑎𝑛𝑑 𝑓(𝑡) as:

-5-
 ∞
1 𝑡―𝑏
𝑊𝑇(𝑓,𝑎,𝑏) = 𝑓(𝑡)Ψ 𝑑𝑡 (1)
𝑎 ―∞ 𝑎

The wavelet transform of the sampled waveforms is obtained by implementing DWT given by:
1 𝑛 ― 𝑘𝑎𝑚
0
𝐷𝑊𝑇(𝑓,𝑚,𝑛) = 𝑓(𝑡)Ψ( ) (2)
𝑎𝑚
0
𝑎𝑚0
𝑘

where 𝑎 and 𝑏 from the previous equation are replaced by 𝑎𝑚 𝑚

0 𝑎𝑛𝑑 𝑘𝑎0 , k and m being integer variables. In a standard DWT,
the coefficients are sampled from a continuous WT on a dyadic grid, a0=2 and b0=1, yielding a00=1, a0-1=2-1, etc.
After detecting the fault, the CPU calculates the time of arrival and the amplitude of the TWs at different measurement
points. The differences in arrival times and the changes in wave amplitudes are used to triangulate the fault location. The
equations for fault location are derived from the principles of wave propagation and the characteristics of the transmission line
as follows [21]:
𝑣 ∗ 𝑡𝑑
𝑥= (3)
2
where 𝑥 is the position of the fault, 𝑣 is the propagation speed of the TW in the line, and 𝑡𝑑 is the time difference between the
first arrival peaks at the two ends of the line.
By integrating the TW and DWT data, the CPU can accurately determine the fault's position along the line, enabling precise
isolation and minimizing the impact on the microgrid's operation.
Travelling Wave
Bus 1 Fault Point Bus 2

Reflected Wave Refracted Wave

Fig. 3. TW-based protection approach.

3.2. Communication Scheme

Efficient communication is crucial for the centralized protection approach to function effectively. In this proposed system,
LoRaWAN (Long Range Wide Area Network) is utilized for communication between the relays, distributed generators, and the
CPU. LoRaWAN is chosen for its long-range, low-power consumption, and robustness, making it ideal for microgrid
applications. LoRaWAN operates on unlicensed radio frequency bands and provides bi-directional communication, ensuring that
data from the relays and generators is reliably transmitted to the CPU. This communication network supports the IEC 61850
protocol, allowing for the exchange of wavelet-transformed data essential for coordinated protection and control. Using
LoRaWAN, the system can handle the following tasks: 1) Ensures that fault detection and location data is promptly sent to the
CPU for analysis. 2) Offers redundancy and reliability in data transmission, critical for maintaining grid stability. 3) Supports a
scalable network architecture, allowing for the addition of new devices and sensors without significant changes to the existing
infrastructure.
LoRaWAN uses Chirp Spread Spectrum (CSS) for message transmission, allowing transfer rates of up to 50 kbps,
depending on the spread factor and channel bandwidth. A significant feature of LoRaWAN is its capacity to receive multiple
messages with different spreading factors simultaneously at LoRa base stations. Each message can include a payload of up to
243 bytes. Operating in sub-GHz frequency bands like 868 MHz in Europe, 915 MHz in North America, and 433 MHz in Asia,
LoRaWAN effectively covers long distances (over 18 km) while maintaining low power consumption. A single LoRa gateway
or base station can potentially cover entire cities or large regions spanning hundreds of square kilometers, although actual
coverage varies based on environmental conditions and obstacles. By adopting LoRaWAN technology and the open LoRaWAN
protocol, superior coverage is achieved compared to other standardized communication technologies [33]. In the proposed study,
standardized time synchronization protocols within the LoRaWAN framework are implemented to maintain network-wide
synchronization. The proposed LoRaWAN network uses regular time synchronization beacons sent by the gateway. These

-6-
beacons help LoRaWAN devices adjust their internal clocks, ensuring synchronization across the network [34]. To address signal
propagation and fading issues in the wireless medium, the Rayleigh fading model is implemented. This model accounts for
channel gain coefficients and represents the system channel impulse response as a Gaussian process with zero mean and unit
variance [35]. Additionally, path loss caused by refraction, diffraction, and reflection is analyzed using the Friis transmission
equation over free space, a well-known method [35].
The proposed supervisory monitoring system, depicted in Fig. 4, involves LoRaWAN devices gathering data and
transmitting it to the LoRaWAN gateway, which then forwards the data to a central monitoring server capable of multiple tasks.
Before each measurement cycle, the central server sends a synchronization request to all LoRaWAN devices. The devices
respond with their current time information, allowing the central server to calculate synchronization errors and delays. If any
device shows errors or delays beyond acceptable thresholds, the server instructs the LoRaWAN gateway to correct these issues.
After synchronization is achieved, the central server sends a measurement request to all devices, which then collects and transmits
data back to the gateway. The flowchart of the presented method is illustrated in Fig. 5.

PowerEdge
6500

LoRa Device LoRa Gateway Server Monitoring system

Fig. 4. Supervisory monitoring system using LoRaWAN.

Start

Initialize Data Acquisition

Detect and Analyze Traveling

Waves (TWs)

Apply Discrete Wavelet

Transform (DWT)

No
(Is fault is detected)

Yes

Analyze Time and Amplitude

of TWs

Calculate fault location using

Eq. (3)

Sent tripping signals

to the related CBs

End

Fig. 5. The TW-DWT-based fault detection and location estimation flowchart.

4. Simulation Results and Discussion

The MATLAB/Simulink environment and the Real Time Innovation (RTI) Connext connectivity framework were used in
this paper to conduct simulation implementation. To facilitate smooth communication between the different DDS members,
Simulink blocks specifically designed for DDS were integrated into the Simulink model. The process of simulation entailed
using MATLAB/Simulink Coder to generate C/C++ code from the model; the code complied with the RTI Connext DDS
requirements for the DDS blocks. After that, the generated code was executed on a system that was capable of running the RTI
Connext DDS or RTI Connext Micro DDS framework. By evaluating the suggested protection system in various fault scenarios,

-7-
its efficacy was evaluated. These tests proved the suggested approach's ability to detect, locate, and isolate faults, guaranteeing
complete protection of the system. In order to precisely locate the fault, the frequency bands have important information about
the fault in the system. A signature signal for each fault scenario is created by combining all these signals into one. Ultimately,
the corresponding CB received tripping signals in order to clear the fault. In this work, several different conditions were analyzed
within the two proposed systems to evaluate the effectiveness of the suggested protection scheme, as follows:
 Scenario I: Different fault types for different fault locations in the ring system.
➢ Case I: Three phase fault.
➢ Case II: Phase to ground fault.
➢ Case III: Phase to phase to ground fault.
 Scenario II: Different fault types for different fault locations in the IEEE 33 bus system.
➢ Case I: Symmetrical Fault (ABC_G) in Line 1-2.
➢ Case II: Unsymmetrical Fault (A_G) in Line 1-2.
➢ Case III: Symmetrical Fault (ABC_G) in Line 18-33

4.1. Scenario I: Different fault types for different fault locations in the ring system:
4.1.1. Case I: Three phase fault
In this case, a three-phase-to-ground fault is applied at 20% of lines 1-4, occurring at 0.5 seconds at F1 as depicted in Fig.
1. The fault resistance is 0.0001 Ω. Fig. 6 shows the system current during the fault. The current in the faulted section
experiences a significant increase, reaching its peak immediately after the fault occurrence at 0.5 seconds. Following the fault
isolation, the current values become zero as the fault occurs at the three phases. Tripping signals are shown in Fig. 7 which
emphasizes the accuracy of the proposed method for fault classification. The current coefficients of all the phases are shown
in Fig. 8. The fault is cleared within 7 ms.

Fig. 6. AC line currents during three phase fault at F1.

Fig. 7. Trip Signals during three phase fault at F1.

-8-
 (a) (b)

(c) (d)

Fig. 8. Current Coefficients during three-phase fault at F1: a) Current coefficients at Phase A. b) Current coefficients at
Phase B. c) Current coefficients at Phase C. d) Current coefficients at Phase ground.

4.1.2. Case II: Phase to ground fault

In this case, a phase-to-ground fault (C-G) is applied at 60% of lines 1-4, occurring at 0.5 seconds at F2 as depicted in Fig.
1. The fault resistance is 0.0001 Ω. Fig. 9 shows the system current during the fault. The current in the faulted section
experiences a significant increase, reaching its peak immediately after the fault occurrence at 0.5 seconds. Following the fault
isolation, the current in the faulted phase drops to zero, while the healthy phases continue to operate normally. This selective
isolation is crucial for maintaining system stability and reliability. Tripping signals, shown in Fig. 10, highlight the accuracy
of the proposed fault classification method, ensuring that only the faulted phase is isolated while the other healthy phases
remain operational. The current coefficients of all the phases are shown in Fig. 11. The fault is cleared within 7.5 ms,
showcasing the system's rapid response and effectiveness in isolating the faulted phase and preserving the functionality of the
remaining phases. This approach not only minimizes disruption but also enhances the overall reliability of the power system.

Fig. 9. AC line currents during three phase fault at F.

-9-
Fig. 10. AC line currents during three phase fault at F2.

(a) (b)

(c) (d)

Fig. 11. Current Coefficients during three-phase fault at F12: a) Current coefficients at Phase A. b) Current coefficients at
Phase B. c) Current coefficients at Phase C. d) Current coefficients at Phase ground.

4.1.3. Case III: Phase to phase to ground fault

In this case, a phase-to-phase-to-ground fault (AB-G) is applied at the end of lines 1-4, occurring at 0.5 seconds at F3 as
depicted in Fig. 1. The fault resistance is 0.0001 Ω. Fig. 12 shows the system current during the fault. The current in the faulted
section experiences a significant increase, reaching its peak immediately after the fault occurrence at 0.5 seconds. Following
the fault isolation, the current in the faulted phase drops to zero, while the healthy phases continue to operate normally. This
selective isolation is crucial for maintaining system stability and reliability. Tripping signals, shown in Fig. 13, highlight the
accuracy of the proposed fault classification method, ensuring that only the faulted phase is isolated while the other healthy
phases remain operational. The current coefficients of all the phases are shown in Fig. 14. The fault is cleared within 7 ms,
showcasing the system's rapid response and effectiveness in isolating the faulted phase and preserving the functionality of the
remaining phases. This approach not only minimizes disruption but also enhances the overall reliability of the power system.

- 10 -
Fig. 12. AC line currents during three phase fault at F3.

Fig. 13. AC line currents during three phase fault at F3.

(a) (b)

(c) (d)
Fig. 14. Current Coefficients during three-phase fault at F3: a) Current coefficients at Phase A. b) Current coefficients at Phase
B. c) Current coefficients at Phase C. d) Current coefficients at Phase ground.

- 11 -
4.2. Scenario II: Different fault types in the IEEE system
4.2.1. Case I: Symmetrical Fault (ABC-G) in Lines 1-2
Line 1-2 is the connection between the generation bus and bus 2 as depicted in Fig. 2, making it the line with the highest
current during both normal operation and fault conditions. Under normal operation, the current in this line is 175 A. However,
during a fault, the current can exceed ten times this value. Any fault in this line results in the isolation of the entire system from
the generation unit. In this case, a symmetrical fault (ABC_G) occurred in the first line at 0.1 sec after the generation unit. The
fault current, as shown in Fig. 15, is extremely high. The tripping signal is shown in Fig. 16. The CPU detected the fault after
1.5 milliseconds and took 11.5 milliseconds to clear it.

Fig. 15. Current phases in line1-2 at ABC-G fault.

Fig. 16. Trip Signal with (ABC-G) fault applied in line1-2.

4.2.2. Case II: Unsymmetrical Fault (A-G) in Lines 1-2

In this case, an unsymmetrical fault (A_G) also occurred in line1-2 at 0.1 sec after the generation unit. Similar to the first
case, the fault current was very high as depicted in Fig. 17. The tripping signal is shown in Fig. 18. The CPU detected the fault
after 1.5 milliseconds and took 11.5 milliseconds to clear it.

Fig. 17. Current phases in bus 12.

- 12 -
Fig. 18. Trip Signal, by applying a central protection scheme with (A-G) fault applied to bus 12.

4.2.3. Case III: Symmetrical Fault (ABC-G) in Line 18-33

Line 18-33 is located farther from the generation unit as shown in Fig.2, running between bus 18 and the end of bus 33.
Under normal conditions, the current in this line is 7 A, but during a fault, the current can be ten times higher. In this case, a
symmetrical fault (ABC_G) occurred in lines 18-33 at 0.1 seconds, far from the generation unit. Consequently, the fault current,
as depicted in Fig.19, is not very high. The tripping signal is shown in Fig. 20. The CPU detected the fault after 2.5 milliseconds
and took 12.5 milliseconds to clear it.

Fig. 19. Current phases in bus 18-33.

Fig. 20. Trip Signal, by applying a central protection scheme with (ABC-G) fault applied to bus 18-33.

Table 3 summarizes the clearing times for various fault types in two scenarios: Scenario I (Ring System) and Scenario II
(IEEE 33 bus system). It is concluded that different fault types at different locations can be detected and cleared using the
proposed method. In general, the radial system (Scenario II) shows higher clearing times across fault types compared to the ring
system (Scenario I). For instance, in Case I with an ABC-G fault, the clearing time is 7 ms in the ring system, while it increases
to 11.5 ms in the radial system. Similarly, in Case II, a C-G fault in the ring system clears in 7.5 ms, whereas the same fault takes
11.5 ms to clear in the radial system. This trend is also observed in Case III, where an AB-G fault clears in 7 ms in the ring
system but takes 12.5 ms in the radial system. The ring system generally demonstrates faster clearing times, likely due to its
interconnected structure, which provides multiple pathways for fault current and faster fault isolation. In contrast, the radial
system’s higher clearing times may be attributed to its single-path configuration, which limits the system's flexibility in isolating
faults, resulting in a delayed response. These findings suggest that while both system configurations can detect and clear faults
effectively using the proposed method, the ring system may be more efficient in terms of fault-clearing speed. This efficiency
could be beneficial in applications where quick response times are critical. On the other hand, the radial system’s longer clearing
times indicate a potential trade-off between simplicity in design and response speed.

- 13 -
Table 3: Comparison between the two scenarios.
Scenario I (Ring System) Scenario II (IEEE 33 bus)
Case
Fault Type Fault Location Clearing Time (ms) Fault Type Fault Location Clearing Time (ms)
Case I ABC-G 20% of the length of line 1-4 7 ABC-G Line 1-2 11.5
Case II C-G 60% of the length of line 1-4 7.5 A-G Line 1-2 11.5
Case III AB-G At the end of lines 1-4 7 ABC-G Line 18-33 12.5

5. Conclusion
This paper presents a CPU based on the TW-DWT protection scheme, validated through extensive simulations on two
systems: an AC microgrid with a ring configuration and the IEEE 33-bus AC distribution system. The TW-DWT scheme
showcases high flexibility, sensitivity, reliability, and directional accuracy, making it particularly effective for microgrids and
offering substantial advantages over traditional protection schemes. The study investigates two distinct scenarios. The first
scenario focuses on the ring microgrid, where three different fault types: three-phase faults, phase-to-ground faults, and phase-
to-phase-to-ground faults are tested across multiple fault locations. In this setup, fault clearance times range between 7 and 7.5
milliseconds. The second scenario explores fault behavior within the IEEE 33-bus system. This includes symmetrical faults
(ABC_G) and unsymmetrical faults (A_G) occurring at various locations, such as between lines 1-2 and 18-33. The fault
clearance times for this scenario vary between 11.5 and 12.5 milliseconds, reflecting the system’s adaptability to complex fault
conditions. The proposed CPU enhances overall system reliability and simplifies control processes, enabling precise and rapid
fault detection and location, along with remote management of microgrid operations. The effectiveness of the CPU in
implementing the TW-DWT scheme demonstrates its potential for advanced microgrid protection. Future research will focus on
integrating artificial intelligence (AI) techniques with the proposed method, specifically artificial neural networks (ANNs), to
further improve fault identification and localization, particularly within the bus sections of the simulated network. These AI-
based modules will complement the existing protection system, enhancing the accuracy and efficiency of fault detection in
increasingly complex operational environments.

References
[1] A. M. ; Aboelezz, B. E. Sedhom;, and M. M.El-Saadawi, “IEC 61850 Communication-Based Pilot Distance Protective IED
for Fault Detection and Location in DC Zonal Shipboard Microgrid,” IEEE Trans. Ind. Appl., vol. 59, no. 5, pp. 5559–
5569, 2023.
[2] A. M. Aboelezz, M. M. El-Saadawi, A. A. Eladl, V. Bures, and B. E. Sedhom, “A Novel Fault Detection and Location
Approach for DC Zonal Shipboard Microgrid based on High-Frequency Impedance Estimation with IEC 61850
Communication Protocol,” IEEE Access, vol. 12, March 2024, pp. 36212–36228, doi: 10.1109/ACCESS.2024.3374198.
[3] A. N. Sheta, G. M. Abdulsalam, B. E. Sedhom, and A. A. Eladl, “Comparative framework for AC-microgrid protection
schemes: challenges, solutions, real applications, and future trends,” Prot. Control Mod. Power Syst., vol. 8, pp. 1–40,
2023, doi: 10.1186/s41601-023-00296-9.
[4] B. E. Sedhom, M. M. El-Saadawi, M. S. El Moursi, M. A. Hassan, and A. A. Eladl, “IoT-based optimal demand side
management and control scheme for smart microgrid,” Int. J. Electr. Power Energy Syst., vol. 127, p. 106674, 2021,
[Online]. Available: https://doi.org/10.1016/j.ijepes.2020.106674.
[5] A. M. ; Aboelezz, B. E. Sedhom, M. A. Saeed, A. A. Eladl, and M. M. El-Saadawi, “Fuzzy Logic Approach Based Fault
Detection in Multi Terminal HVDC Transmission Systems,” 2023 24th Int. Middle East Power Syst. Conf., no. 19-21
December, 2023, doi: 10.1109/MEPCON58725.2023.10462447.
[6] M. A. Aftab, S. M. S. Hussain, I. Ali, and T. S. Ustun, “Dynamic protection of power systems with high penetration of
renewables: A review of the traveling wave based fault location techniques,” Int. J. Electr. Power Energy Syst., vol. 114,
p. 105410, 2020, doi: 10.1016/j.ijepes.2019.105410.
[7] A. N. Sheta, B. E. Sedhom, A. Pal, M. S. El Moursi, and A. A. Eladl, “Stability-Constrained Settings of Directional
Overcurrent Relays with Shifted User-Defined Characteristics for Distribution Networks with DERs,” IEEE Trans. Power
Deliv., vol. 39, no. 4, pp. 2401–2413, 2024, doi: 10.1109/TPWRD.2024.3403921.
[8] A. A. Eladl, M. A. Saeed, B. E. Sedhom, A. Y. Abdelaziz, and V. Bureš, “LoRaWAN-based IoT protocol for three levels
central protection scheme in MT-HVDC networks with ANFIS-PSO restoration algorithm,” Energy Reports, vol. 10, no.
November, pp. 4556–4573, 2023, doi: 10.1016/j.egyr.2023.11.013.
[9] A. M. Aboelezz, B. E. Sedhom, and M. M. El-Saadawi, “Pilot Distance Protection Scheme for DC Zonal Shipboard
Microgrid,” in 2021 4th International Symposium on Advanced Electrical and Communication Technologies (ISAECT),
Jan. 2021, pp. 01–06, doi: 10.1109/ISAECT53699.2021.9668537.
[10] M. Uzair, L. Li, M. Eskandari, J. Hossain, and J. G. Zhu, “Challenges, advances and future trends in AC microgrid
protection: With a focus on intelligent learning methods,” Renew. Sustain. Energy Rev., vol. 178, no. February, 2023, doi:
10.1016/j.rser.2023.113228.
[11] W. M. Hamanah, M. I. Hossain, M. Shafiullah, and M. A. Abido, “AC Microgrid Protection Schemes: A Comprehensive
Review,” IEEE Access, vol. 11, pp. 76842–76868, 2023, doi: 10.1109/ACCESS.2023.3298306.
[12] J. I. D. Cisneros-Saldana, S. Samal, M. M. Begovic, and S. R. Samantaray, “On Protection Schemes for AC Microgrids:
Challenges and Opportunities,” IEEE Trans. Ind. Appl., vol. 60, no. 3, pp. 4843–4854, 2024, doi:

- 14 -
 10.1109/TIA.2024.3353721.
[13] M. Bakkar, S. Bogarra, F. Córcoles, A. Aboelhassan, S. Wang, and J. Iglesias, “Artificial Intelligence-Based Protection for
Smart Grids,” Energies, vol. 15, no. 13. 2022, doi: 10.3390/en15134933.
[14] A. M. Aboelezz, B. E. Sedhom, and M. M. El-Saadawi, “Intelligent Distance Relay based on IEC 61850 for DC Zonal
Shipboard Microgrid Protection,” in 2022 Second International Conference on Power, Control and Computing
Technologies (ICPC2T), Mar. 2022, pp. 1–5, doi: 10.1109/ICPC2T53885.2022.9776712.
[15] N. Hussain, M. Nasir, and J. C. Vasquez, “Recent Developments and Challenges on AC Microgrids Fault Detection and
Protection Systems – A Review,” Energies, vol. 13, no. 9, p. 2149, 2020.
[16] F. Zhang, L. Mu, and W. Guo, “An Integrated Wide-Area Protection Scheme for Active Distribution Networks Based on
Fault Components Principle,” IEEE Trans. Smart Grid, vol. 10, no. 1, pp. 392–402, 2019, doi: 10.1109/TSG.2017.2741060.
[17] V. C. Nikolaidis, E. Papanikolaou, and A. S. Safigianni, “A communication-assisted overcurrent protection scheme for
radial distribution systems with distributed generation,” IEEE Trans. Smart Grid, vol. 7, no. 1, pp. 114–123, 2016, doi:
10.1109/TSG.2015.2411216.
[18] X. Zheng, Y. Zeng, M. Zhao, and B. Venkatesh, “Early Identification and Location of Short-Circuit Fault in Grid-
Connected AC Microgrid,” IEEE Trans. Smart Grid, vol. 12, no. 4, pp. 2869–2878, 2021, doi: 10.1109/TSG.2021.3066803.
[19] J. U. N. Nunes, A. S. Bretas, N. G. Bretas, A. R. Herrera-Orozco, and L. U. Iurinic, “Distribution systems high impedance
fault location: A spectral domain model considering parametric error processing,” Int. J. Electr. Power Energy Syst., vol.
109, no. July 2018, pp. 227–241, 2019, doi: 10.1016/j.ijepes.2019.02.012.
[20] J. A. Morales, E. Orduña, C. Rehtanz, R. J. Cabral, and A. S. Bretas, “Comparison between Principal Component Analysis
and Wavelet Transform ’Filtering Methods for Lightning Stroke Classification on Transmission Lines,” Electr. Power Syst.
Res., vol. 118, pp. 37–46, 2015, doi: 10.1016/j.epsr.2014.05.018.
[21] H. A. Abd El-Ghany, A. M. Azmy, and A. M. Abeid, “A General Travelling-Wave-Based Scheme for Locating
Simultaneous Faults in Transmission Lines,” IEEE Trans. Power Deliv., vol. 35, no. 1, pp. 130–139, 2020, doi:
10.1109/TPWRD.2019.2931178.
[22] X. Dong, J. Wang, S. Shi, B. Wang, B. Dominik, and M. Redefern, “Traveling wave based single-phase-to-ground
protection method for power distribution system,” CSEE J. Power Energy Syst., vol. 1, no. 2, pp. 75–82, 2015, doi:
10.17775/cseejpes.2015.00022.
[23] Q. Jia, X. Dong, and S. Mirsaeidi, “A traveling-wave-based line protection strategy against single-line-to-ground faults in
active distribution networks,” Int. J. Electr. Power Energy Syst., vol. 107, no. December 2018, pp. 403–411, 2019, doi:
10.1016/j.ijepes.2018.11.032.
[24] A. P. Shafei, J. F. A. Silva, and J. Monteiro, “Convolutional Neural Network Approach for Fault Detection and
Characterization in Medium Voltage Distribution Networks,” e-Prime - Adv. Electr. Eng. Electron. Energy, p. 100820,
2024, doi: 10.1016/j.prime.2024.100820.
[25] M. Fayazi, A. Saffarian, M. Joorabian, and M. Monadi, “An AI-based fault detection and classification method for hybrid
parallel HVAC/HVDC overhead transmission lines,” Electr. Power Syst. Res., vol. 238, no. September 2024, p. 111083,
2025, doi: 10.1016/j.epsr.2024.111083.
[26] H. Wang, C. Huang, H. Yu, J. Zhang, and F. Wei, “Method for fault location in a low-resistance grounded distribution
network based on multi-source information fusion,” Int. J. Electr. Power Energy Syst., vol. 125, no. October 2020, p.
106384, 2021, doi: 10.1016/j.ijepes.2020.106384.
[27] A. M. Aboelezz et al., “Efficient fault detection , localization , and isolation in MT-HVDC systems based on distance
protection and LoRaWAN communication,” Energy Reports, vol. 10, pp. 4183–4197, 2023, doi:
10.1016/j.egyr.2023.10.083.
[28] S. H. Dolatabadi, M. Ghorbanian, P. Siano, and N. D. Hatziargyriou, “An Enhanced IEEE 33 Bus Benchmark Test System
for Distribution System Studies,” IEEE Trans. Power Syst., vol. 36, no. 3, pp. 2565–2572, 2021, doi:
10.1109/TPWRS.2020.3038030.
[29] T. S. Ustun, C. Ozansoy, and A. Zayegh, “A microgrid protection system with central protection unit and extensive
communication,” 2011 10th Int. Conf. Environ. Electr. Eng. EEEIC.EU 2011 - Conf. Proc., no. May 2011, 2011, doi:
10.1109/EEEIC.2011.5874777.
[30] F. Wilches-Bernal et al., “A Survey of Traveling Wave Protection Schemes in Electric Power Systems,” IEEE Access, vol.
9, pp. 72949–72969, 2021, doi: 10.1109/ACCESS.2021.3080234.
[31] A. Sheta, G. Abdulsalam, and A. Eladl, “A Survey of Fault Location Techniques for Distribution Networks,” Bull. Fac.
Eng. Mansoura Univ., vol. 45, no. 2, pp. 12–22, 2020, doi: 10.21608/bfemu.2020.98825.
[32] T. Hubana, “Artificial Intelligence based Station Protection Concept for Medium Voltage Microgrids,” 2020 19th Int. Symp.
INFOTEH-JAHORINA, INFOTEH 2020 - Proc., March 2020,, pp. 18–20, doi: 10.1109/INFOTEH48170.2020.9066305.
[33] N. C. Gaitan, “A long-distance communication architecture for medical devices based on lorawan protocol,” Electron., vol.
10, no. 8, 2021, doi: 10.3390/electronics10080940.
[34] H. Alahmadi, F. Bouabdallah, and A. Al-Dubai, “A novel time-slotted LoRa MAC protocol for scalable IoT networks,”
Futur. Gener. Comput. Syst., vol. 134, pp. 287–302, 2022, doi: 10.1016/j.future.2022.04.003.
[35] H. Rafiqi, G. Mahendru, and S. H. Gupta, “Effect of Relay-based Communication on Probability of Detection for Spectrum
Sensing in LoRaWAN,” Wirel. Pers. Commun., vol. 130, no. 4, pp. 2345–2366, 2023, doi: 10.1007/s11277-023-10273-y.

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships that

- 15 -
could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered as
potential competing interests:

- In this setup, fault clearance times range between 7 and 7.5 milliseconds. The second scenario explores fault behavior
within the IEEE 33-bus system.
- This includes symmetrical faults (ABC_G) and unsymmetrical faults (A_G) occurring at various locations, such as
between lines 1-2 and 18-33. The fault clearance times for this scenario vary between 11.5 and 12.5 milliseconds,
reflecting the system’s adaptability to complex fault conditions.
- The proposed CPU enhances overall system reliability and simplifies control processes, enabling precise and rapid
fault detection and location, along with remote management of microgrid operations.
- The effectiveness of the CPU in implementing the TW-DWT scheme demonstrates its potential for advanced microgrid
protection.
- Future research will focus on integrating artificial intelligence (AI) techniques with the proposed method, specifically
artificial neural networks (ANNs), to further improve fault identification and localization, particularly within the bus
sections of the simulated network.

- 16 -