Real-Time Power System State Estimation and Forecasting via Deep Unrolled Neural Networks

Liang Zhang, Gang Wang, Georgios B. Giannakis

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Contemporary power grids are being challenged by rapid and sizeable voltage fluctuations that are caused by large-scale deployment of renewable generators, electric vehicles, and demand response programs. In this context, monitoring the grid's operating conditions in real time becomes increasingly critical. With the emergent large scale and nonconvexity, existing power system state estimation (PSSE) schemes become computationally expensive or often yield suboptimal performance. To bypass these hurdles, this paper advocates physics-inspired deep neural networks (DNNs) for real-time power system monitoring. By unrolling an iterative solver that was originally developed using the exact ac model, a novel model-specific DNN is developed for real-time PSSE requiring only offline training and minimal tuning effort. To further enable system awareness, even ahead of the time horizon, as well as to endow the DNN-based estimator with resilience, deep recurrent neural networks (RNNs) are also pursued for power system state forecasting. Deep RNNs leverage the long-term nonlinear dependencies present in the historical voltage time series to enable forecasting, and they are easy to implement. Numerical tests showcase improved performance of the proposed DNN-based estimation and forecasting approaches compared with existing alternatives. In real load data experiments on the IEEE 118-bus benchmark system, the novel model-specific DNN-based PSSE scheme outperforms nearly by an order-of-magnitude its competing alternatives, including the widely adopted Gauss-Newton PSSE solver.

Original languageEnglish (US)
Article number8754766
Pages (from-to)4069-4077
Number of pages9
JournalIEEE Transactions on Signal Processing
Volume67
Issue number15
DOIs
StatePublished - Aug 1 2019

Fingerprint

State estimation
Recurrent neural networks
Electric power system measurement
Electric potential
Electric vehicles
Time series
Physics
Tuning
Deep neural networks
Monitoring
Experiments

Keywords

  • Power system state estimation
  • data validation
  • forecasting
  • least-absolute-value
  • proximal linear algorithm
  • recurrent neural networks

Cite this

Real-Time Power System State Estimation and Forecasting via Deep Unrolled Neural Networks. / Zhang, Liang; Wang, Gang; Giannakis, Georgios B.

In: IEEE Transactions on Signal Processing, Vol. 67, No. 15, 8754766, 01.08.2019, p. 4069-4077.

Research output: Contribution to journalArticle

@article{450147952c454e669f7c76600522c76a,
title = "Real-Time Power System State Estimation and Forecasting via Deep Unrolled Neural Networks",
abstract = "Contemporary power grids are being challenged by rapid and sizeable voltage fluctuations that are caused by large-scale deployment of renewable generators, electric vehicles, and demand response programs. In this context, monitoring the grid's operating conditions in real time becomes increasingly critical. With the emergent large scale and nonconvexity, existing power system state estimation (PSSE) schemes become computationally expensive or often yield suboptimal performance. To bypass these hurdles, this paper advocates physics-inspired deep neural networks (DNNs) for real-time power system monitoring. By unrolling an iterative solver that was originally developed using the exact ac model, a novel model-specific DNN is developed for real-time PSSE requiring only offline training and minimal tuning effort. To further enable system awareness, even ahead of the time horizon, as well as to endow the DNN-based estimator with resilience, deep recurrent neural networks (RNNs) are also pursued for power system state forecasting. Deep RNNs leverage the long-term nonlinear dependencies present in the historical voltage time series to enable forecasting, and they are easy to implement. Numerical tests showcase improved performance of the proposed DNN-based estimation and forecasting approaches compared with existing alternatives. In real load data experiments on the IEEE 118-bus benchmark system, the novel model-specific DNN-based PSSE scheme outperforms nearly by an order-of-magnitude its competing alternatives, including the widely adopted Gauss-Newton PSSE solver.",
keywords = "Power system state estimation, data validation, forecasting, least-absolute-value, proximal linear algorithm, recurrent neural networks",
author = "Liang Zhang and Gang Wang and Giannakis, {Georgios B.}",
year = "2019",
month = "8",
day = "1",
doi = "10.1109/TSP.2019.2926023",
language = "English (US)",
volume = "67",
pages = "4069--4077",
journal = "IEEE Transactions on Signal Processing",
issn = "1053-587X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "15",

}

TY - JOUR

T1 - Real-Time Power System State Estimation and Forecasting via Deep Unrolled Neural Networks

AU - Zhang, Liang

AU - Wang, Gang

AU - Giannakis, Georgios B.

PY - 2019/8/1

Y1 - 2019/8/1

N2 - Contemporary power grids are being challenged by rapid and sizeable voltage fluctuations that are caused by large-scale deployment of renewable generators, electric vehicles, and demand response programs. In this context, monitoring the grid's operating conditions in real time becomes increasingly critical. With the emergent large scale and nonconvexity, existing power system state estimation (PSSE) schemes become computationally expensive or often yield suboptimal performance. To bypass these hurdles, this paper advocates physics-inspired deep neural networks (DNNs) for real-time power system monitoring. By unrolling an iterative solver that was originally developed using the exact ac model, a novel model-specific DNN is developed for real-time PSSE requiring only offline training and minimal tuning effort. To further enable system awareness, even ahead of the time horizon, as well as to endow the DNN-based estimator with resilience, deep recurrent neural networks (RNNs) are also pursued for power system state forecasting. Deep RNNs leverage the long-term nonlinear dependencies present in the historical voltage time series to enable forecasting, and they are easy to implement. Numerical tests showcase improved performance of the proposed DNN-based estimation and forecasting approaches compared with existing alternatives. In real load data experiments on the IEEE 118-bus benchmark system, the novel model-specific DNN-based PSSE scheme outperforms nearly by an order-of-magnitude its competing alternatives, including the widely adopted Gauss-Newton PSSE solver.

AB - Contemporary power grids are being challenged by rapid and sizeable voltage fluctuations that are caused by large-scale deployment of renewable generators, electric vehicles, and demand response programs. In this context, monitoring the grid's operating conditions in real time becomes increasingly critical. With the emergent large scale and nonconvexity, existing power system state estimation (PSSE) schemes become computationally expensive or often yield suboptimal performance. To bypass these hurdles, this paper advocates physics-inspired deep neural networks (DNNs) for real-time power system monitoring. By unrolling an iterative solver that was originally developed using the exact ac model, a novel model-specific DNN is developed for real-time PSSE requiring only offline training and minimal tuning effort. To further enable system awareness, even ahead of the time horizon, as well as to endow the DNN-based estimator with resilience, deep recurrent neural networks (RNNs) are also pursued for power system state forecasting. Deep RNNs leverage the long-term nonlinear dependencies present in the historical voltage time series to enable forecasting, and they are easy to implement. Numerical tests showcase improved performance of the proposed DNN-based estimation and forecasting approaches compared with existing alternatives. In real load data experiments on the IEEE 118-bus benchmark system, the novel model-specific DNN-based PSSE scheme outperforms nearly by an order-of-magnitude its competing alternatives, including the widely adopted Gauss-Newton PSSE solver.

KW - Power system state estimation

KW - data validation

KW - forecasting

KW - least-absolute-value

KW - proximal linear algorithm

KW - recurrent neural networks

UR - http://www.scopus.com/inward/record.url?scp=85069758394&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069758394&partnerID=8YFLogxK

U2 - 10.1109/TSP.2019.2926023

DO - 10.1109/TSP.2019.2926023

M3 - Article

AN - SCOPUS:85069758394

VL - 67

SP - 4069

EP - 4077

JO - IEEE Transactions on Signal Processing

JF - IEEE Transactions on Signal Processing

SN - 1053-587X

IS - 15

M1 - 8754766

ER -