High-precision Transient Stability Assessment Using Convolution Neural Network

Convolutional neural network-based power system transient stability assessment and instability mode prediction

Fast and accurate fault location detection methods are crucial for power system maintenance crews to isolate permanent faults, which in turn helps to minimize downtime and improve system reliability. Therefore, it is vital from the perspective of power system reliability planners to develop an approach that can quickly detect and locate faults. This paper proposes an approach to locate faults in transmission networks based on machine learning using high-precision data gathered through Real-time Digital Simulators (RTDS). The data-driven machine learning algorithms are provided transient data collected during the fault to reliably predict fault locations, via temporal voltage phasors captured through Phasor Measurement Units (PMUs). A Python automated real-time simulation is conducted to simulate hundreds of three-phase fault scenarios considering different loading levels, fault locations, durations, and resistances. To facilitate the data gathering process, a lab-scale data acquisition and storage framework is developed with the integration of RTDS, virtual Phasor Data Concentrator (OpenPDC), and MySQL. The proposed approach is demonstrated through a comprehensive study on the WECC 9-Bus System. A comparative study among different machine learning algorithms is conducted to determine the superior algorithm for fault location detection.

Spatial-temporal adaptive transient stability assessment for power system under missing data

There have been quite a few attempts to compute synchronous generator parameters based on voltage and current synchrophasors taken under power system transients. However, we have not seen any publications with thorough analysis as to how soon the phasor measurement unit reacts to disturbance conditions, which components of the transient are filtered out and which are passed through, as well as what the total vector error is. The goal of this research is to determine all of these characteristics of a phasor measurement unit when playing back transient oscillograms for a stator short circuit obtained through mathematical modeling. The transient oscillograms have been derived via both a full Park-Gorev system of flux linkage equations as well as the MATLAB/Simulink Synchronous Machine block. Physical modeling was then conducted via a real-time digital simulator (RTDS) along with a dedicated phasor measurement unit ENIP-2 (PMU), and the stator current phasors were recorded. Our analysis has shown that both RTDS and ENIP-2 (PMU) almost entirely filter out the exponentially decaying DC component of the fault current while closely following the periodical signal envelope. The total vector error has been estimated to become below 1–2 % after around 0,02–0,03 s into the fault when selecting the “P” class filters according to IEEE C37.118. We have come to a conclusion that synchrophasor measurements under power system disturbances could be utilized for estimating the synchronous, transient, and subtransient generator parameters. The selected synchronous machine model in the form of flux linkage equations is correct, as the obtained transient oscillograms are exactly the same as those produced by Simulink. “P” class phasor measurements can be recommended for representing transients in the stator circuit of a synchronous generator. The results of this investigation are meant to be employed for synchronous machine parameter estimation based on phasors sourced from RTDS and, hopefully, from phasor measurement units installed at power plants.

Transient stability assessment (TSA) plays an essential role in the safe and stable operation of power system. The traditional time-domain simulation method and direct method have great limitation in modeling and computing, which can not meet the need of fast and accurate assessment of power system stability. Therefore, combined physics model with data-driven algorithm, this paper proposes a physics-informed deep learning paradigm to solve this problem. In the data-driven algorithm, recurrent neural networks are trained to predict the rotor speed and power angle with the guiding of the power system’s physical information. Finally, the simulation on IEEE14 system is carried out by using PSS/E software, which proved that the proposed method can predict the power angle trajectory quickly and accurately.

Deep learning-based transient stability assessment framework for large-scale modern power system

The subject of the article is the technique for determination of the automatic excitation controller (AEC) parameters based on measurements from phasor measurement units (PMU). This paper shows the possibility of using PMU measurements for determining the synchronous generator AEC parameters. The article considers the type of the AEC which is ARV-MT of the Russian production. By using the simulation in the hardware and software system RTDS (Real-Time Digital Simulator), PMU measurements have been obtained for different modes of the AEC operation. Processing of the results has been carried out in MATLAB. The developed technique is based on the processing of obtained measurements by optimization algorithm. As a result of the comparative analysis, the most suitable type of optimization has been chosen. The possibility of PMU measurements use for determining the AEC parameters in different modes of its operation is shown. Also, the results of application of the technique for cases of a different power level noise in the circuit is investigated. The developed technique determines the AEC parameters with a sufficient accuracy by using PMU measurements. Thus, the technique allows one to extend the field of PMU application, and, being further developed, will allow one to monitor the AEC and calculate its parameters in real time.

Several major blackouts in the past have shown the need for advanced wide‐area monitoring and control (WAMC) techniques. Advanced synchrophasor visualization and monitoring can be realized with time‐referenced power system data that are collected by phasor measurement units (PMUs) using synchronized clocks. Presently, PMUs are being deployed around the world at a rapid rate for various power system applications. When a necessary number of PMUs are installed at optimal locations throughout the network, the complete state of the system can be observed. PMUs take real‐time measurements to determine the state of the power system and can be used to enhance state estimation efficiency. With PMUs in place, there are many control and protection schemes that can be implemented successfully to take preventive and corrective actions. These remedial action schemes (RASs) are classified by the IEEE Power System Relaying Committee (PSRC) as system integrity protection schemes (SIPSs). This work involved the development of a real‐time hardware test bed in order to analyze the transient stability of a simulated power system by using synchrophasors to visualize system stress across a transmission line with and without load‐shedding schemes. The real time digital simulator (RTDS®) was used to model the power system in real time. PMUs, a satellite‐synchronized clock, and a synchrophasor vector processor (SVP) were used to test the synchrophasor application. Copyright © 2010 John Wiley & Sons, Ltd.

Transient stability assessment (TSA) based on machine learning techniques is a classical problem of classification. In this paper, a new transient stability assessment model using the weighted random forest (WRF) is proposed. Random forest, which runs fast and can scale to very high dimensional datasets, is one of the most popular methods applied to TSA after faults occurred. However, in the actual operation of modern power system, unstable samples constitute only a very small minority of the cases. Besides, the traditional random forest algorithm does not work very well on imbalanced data. To tackle the problem, weighted random forest is adopted to lean the interest towards to the correct classification of rare class, and improve the prediction performance. Case study on the New England 39-bus test power system exhibits that WRF has better performance than the original random forest.

In this paper, an online power system transient stability assessment (TSA) problem is mapped as a two-class classification problem and a novel data mining algorithm the core vector machine (CVM) is proposed to solve the problem based on phasor measurement units (PMUs) big data. First of all, an offline training, online application framework is proposed, which contained four sub-steps, namely features selection, offline training, online application, and assessment evaluation. First, 24 features are selected to present the system status. Then in the offline training procedure, the PMU big data is generated by time domain simulation, and a CVM model is trained and tested. In the online application procedure, an interface between PMU data center and feature calculation program is set up to collect real time specific PMU big data and the CVM trained is applied to the TSA problem. Last but not least, the evaluation indices are calculated. Compared with other support vector machines, the proposed CVM based assessment algorithm has the higher precision, meanwhile, it has the least time consumption and space complexity. As long as online PMU big data are received, TSA can be done simultaneously. Case studies on the IEEE New England 39-bus system, and real systems in China and the U.S., exhibit the speed and effectiveness of the proposed algorithm.

Modern power systems are large in size and complex in features; the data collected by Phasor Measurement Units (PMUs) are often noisy and contaminated; and the machine learning models that have been applied to the transient stability assessment (TSA) of power systems are not sufficiently capable of capturing long-distance dependencies. All these issues make it difficult for data mining-based power system TSA methods to have sufficient accuracy, timeliness, and robustness. To solve this problem, this paper proposes a power system TSA model based on the transformer and neighborhood rough set. The model first uses the neighborhood rough set to deal with the redundant features of the power system trend data and then uses the transformer model to train the TSA model, in which various normalization methods such as Batch Normalization and Layer Normalization are introduced in the process to obtain better evaluation performance and speed up the convergence rate of the model. Finally, the model is evaluated by two evaluation indicators, F1−measure and accuracy, with values of 99.61% for accuracy and 0.9972 for F1−measure, as soon as the tests on noise contamination and missing data test results on the IEEE39 system show that the NRS-Transformer model proposed in this paper is superior in terms of prediction accuracy, training speed, and robustness.

Transient stability assessment is the key point to ensure the safe and stable operation of the power system. Based on the data collected from the phasor measurement unit (PMU), a transient stability assessment method combining stacked automatic encoder (SAE) and support vector machine (SVM) is proposed. Multi-layer abstract learning is performed on the original features by the SAE, and the extracted feature is used to train the SVM and test. Simulation on the New England 39-bus test system show that the proposed model has high accuracy.

Transient stability assessment (TSA) is a cornerstone for resilient operations of today’s interconnected power grids. This paper is a confluence of quantum computing, data science and machine learning to potentially address the power system TSA issue. We devise a quantum TSA (QTSA) method to enable efficient data-driven transient stability prediction for bulk power systems, which is the first attempt to tackle the TSA issue with quantum computing. Our contributions are three-fold: 1) A high expressibility, low-depth quantum circuit (HELD) is designed for accurate and noise-resilient TSA; 2) A quantum natural gradient descent algorithm is developed for efficient HELD training and 3) A systematical analysis on QTSA’s performance under various quantum factors is performed. QTSA underpins a foundation of quantum machine learning-enabled power grid stability analytics. It renders the intractable TSA straightforward and effortless in the Hilbert space, and therefore provides stability information for power system operations. Extensive experiments on quantum simulators and real quantum computers verify the accuracy, noise-resilience, scalability and universality of QTSA.

Missing data caused by phasor measurement unit (PMU) faults and communication errors leads to power system unobservability, making transient stability assessment (TSA) challenging. To address this issue, a source-free domain adaptation method based on embedding space recovery is presented for TSA. First, an end-to-end multi-task learning framework is proposed, which ensures the integrity of transient features and the distinguishability of system states by sharing spatio-temporal features to recover missing feature spaces and align transient features across domains. Next, a spatio-temporal variational autoencoder (STVAE) is introduced, which avoids the accumulation of imputation errors by recovering the feature space of missing data, using spatial modeling from the power system topology and graph convolutional networks to guide temporal feature recovery. Additionally, an evaluation metric approximation mechanism is proposed to address sample imbalance by converting confusion matrix-based metrics into differentiable loss functions, allowing for direct optimization and improved model performance. This method enables the stability assessment of power systems with missing data, ensuring both rapid assessment and data privacy. Finally, simulation results using the 39-bus and 145-bus systems validate the effectiveness of the proposed method.

The use of phasor measurement units (PMU) has enabled monitoring the power system dynamics online. There are many PMUs installed, which will give a comprehensive view in real or near real time of the entire system. In general, there are many operating conditions of power systems, which requires different tools to analyze the synchrophasor data provided by PMUs. Changes in power system monitoring must be detected as quickly as possible, so that the system operators can choose a different tool for data analysis. Therefore, in this paper, changes in the working conditions of the power system are detected using a subspace state-space based change-point detection method. The detection algorithm proposed in this paper is sequential detection, which aims at both detection accuracy and detection speed. We have tested our proposed algorithm to detect many various kinds of changes in power system operations and verified that the algorithm is able to detect changes successfully, including very small ones in the system such as the low-amplitude forced oscillations buried in the system's ambient conditions in a real-time manner. The algorithm is tested in a set of real-world data to show its practical usage in large systems with imperfect data.