Learning-Based Real-Time Event Identification Using Rich Real PMU Data

Yuan, Yuxuan; Guo, Yifei; Dehghanpour, Kaveh; Wang, Zhaoyu; Wang, Yanchao

doi:10.1109/tpwrs.2021.3081608

Cited by 28 publications

(14 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The increasing pressure around environment protection [81], the issue of renewable energy accommodation [82], and the need for better and faster event detection and identification [83] suggests the need to improve the existing metrics for critical load prioritization, as proposed in the research presented here.…”

Section: Resultsmentioning

confidence: 99%

Resilience Framework, Methods, and Metrics for the Prioritization of Critical Electrical Grid Customers

et al. 2022

View full text Add to dashboard Cite

One of the main objectives of new operating regimes, such as transactional power systems, is to make the power grid more resilient to catastrophes and disturbances, while at the same time making it possible to supply electricity to the largest number of customers. Although this is true, it is well known among power system operators that not all customers are the same. The consequence of this is that any discussion around the impact of transactional power systems on power system resilience should consider the needs of its critical users (such as control centers, fire stations, and hospitals) over other users. In assessing power system resilience, a metric is needed that gives “bonus points” to those systems that, under all circumstances, can continue to provide electricity to their critical users. In order to serve as a parameter in the assessment of power grid resilience, the research presented here discusses the proportion of critical loads existing in critical infrastructures. Once the critical loads are characterized, the next step is the inclusion of these loads in resilience metrics. This paper proposes resiliency metrics in which certain customers (those categorized as critical) are assigned a higher weight than others. One thing to keep in mind is the fact that there is no one-size-fits-all approach for all power systems, and that the assignment of such weights to customers can vary significantly from one operator to another based on their unique systems and the current and expected states of their critical customers.

show abstract

Section: Resultsmentioning

confidence: 99%

Resilience Framework, Methods, and Metrics for the Prioritization of Critical Electrical Grid Customers

et al. 2022

View full text Add to dashboard Cite

show abstract

“…• Approximating partial system constraints by penalty or prior • Crafting interpretable hidden features and latent space (that encode physical characteristics) [14] [15]…”

Section: B An Evaluation Methodologymentioning

confidence: 99%

“…3) Extracting meaningful features or crafting an interpretable latent space: Many works exist in this class. For example, [14] [15] learned the latent representation of sensor data in a graph to capture temporal dependency [14] or spatial sensor interactions [15]. [16] applied influence model to learn, for all edge pairs, the pairwise influence matrices which are then used for the prediction of line cascading outages.…”

Section: Related Workmentioning

confidence: 99%

“…• reducing the search space [1] [2] [3] • enforcing system constraints and technical limits [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] • crafting meaningful features and defining an interpretable latent space [14] [15] [16] [17] While these physics-informed ML methods address challenges with general ML methods, critical limitations still exist regarding scalability [18], generalization [19], interpretability [20], etc., especially in methods designed with insufficient knowledge of the physical system. Firstly, many of these methods do not consider the dynamic nature of grid conditions, in terms of network topology, loads, and generations, requiring their model to be retrained upon any changes.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Towards Practical Physics-Informed ML Design and Evaluation for Power Grid

Li¹,

Pandey²,

Pileggi³

2022

Preprint

View full text Add to dashboard Cite

When applied to a real-world safety critical system like the power grid, general machine learning methods suffer from expensive training, non-physical solutions, and limited interpretability. To address these challenges for power grids, many recent works have explored the inclusion of grid physics (i.e., domain expertise) into their method design, primarily through including system constraints and technical limits, reducing search space and defining meaningful features in latent space. Yet, there is no general methodology to evaluate the practicality of these approaches in power grid tasks, and limitations exist regarding scalability, generalization, interpretability, etc. This work formalizes a new concept of physical interpretability which assesses how a ML model makes predictions in a physically meaningful way and introduces an evaluation methodology that identifies a set of attributes that a practical method should satisfy. Inspired by the evaluation attributes, the paper further develops a novel contingency analysis warm starter for MadIoT cyberattack, based on a conditional Gaussian random field. This method serves as an instance of an ML model that can incorporate diverse domain knowledge and improve on these identified attributes. Experiments validate that the warm starter significantly boosts the efficiency of contingency analysis for MadIoT attack even with shallow NN architectures.

show abstract

“…First, there are studies that use supervised learning, i.e. event classification, such as those in [8]- [10]. They require prior labeling of the events in the training data set, which may not be doable in practice.…”

Section: Literature Reviewmentioning

confidence: 99%

GraphPMU: Event Clustering via Graph Representation Learning Using Locationally-Scarce Distribution-Level Fundamental and Harmonic PMU Measurements

Aligholian¹,

Mohsenian‐Rad²

2022

Preprint

View full text Add to dashboard Cite

This paper is concerned with the complex task of identifying the type and cause of the events that are captured by distribution-level phasor measurement units (D-PMUs) in order to enhance situational awareness in power distribution systems. Our goal is to address two fundamental challenges in this field: a) scarcity in measurement locations due to the high cost of purchasing, installing, and streaming data from D-PMUs; b) limited prior knowledge about the event signatures due to the fact that the events are diverse, infrequent, and inherently unscheduled. To tackle these challenges, we propose an unsupervised graph-representation learning method, called GraphPMU, to significantly improve the performance in event clustering under locationally-scarce data availability by proposing the following two new directions: 1) using the topological information about the relative location of the few available phasor measurement units on the graph of the power distribution network; 2) utilizing not only the commonly used fundamental phasor measurements, bus also the less explored harmonic phasor measurements in the process of analyzing the signatures of various events. Through a detailed analysis of several case studies, we show that GraphPMU can highly outperform the prevalent methods in the literature.

show abstract

Learning-Based Real-Time Event Identification Using Rich Real PMU Data

Cited by 28 publications

References 42 publications

Resilience Framework, Methods, and Metrics for the Prioritization of Critical Electrical Grid Customers

Resilience Framework, Methods, and Metrics for the Prioritization of Critical Electrical Grid Customers

Towards Practical Physics-Informed ML Design and Evaluation for Power Grid

GraphPMU: Event Clustering via Graph Representation Learning Using Locationally-Scarce Distribution-Level Fundamental and Harmonic PMU Measurements

Contact Info

Product

Resources

About