Search

Studies on transient stability constrained optimal power flow (TSCOPF) have become crucial for power systems to guarantee their dynamic securities against credible contingencies, while their optimum operations are to be continuously projected under changing conditions. However, the current approach to the TSCOPF problem is not sufficient to meet the expectations of a modern power system because it suffers from uncertainties mainly due to the rapid and large integration of distributed energy sources. This study proposes a novel method using the information-gap decision theory (IGDT) technique to solve the TSCOPF problem in the presence of uncertainties due to the penetration of wind farms. The IGDT is a non-probabilistic decision-making method that can be easily implemented to handle uncertainty in optimisation problems. While presenting applicable strategies, it does not require any information about the historical data, probability density function or membership function of the uncertain parameters. The proposed method offers an analysis for the economic dispatch in a power system with wind energy resources while providing robustness against transient instabilities and uncertainties in power generation. To demonstrate the effectiveness of the proposed method, it is implemented on New England 39-bus and IEEE 118-bus test systems.

In this paper, a new real-time defensive islanding method, which is adaptive to the operating conditions is proposed. In the method, a number of candidate islanding schemes are generated using both model- and measurement-based islanding algorithms after detecting a severe fault in the system by means of a new severity index based on generator bus voltage frequency measurements. Of model-based algorithms, slow coherency-based islanding, in which the prefault measurements are utilized, is adopted. On the other hand, ${K}$ -means, hierarchical, and fuzzy relational eigenvector centrality-based clustering are employed as measurement-based islanding algorithms, where the postfault measurements of the evolving dynamics after the severe fault are utilized. A faster-than-real-time software platform is, then, employed to validate the success of the candidate schemes in healing the system. Among the successful schemes, the one resulting in the least load imbalance is chosen to be applied. All the computations from the detection of the fault to the application of islanding are performed in real-time, directly after the occurrence of the fault. The proposed method is demonstrated on the 37-generator 127-bus WSCC power test system, and on a model of the Turkish power system to assess the method’s performance.

Post-fault prediction of transient stability of power systems has a great impact on the performance of wide area monitoring, protection and control systems. Situational awareness capabilities of a power system are improved by fast detection of instabilities after severe fault occurrences. This allows sufficient time to take necessary corrective control actions. In this paper, a novel method based on stacked sparse autoencoder is proposed to predict the post-fault transient stability status of the power system directly after clearing the fault. A dataset is generated off-line to train a stacked sparse autoencoder, and then the trained stacked sparse autoencoder is used in an online application of predicting any transient instability. The stacked sparse autoencoder is fed by the inputs, which are specific points extracted from the fault-on voltage magnitude measurements collected from the phasor measurement units. The effectiveness of the proposed method is demonstrated and compared with the conventional approaches that adopt multilayer perceptrons or post-fault measurements as it is applied to the 127-bus WSCC test system and to the Turkish power system.

One of the main problems in the utilization of machine learning-based classifiers for transient stability prediction is their long training times with the comprehensive large-sized datasets. Using a small-sized dataset to decrease the training time is not reasonable since the dataset should be representative of all types of severe faults. In this paper, a novel methodology based on transfer learning is proposed for real-time post-contingency transient stability prediction to overcome the difficulties about the long training times of these classifiers. In the proposed method, first, a small dataset which contains only the three-phase fault contingencies for various operating points is selected to train a convolutional neural network (CNN) classifier, and then, an additional dataset which involves two-phase-to-ground fault scenarios is used to update the trained CNN using the transfer learning approach instead of retraining the model from ground up. To demonstrate the efficiency of the proposed method, it is applied to the 127-bus test system.

Transient instabilities triggered by critical faults can lead to rapidly developing blackouts in power systems. With the data driven methods using synchrophasor measurements collected from PMUs, it is possible to predict the instabilities just after the clearance of a critical fault. In this study, the performance of a classifier to be used for an early prediction of transient instability is enhanced by modifying the loss function used during the offline training phase. This study proposes to assign weights to the terms in the binary cross entropy loss function associated with each class, as misclassification of different classes generally causes different costs to the utility and its customers. The proposed method is able to determine the optimum values of the weights according to a pre-selected tolerance value, which represents how far the accuracy is away from being acceptable. The efficacy of the proposed method is examined on the 127-bus WSCC test system.

Real-time transient stability assessment (TSA) is an important task which ensures the stability, and therefore, enhances the reliability of power systems. Various types of learning-based methodologies in which machine learning and deep learning algorithms are adopted for real-time TSA exist in literature. Convolutional neural network (CNN) is a deep-learning-based method which mostly demonstrates high performance for image classification. However, employing the conventional structure of CNN classifier for time series data may result in high computational complexity or low prediction accuracy. In this paper, a novel methodology is proposed for real-time stability prediction of a power system in which voltage angle measurements obtained from PMUs are utilized to train a multichannel deep CNN (MC-DCNN), which is a modified version of CNN classifier and appropriate for multivariate time series data. To evaluate the performance of the proposed method for real-time transient stability prediction, it is applied to the 127-bus WSCC test system.

Recently, technological developments in PMUs have led to a widespread deployment of PMUs in electrical power systems to carry out many online applications, thanks to their high reporting rates and synchronized measurements. Since it may not be feasible to deploy a PMU on every bus in the system, a selected number of buses can be chosen for the deployment of PMUs to achieve a certain task. In this paper, a new PMU placement method based on feature selection algorithms for obtaining a high accuracy in early predicting of transient instabilities is proposed. The optimal PMU locations are selected according to two different feature selection methods, which are filtering based and feature importance based algorithms. The feature selection methods are implemented with the use of a classifier based on LightGBM. The proposed method is demonstrated on the 127-bus Western Systems Coordinating Council test system.

Detection of evolving transient instabilities in power systems is of high importance in order to maintain the system’s security and integrity. With the recent developments on wide area monitoring systems, employing machine learning models for transient stability assessment has drawn a great attention. Nevertheless, as the power systems are being designed and operated in a secure and robust manner, the ratio of the contingencies that lead to transient instability to the ones that do not is relatively low. This makes the learning problem harder for such systems as the training data would be inherently imbalanced. In this work, we exploit the resampling techniques to tackle the imbalanced learning problem by utilizing three different over-sampling methods: random over-sampling, SMOTE and ADASYN. The XGBoost classifier model is adopted within the proposed framework to compare the performance improvements through each over-sampling method. The results obtained in the Nordic power system show notable improvements, especially when unequal misclassification costs are considered.

Online monitoring of the power system is a vital application for enhancing the situational awareness capabilities of the system. Rapid integration of phasor measurement units in the network enables transmission system operators to analyze the events in real time due to their high reporting rates. Real-time detection and classification of the fault related events as no-fault, fault-incidence, fault-on and post-fault stage with no further disturbance, is an important requirement in order to decide on the control actions to protect the system from any instability. In this paper, a sliding window based continuous online monitoring method of the power system using convolutional neural networks is proposed. The effectiveness of the proposed method is validated on the 127-bus Western Systems Coordinating Council test system.

A fast and accurate transient stability assessment is essential for maintaining the reliability and integrity of the power system where quick corrective control actions are to be taken to prevent cascading outages. In this work, two separate classifiers based on long short-term memory networks are proposed to predict the stability status of a power system after being subjected to a credible fault. These classifiers use the immediate post-fault measurements of bus voltage magnitudes and rate of change of frequencies obtained from phasor measurement units during the first few cycles of the post-fault period. The proposed method does not assume that every bus is equipped with a PMU. It is found out that, even with only 30% of buses equipped with PMUs, the proposed classifiers are able to predict the stability status with an accuracy above 97%. Robustness of the classifiers is also investigated under noisy conditions and against the scenario of missing PMU measurements. A comparison between the two proposed classifiers is conducted by means of a t-test.

In this study, a novel methodology based on signal processing and machine learning approaches is proposed for real-time transient stability prediction (TSP) in power systems where the signals obtained from PMUs are utilized. The proposed method for TSP takes the computed energy of PMU signals in a window of measurements, as an input to a classifier to predict the stability of the system. Several types of classifiers, which are multi-layered perceptrons (MLPs), decision trees, and naive Bayes classifiers, are employed. Two alternative approaches of choosing the window of measurements used for TSP are proposed, where an MLP-based fault detection process is also proposed to form the proper window of measurements. One approach is to use a fixed window of only post-fault measurements, whereas the other approach is to use an expanding window of measurements covering pre-fault, fault-on and post-fault stages. Utilization of the energy concept in TSP gives the flexibility to process signals in different sizes while providing predictions robust to measurement noises and missing data. It also makes feature selection methods directly applicable, making the TSP possible with less PMUs. The proposed methods are applied to two different test systems and a large-scale model of the Turkish power system.

Inrecent years, many blackouts occurred in power systems of different parts ofthe world, affecting millions of people and causing great economic losses. Powersystem stability, which has a critical role in the design and operation ofelectrical power systems, maintains its importance today. Monitoring thestability status of a power system in real time is regarded as a primary taskin preventing system blackouts. This allows of a sufficient amount of time totake appropriate corrective control actions. In this study, the pre-faultvoltage magnitudes and angles taken from the phasor measurement units (PMU),clearing time of the fault and topology information of the transmission linethat has been tripped for clearing the fault are used to predict the transientinstabilities by two different methods based on the decision trees. The successand the effectiveness of the proposed machine learning models are shown as theyare applied to the 127-bus Western Systems Coordinating Council (WSCC) testsystem., Geçtiğimiz yıllarda, dünya çapında farklı güç sistemlerinde,çok sayıda geniş çaplı enerji kesintileri meydana gelmiş, bu kesintilermilyonlarca tüketicinin olumsuz etkilenmesine neden olmuş ve büyük miktardamali zararlara neden olmuştur. Elektrik güç sistemi tasarımı ve işletmesinde kritikbir role sahip sistem kararlılığı, günümüzdeki önemini korumaktadır. Bir güçsisteminin kararlılık durumunu gerçek zamanlı olarak izlemek, sistemçökmelerini önlemede birincil öneme sahip bir görev olarak kabul edilmektedir. Şebekeninkararlılık durumunun gerçek zamanlı olarak izlenmesi, geniş alan izleme, korumave kontrol sistemlerinin verimliliği açısından önemli bir fonksiyondur. Bufonksiyon ile düzeltici kontrol eylemlerinin zamanında gerçekleştirilebilmesisağlanabilir. Bu çalışmada, güç sisteminde meydana gelebilecek arızalaröncesinde fazör ölçüm birimlerinden alınan gerilimlere ait genlik ve açılarınyanı sıra, arızanın temizlenme süresi ve arızanın temizlenmesi için devredençıkarılan iletim hattına ait topoloji bilgileri de kullanılarak, geçici halkararsızlıklarının kestirimi, karar ağaçlarına dayalı iki farklı yöntem ilegerçekleştirilmiştir. Önerilen makine öğrenmesi modellerinin başarımları veetkinlikleri 29 jeneratörlü 127 baralı WSCC (Batı Eyaletleri KoordinasyonKurulu) test sisteminde uygulanarak gösterilmiştir.

In this paper, the performance degradation of a generating unit in a multi-area power system regulated by an area-wise decentralized load frequency control is detected and isolated via a scheme of unknown input observers. This scheme is robust to the changes in the load demand, which are considered as the unknown inputs. The actuator faults studied in this work are meant to be the performance degradations in generating units under load frequency control. Both stuck and non-stuck actuator faults are analyzed and the performances of the proposed method in both cases are tested through computer simulations on a two-area power system composed of five generating units. The results show that a performance degradation in any generating unit can be detected and isolated successfully provided that the faults in the sensors measuring generator outputs are absent.

This paper proposes a new method of optimization, the Big Bang–Big Crunch (BB–BC) method, for the optimization of preventive and corrective control actions to enhance the dynamic security of a power system against transient instabilities. The control actions, generation rescheduling and load shedding, are considered as constrained optimization problems with static and dynamic security constraints. These optimization problems are solved through the BB–BC method to minimize some operational costs related to the control actions. To reduce the size of the search spaces and the computational burden, decision trees and correlation coefficients are used as feature selection tools, which determine the most effective generators and loads for shaping the system’s transient stability. The proposed method is applied to a test system and compared with genetic algorithms, particle swarm optimization, differential evolution and active set method. The BB–BC method is promising since it gives comparable results with the other population based optimization methods.

Defensive islanding is one of the corrective control methods for securing the power system from severe cascading contingencies. Since the power system is operating in continuously changing operating conditions, the defensive islanding scheme must be adapting to these changes to arm the system against any potential instability. In this paper, we explore the effectiveness of a defensive islanding method to adapt to these changes by periodically updating the islanding scheme based on wide-area measurements. The proposed method uses the bus voltage angles measured by phasor measurement units to determine the islands via clustering algorithms. Among them, K-means clustering algorithm is adopted and explored as its performance during the course of the islanding scheme is studied. The proposed method is demonstrated on a 16-generator 68-bus power system and its performance is discussed.

Early prediction of the transient stability of power systems after fault occurrences has a great impact on the performance of wide area protection and control systems designed against transient instabilities. In this paper, an artificial neural networks based methodology is proposed for predicting the power system stability directly after clearing the fault. A dataset is generated to train a multilayer perceptron off-line, which is then used for early online prediction of any transient instability. The neural network is fed by the inputs, which are the pre-fault, during-fault, and post-fault voltage magnitude measurements collected from the phasor measurement units. The success and the effectiveness of the proposed method are demonstrated, as it is applied to the 37-generator 127-bus power test system and an accuracy above 99% is obtained in the early prediction of transient instabilities.

In this study, in order to reduce the abnormal voltage drop or rise in the distribution systems with distributed generators, we consider a multi-objective function to be optimized in such a way that the voltage profile is enhanced while the total active power loss in the network is minimized under some operational constraints. Capacitors and static VAR compensators are also coordinated with the distributed generators in the network to acquire a better enhancement in the voltage profile. The purpose of this study is to obtain the optimum sizes and locations of these system components so that the active power losses and the deviations in bus voltages from their nominal values are minimized. The problem is formulated as a multi-objective optimization problem and solved by the goal attainment method. The studies are demonstrated on an IEEE 34-bus test system where the direct load flow method is utilized for load flow calculations.

Among the power system corrective controls, defensive islanding is considered as the last resort to secure the system from severe cascading contingencies. The primary motive of defensive islanding is to limit the affected areas to maintain the stability of the resulting subsystems and to reduce the total loss of load in the system. The slow coherency based islanding can successfully be applied for the defensive islanding. In this paper, two partitioning methods are proposed, K-means clustering algorithm and fuzzy relational eigenvector centrality-based clustering algorithm. The proposed methods are using the data measured by phasor measurement units to determine the islands to be used in the defensive islanding. The proposed methods are demonstrated on the 16-generator 68-bus power system and their performances are discussed as their results are compared.

This paper proposes a new methodology to improve the population based optimization techniques applied for preventive control actions enhancing power system security. The preventive control studied includes both generation rescheduling and load curtailment. We first investigate how the size of the search space affects and improves the best solution obtained in the optimization process. Then, we develop a new methodology that involves a number of optimization algorithms running consecutively as the size of the search space of each algorithm is reduced according to the objective function. The extensive computational requirement for dynamic security assessment during the optimization processes is overcome by the application of neural networks. The methodology is successfully applied for solving the security constrained optimization problem of a 16-generator 68-bus test system with both continuous and discrete decision variables using consecutive differential evolution optimization algorithms.

In this paper, a preventive control method against transient instabilities due to a set of critical contingencies that could occur in a large power system is proposed. The generation rescheduling, adopted as the preventive control, is based on the coherency between the generators. The rescheduling is done by attempting to bring the generators’ rotor speeds equal after a three phase fault that might cause instability. The proposed methodology involves off-line simulations to determine the system response and to check the severity of each contingency. For each critical contingency, a scaling factor is assigned to scale the speed trajectory of the contingency. Active-set sequential quadratic programming (SQP) is used to optimize the scaling factors in such a way that the rescheduling method based on the scaling factors improves the dynamic security and restores the system’s stability for all contingencies that are taken into account.

In this paper, a new dynamic security assessment and generation rescheduling method utilizing genetic algorithms (GAs) which are integrated with probabilistic neural networks (PNNs) and adaptive neuro fuzzy inference systems (ANFISs) is proposed for the preventive control of large power systems against transient instabilities. By the proposed approach, PNNs are employed in a feasible manner to calculate the security regions accurately during the assessment and control. The security constrained generation rescheduling is implemented through a GA which optimizes the total fuel cost or the generation shifting during the preventive control. The steady-state solutions of the variables required for the GA are smoothly performed by the use of an ANFIS. The proposed methods are demonstrated on the 17-generator 163-bus Iowa power system and on the 50-generator 145-bus IEEE test system successfully and the effectiveness of the approaches is discussed.

Among the power system corrective controls, defensive islanding is considered as the last resort to secure the system from severe cascading contingencies. The primary motive of defensive islanding is to limit the affected areas as soon as possible to maintain the stability of the resulting subsystems and to reduce the total loss of load in the system. The design of the defensive islanding must address the questions of where and when to island the system to ensure minimum impact and the recovery of the affected portion of the system. In this paper, the critical islanding time is explored as a security index for applying the defensive islanding where the boundaries of the islands are decided by using the slow coherency concept between generators. The computations are performed on a test system to specify the maximum allowed time for operators to apply islanding before falling in blackout due to some critical contingencies.