2024 3rd International Conference on Energy and Electrical Power Systems (ICEEPS)

Critical Events/Moments Identification Method for

Power System Transient Stability Analysis
Jiayang Zhou1, Shibin Bai2, Liang Zheng3, Zhihao Tian2, Zelun Sun3, Jiabin Zhang3, Songhao Yang1*
1School of Electrical Engineering, Xi’an Jiaotong University, Xi’an, China.

2State Grid Ningxia Electric Power Co., Ltd., Yinchuan, China.

3NARI Group Corporation (State Grid Electric Power Research Institute), Nanjing, China.

*[email protected]
2024 3rd International Conference on Energy and Electrical Power Systems (ICEEPS) | 979-8-3503-7513-8/24/$31.00 ©2024 IEEE | DOI: 10.1109/ICEEPS62542.2024.10693082

Abstract—The continuous integration of renewable energy DC blocking, etc. And critical moments refer to the times
generation (REG) into power systems has increased the when the dominant stability issue changes from one to another,
complexity of their transient characteristics. This complexity is or the time of transition from stable to unstable.
manifested in the coupling and transition of various transient
stability issues, typically driven by critical events such as grid- In the research of power system transient stability,
disconnection of REG and DC communication failures. Existing IEEE/CIGRE Joint Task Force on Stability Terms and
methods for analyzing individual transient stability issues are Definitions[1] proposed definitions and classifications of
inadequate to handle such complex scenarios. In response to this power system stability that laid the groundwork for subsequent
problem, this paper proposes a method for identifying critical study of power system stability in 2004. [2] introduced a
events/moments for transient stability analysis. Firstly, model for transient rotor angle and voltage stability
evaluation metrics for different stability issues are introduced, discrimination and employed a deep convolutional neural
establishing a quantitative evaluation scheme for transient network for training. [3],[4] proposed methods for power
stability. Secondly, criteria for identifying the dominant stability system transient voltage stability assessment based on
issue of the system are provided, and a scheme for identifying convolutional neural network (CNN) and temporal
critical events and moments is proposed. Finally, the effectiveness convolution network (TCN) respectively. [5] presented a
of the proposed method is validated through an improved IEEE unified scheme for transient stability simulation and
39-bus system.
assessment based on MATLAB and the PSD-BPA. This
Keywords-critical events/moments; transient stability; cascading
scheme enabled the simulation of numerous power system
failure simulation; renewable energy grid-disconnection; DC transient scenarios.
blocking In the research of cascading failures, three types of models
have been proposed: dynamic models, quasi-dynamic models
I. INTRODUCTION and steady-state models [6]. Song et al.[7] introduced a
nonlinear dynamic model of cascading failure, called
According to the "dual carbon" goals and the requirements
cascading outage simulator with multi-process integration
of sustainable energy development strategies, large-scale of
capabilities (COSMIC). This model can be used to study a
REG gradually replaces synchronous generators, integrated
wide variety of different mechanisms of cascading outages and
into power systems widely. The integration of REG and DC
provide detailed behaviors of the system under outages. Yao et
transmission makes the dynamic process of the power system
al.[8] proposed a multi-timescale cascading outage model
increasingly complex. The occurrence of cascading failures
using a quasi-dynamic simulation method. This model
has become more frequent, tightly linking the response
categorizes the dynamic processes into three timescales:
processes across different timescales. Traditional stability
Short-Term, Mid-Term and Long-Term. And they established
issues such as rotor angle, voltage and frequency intertwine,
a framework for simulating interactions among dynamics in
and the transient process is more complex and ever-changing.
different timescales. With the widespread integration of
This significantly affects stability analysis and increases the
renewable energy in power system, Zhu et al.[9] considered
complexity of simulation for modern power systems.
the impact of over/under-voltage and proposed a GCN-LSTM
Due to the intertwining and mutual transformation of model to map the operating states of power system, and then
stability issues, traditional methods for transient stability provide a framework for real-time cascading failure risk
analysis, which are based on single timescales and individual evaluation.
stability issues, can no longer adapt to the increasingly
complex transient behaviors of modern power systems. The rest of the paper is organized as follows: in section II,
based on the existing evaluation metrics for individual stability
Therefore, it is imperative to investigate the transient response
issue, a normalized metric for transient stability is proposed.
of modern power systems and identify critical events/moments
Section III firstly introduces criteria for identifying dominant
for further transient stability analysis.
stability issue, and then a scheme for critical events/moments
In this paper, critical events refer to the faults or stability identification is elaborated. In section IV, a model of the
controls that lead to the change of dominant stability issue or hybrid AC-DC power system with integrated renewable
the loss of stability, such as fault, grid-disconnection of REG, energy is constructed, and simulation based on the interaction
between MATLAB and BPA is used in this study to generate
This work is funded by the Science and Technology Project of SGCC fault scenarios. In section V, the proposed method for critical
(No.5108-202218280A-2-266-XG).
events/moments identification is applied in the improved IEEE

979-8-3503-7513-8/24/$31.00 ©2024 IEEE 681

39-bus system for critical events/moments identification and 1 m
(4)
transient stability analysis. f COI =
H∑
∑H
i =1
i fi

II. EVALUATION METRICS FOR POWER SYSTEMS TRANSIENT where H i is the inertia constant of generator i , m is the total
STABILITY number of generators, f i is the frequency of generator i .

A. Evaluation Metrics for Different Stability Issues The average frequency of the power system can be
calculated by formula (5).
1) Metric for rotor angle transient stability issue
The transient rotor angle issue primarily manifests when 1 N

the power system is subjected to a severe disturbance, causing

f avg =
N
∑f
i =1
i
(5)
oscillations of the rotor angles among generators. As
oscillations continue, the relative difference in rotor angles where N is the total number of buses in the power system, f i
increases, ultimately leading to instability. Therefore, transient is the frequency of bus i .
stability of rotor angle can be assessed by determining whether
the difference in rotor angles between any two synchronous 3) Metric for voltage transient stability issue
generators at any given time remains below a specified The voltage stability refers to the ability of a power system
threshold, satisfying the inequality constraint (1). to maintain steady voltages at all buses in the system after
being subjected to a disturbance from a given initial operating
δ i (tk ) − δ j (tk ) ≤ δ set i, j ∈ N Gen , tk > tc (1) condition. Instability that may result occurs in the form of a
progressive fall or rise of voltages of some buses.
where N Gen denotes the set of synchronous generators in the
Transient stability of voltage can be assessed based on the
system, tc is the time that fault is cleared, δ set is the instability drop of bus voltage and the recovery time after the disturbance.
rotor angle difference threshold, which can be set to 180° . The Transient Voltage Stability Index (TVSI) is proposed, and
The formula indicates that, from the moment a fault is cleared, can be calculated by formula (6).
if the absolute value of the power angle difference between
1 , Vmin > Vcr ∩ TS = 0
any two generators in the system does not exceed δ set , the  (6)
VTVSI 1 , Vmin ≤ Vcr ∩ 0 < TS < Tcr
=
power system is transient rotor angle stable. −1 , V < V ∩ T ≥ T
 min cr S cr
However, the transient angle stability assessment based on
formula (1) can only qualitatively determine whether the where Vmin is the minimum voltage of all buses, Vcr is the low
system is stable or not. It cannot provide a quantitative voltage threshold, TS is the time when Vmin is lower than the
evaluation. Therefore, the Transient Angle Stability Index
set threshold, Tcr is the permissible limit time for Vmin
(TASI) is proposed, which allows for the quantitative
assessment of the transient rotor angle stability in the power remaining below Vcr . VTVSI = 1 signifies transient voltage stable,
system. As shown in formula (2). while VTVSI = −1 signifies transient voltage unstable.
360° − ∆δ max
=ATASI ≥ ηset (2) B. Normalized Metrics for Transient Stability Evaluation
360° + ∆δ max
According to evaluation metrics for individual stability
where ∆δ max is the maximum rotor angle difference between issue mentioned above, normalized metrics for power system
any two generators in the transient process, and ηset is the set transient stability evaluation are carried out, as follows:
threshold. When the inequality is satisfied, it indicates that the Firstly, for each stability issue, a metric is selected. These
system is transient rotor angle stable. Otherwise, it indicates metrics are able to reflect the system transient stability
that the system has lost angle stability. Furthermore, the closer intuitively in real-time.
ATASI is to 1, the greater the stability margin of the system.
Secondly, two thresholds are set for each metric. One
2) Metrics for frequency transient stability issue indicates the occurrence of the stability issue, the other
The frequency stability issue primarily manifests as indicates the loss of stability, as shown in TABLE I.
sustained frequency swings leading to tripping of generating
TABLE I. METRICS AND THRESHOLDS FOR EACH STABILITY ISSUE
units or loads. In severe cases, may cause large excursions of
frequency, power flows, voltage and other variables. Transient Stability Issue Metric
Occurrence of Loss of
frequency of the system can be assessed by the Center of Stability Issues Stability
Inertia (COI) frequency [10] or the average frequency of the Maximum rotor angle
Rotor angle difference, ∆δ max 180° 270°
power system.
Average frequency
The COI frequency can be calculated by formula (3)-(4). Frequency deviation, ∆f avg 0.2Hz 0.5Hz
m Maximum voltage
H ∑ = ∑ Hi (3) Voltage deviation, ∆U max
0.1p.u. 0.3p.u.
i =1
 Thirdly, metrics of Instability Degree for angle, frequency Start
and voltage are defined, and can be calculated and normalized
Calculate Instability Degree Sa,Sf,SU,S by formula(7)-(10)
as shown in formula (7)-(9).
 0 , ∆δ max ≤ 180°
S = 0? No
 ∆δ − 180°
=Sa 
 max
,180° < ∆δ max < 270°
(7)
Yes
 270° − 180° S = 1? No
 1 , ∆δ max ≥ 270° System stable
Yes S = Sa? No
 0 , ∆f avg ≤ 0.2Hz
 ∆f − 0.2Hz System
 avg (8) End
unstable
Yes S = Sf? No
=Sf  , 0.2Hz < ∆f avg < 0.5Hz
 0.5Hz − 0.2Hz Rotor angle Voltage
Yes
 1 , ∆f avg ≥ 0.5Hz issue issue
Sa = 1? No Frequency
 0 , ∆U max ≤ 0.1 issue
 ∆U − 0.1 Yes Sf = 1? No
=

SU  max
, 0.1 < ∆U max < 0.3
(9)
Rotor angle Voltage End
 0.3 − 0.1 unstable
Yes
unstable
 1 , ∆U max ≥ 0.3 Frequency
unstable
In formula (7)-(9), metrics of Instability Degree S a , S f
End
and SU range from 0 to 1. A value of 0 indicates that stability
Figure 1. Flowchart for critical events/moments identification
issue does not exhibit, while a value of 1 indicates transient
instability. And the closer it is to 1, the more severe the
stability issue is. IV. SIMULATION

III. IDENTIFICATION SCHEME OF CRITICAL EVENTS/MOMENTS A. Modeling of Hybrid AC-DC Power Systems with
Integrated Renewable Energy
A. Criteria for the Dominant Stability Issue Identification In this paper, a model of hybrid AC-DC power systems
In modern power systems, stability issues do not appear in with integrated renewable energy is developed based on PSD-
isolation. When a fault occurs, the transient process can often BPA. Several wind farms and an HVDC transmission line are
involve a transition from one type of stability issue to another. incorporated into the original IEEE 39-bus test system. In
Therefore, it is necessary to comprehensively consider order to add an HVDC transmission line, efforts have been
different stability at the same time and identify the dominant made to establish a sending end and a receiving end within the
issue. power system. The power output of several generators and the
power load of several buses have been adjusted accordingly.
The identification of the dominant stability issue is based
on metrics of instability degree proposed in II.B. The Ultimately, a hybrid AC-DC system with integrated
Combined Instability Degree can be obtained by formula (10). renewable energy is constructed. The parameters of wind
farms and the HVDC transmission system are shown in
S = max{S a , S f , SU } (10) TABLE II. and TABLE III. The improved IEEE 39-bus
system is illustrated in Figure 2.
If S = 0 , the system is stable; if S = 1 , the system is
judged as unstable, and the stability issue corresponding to the TABLE II. PARAMETERS OF WIND FARMS
Instability Degree that reaches 1 is identified as the dominant
instability mode. If 0 < S < 1 , the stability issue corresponding Original Bus Name New Bus Name Rated Power
to the Instability Degree that is closest to 1 among three bus-35 Wind Farm 1 (WF1) 750MW
metrics is identified as the dominant stability issue, and S
bus-36 Wind Farm 2 (WF2) 560MW
indicates the transient instability degree for the power system.

B. Method for Critical Events/Moments Identification TABLE III. PARAMETERS OF HVDC TRANSMISSION LINE

The definition of critical is mentioned in section I. The Rectifier Station Inverter Station Rated Voltage Rated Power
method for critical events/moments is show in Figure 1.
bus-22 bus-3 ±400kV 500MW

B. Simulation Based on MATLAB and BPA Interaction

In this study, a batch simulation method based on the
interaction between MATLAB and BPA is employed,
enabling the generation and simulation of fault scenarios,
extraction and processing of simulation results, stability
analysis, and output of the analysis results. The flowchart for
the iteration process between MATLAB and BPA is shown in V. CRITICAL MOMENTS/EVENTS IDENTIFICATION AND
Figure 3. STABILITY ANALYSIS IN TYPICAL SCENARIOS
G G Receiving End
The MATLAB and BPA interactive simulation method
30 37 introduced in IV.B is used in this paper, and a total of 102
25 26 28 29 fault scenarios for three-phase short-circuit faults, single-phase
27
transient faults, and single-phase permanent faults at the
2
38
midpoints of all lines in the system are generated. The critical
1 events/moments identification method proposed in III.B is
3 18 17 G
G then applied, and is used for transient stability analysis. Here
39 16 21 are two typical scenarios.
15
DC Line
A. Scenario 1: Initial Fault Leading to the Grid-
WF
4 14
24 Disconnection of REG
WF2
Among the 102 generated fault scenarios, 28 involve wind
5 13 23 turbines grid-disconnection events. These scenarios exhibit
9 6
12 19 similar transient processes. Here is an example for analysis,
20
with the detailed initial fault information in TABLE IV. The
7 11 22
10 critical events/moments list and curve are shown in TABLE V.
and Figure 4. respectively.
8 31 32 34 33 WF1
G G G G WF
TABLE IV. INFORMATION OF INITIAL FAULT IN SCENARIO 1
Sending End
Fault Occurrence Fault
Figure 2. The improved IEEE 39-bus system Type of Fault Location
Time Duration
Three-phase Line between bus-9
200ms 100ms
Start short-circuit and bus-39

TABLE V. LIST OF CRITICAL EVENTS/MOMENTS AND ANALYSIS

Construct power system RESULT IN SCENARIO 1
models in BPA
Major Transient
Time(s) Critical Events
Stability Issue
0.20 Fault occurred -
Yes No
Operation modes traversal Wind Farm 1,2 low voltage ride
0.22 -
completed through
0.30 Fault cleared -
Modify DAT files
wind turbines in Wind Farm 1,2
in MATLAB 0.30~3.30 Voltage
grid-disconnected at 2.20s.
3.30~4.22 - Frequency
4.22~6.00 - Frequency unstable
Read DAT files, perform
power flow calculation,
generate BSE files
180

135

90

45

Yes 0

Fault scenarios traversal 0 0.3 1 2 2.2 3 3.3 4 4.2 5 6

completed
No
1

0.75

Modify SWI files in 0.5

MATLAB 0.25

0 0.3 1 2 2.2 3 3.3 4 4.2 5 6

Read BSE and SWI files, 51

perform transient simulation, 50

generate SWX files 49

48

0 0.3 1 2 2.2 3 3.3 4 4.2 5 6

PSD-BPA
Read SWX files, analyze
transient stability

Figure 4. Monitoring Curve of System in scenario 1

End
MATLAB
In Figure 4. , FCT means Fault Cleared Time, and CM
Figure 3. Simulation flowchart means Critical Moment. Both wind turbines in Wind Farm 1
and 2 entered Low Voltage Ride Through (LVRT) due to the
sustained grid-connected voltage drop under 0.9p.u. . By 2.20s, to the occurrence of DC blocking during the fault. After the
the duration of LVRT exceeded the limit for wind turbines, fault was cleared, the system remained stable for a short
leading to the grid-disconnection of wind turbines. In this period of time. However, due to the blocking of the DC
scenario, transient voltage and frequency stability issues were transmission line, an issue of power imbalance between the
the primary concerns. Initially, the occurrence of initial fault sending and receiving ends occurred and became apparent,
caused the sustained voltage drop, resulting in transient leading to a continuous increase of rotor angle difference of
voltage issue. Subsequently, the voltage issue caused the grid- synchronous generators between the sending and receiving
disconnection of wind turbines, leading to power loss of the ends, and causing sustained voltage drop of several buses.
system and a continuous decline in frequency. Gradually, the Ultimately, the synchronous generators within the system lost
transient stability issue shifted from voltage issue to more synchronization, and system became transient unstable.
severe frequency issues. Ultimately, the system experienced
transient frequency instability. VI. CONCLUSION
This paper proposes a scheme for identifying critical
B. Scenario 2: Initial Fault Occurred Near the DC events and moments, enabling a comprehensive evaluation of
Subsystem power system transient stability. Additionally, an approach for
If the initial fault occurs near the DC subsystem, batch simulation of complex transient scenarios based on the
commutation failure and DC blocking are likely to happen. interaction between MATLAB and BPA is presented.
Here is an example for analysis, with the detailed initial fault Subsequently, simulations of transient scenarios are conducted
information in TABLE VI. The critical events/moments list on an improved IEEE 39-bus system. Finally, critical
and curve are shown in TABLE VII. and Figure 5. moments/events of two typical scenarios are identified. And
respectively. the effectiveness of the proposed method is validated.

TABLE VI. INFORMATION OF INITIAL FAULT IN SCENARIO 2 REFERENCES

Fault Occurrence Fault [1] P. Kundur et al., "Definition and classification of power system stability
Type of Fault Location
Time Duration IEEE/CIGRE joint task force on stability terms and definitions," in IEEE
Three-phase Line between Transactions on Power Systems, vol. 19, no. 3, pp. 1387-1401, Aug.
200ms 100ms
short-circuit bus-2 and bus-25 2004.
[2] Y. Zhou et al., "Transient Rotor Angle and Voltage Stability
TABLE VII. LIST OF CRITICAL EVENTS/MOMENTS AND ANALYSIS Discrimination Based on Deep Convolutional Neural Network with
RESULT IN SCENARIO 2 Multiple Inputs," 2021 IEEE 4th International Electrical and Energy
Conference (CIEEC), Wuhan, China, 2021, pp. 1-6.
Major Transient
Time(s) Critical Events [3] C. Qin, Y. Jia, Z. Li, G. Li, L. Li and Y. Zhu, "Identification of Transient
Stability Issue
Voltage Stability Weakness in Power Systems Based on Grad-CAM,"
0.20 Fault occurred - 2022 7th International Conference on Power and Renewable Energy
0.23 DC blocking - (ICPRE), Shanghai, China, 2022, pp. 53-57.
0.30 Fault cleared -
[4] Y. Chen, H. Xie, L. Zhang and N. Zhou, "Research on Data-Driven
0.30~2.51 - Stable
Transient Voltage Stability Assessment Considering Uncertainty," 2022
2.51~4.25 - Rotor angle and voltage
IEEE International Conference on Power Systems Technology
Rotor angle, voltage (POWERCON), Kuala Lumpur, Malaysia, 2022, pp. 1-6.
4.25~6.00 -
and frequency unstable
[5] S. Yang, X. Fan, B. Zhang, Z. Hao and H. Masahide, "A Unified
Scheme for Power System Transient Stability Simulation and Transient
Stability Assessment," 2019 IEEE 8th International Conference on
Advanced Power System Automation and Protection (APAP), Xi'an,
270

China, 2019, pp. 1682-1686.

180

90

0 [6] W. Ju, K. Sun and R. Yao, "Simulation of Cascading Outages Using a

Power-Flow Model Considering Frequency," in IEEE Access, vol. 6, pp.
0 0.3 1 2 2.51 3 4 4.25 5 6

37784-37795, 2018.
1 [7] J. Song, E. Cotilla-Sanchez, G. Ghanavati and P. D. H. Hines, "Dynamic
0.75
Modeling of Cascading Failure in Power Systems," in IEEE
Transactions on Power Systems, vol. 31, no. 3, pp. 2085-2095, May
0.5

0.25

0 0.3 1 2 2.51 3 4 4.25 5 6

2016.
[8] R. Yao, S. Huang, K. Sun, F. Liu, X. Zhang and S. Mei, "A Multi-
51
Timescale Quasi-Dynamic Model for Simulation of Cascading Outages,"
50
in IEEE Transactions on Power Systems, vol. 31, no. 4, pp. 3189-3201,
49
July 2016.
48 [9] Y. Zhu, Y. Zhou, W. Wei and L. Zhang, "Real-Time Cascading Failure
Risk Evaluation With High Penetration of Renewable Energy Based on a
0 0.3 1 2 2.51 3 4 4.25 5 6

Graph Convolutional Network," in IEEE Transactions on Power

Systems, vol. 38, no. 5, pp. 4122-4133, Sept. 2023.
[10] X. Ye, J. D. Morales and J. V. Milanović, "Truncated Transient Stability
Figure 5. Monitoring Curve of System in scenario 2 Index for On-line Power System Transient Stability Assessment," 2021
IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe),
In Figure 5. , FCT means Fault Cleared Time, and CM Espoo, Finland, 2021, pp. 1-5.
means Critical Moment. Consecutive commutation failures led