Pith. sign in

REVIEW 5 major objections 5 minor 48 references

Topology-Aware Graph Neural Network-based State Estimation for PMU-Unobservable Power Systems

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A graph neural network state estimator that replaces missing PMU measurements with Gaussian mixture distributions, claimed to stay accurate through topology changes and PMU failures without retraining.

desk verdict A useful GNN-SE application with strong empirical results, but Theorem 3's topology-change bound is not established for the actual architecture; worth refereeing with major revision. read the letter →

arxiv 2506.03493 v1 pith:ALXHNIJW submitted 2025-06-04 eess.SY cs.SY

Shiva Moshtagh , Behrouz Azimian , Mohammad Golgol , Anamitra Pal This is my paper
classification eess.SYcs.SY
keywords powersystemstateestimationgraphneuralnetworksphasormeasurementunitsPMUunobservabilitytopologychangerobustnessmulti-headattentionGaussianmixturemodelsbaddatadetection
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a graph neural network can estimate power system states even when only a minority of buses carry phasor measurement units, by replacing missing PMU data with distributions learned from historical data. It proposes CGNN-SE, which combines a Gaussian-mixture expected-activation graph-convolution layer with a multi-head graph attention layer, and reports that this estimator outperforms optimization-based and earlier learning-based estimators on several test systems. It also derives an upper bound on the change in the estimator output caused by a topology change. If the claim is right, time-synchronized state estimation would no longer require full PMU coverage or retraining after line openings and PMU failures.

What carries the argument

The central mechanism is an expected-activation graph-convolution layer followed by a multi-head graph attention layer. In the first layer, each node's feature vector is treated as a Gaussian mixture, with known PMU readings entered as zero-variance means and missing readings entered as learned Gaussian-mixture parameters; Theorem 2 replaces the ReLU of the convolved Gaussian mixture with a closed-form expression using the auxiliary function $N_R(z)$, so missing-feature parameters and network weights are optimized together. The multi-head attention layer then assigns each neighbor adaptive importance, and Theorem 3 bounds the output difference after a topology perturbation in terms of the change in the adjacency matrix, layer count, feature width, and weight norms.

What would settle it

Take the trained 118-bus CGNN-SE, open an overloaded line that the paper classifies as worst-case, recompute the actual output difference against the base topology, and compare it with the right-hand side of (19); an observed difference larger than the bound for a single line outage would refute the topology-adaptivity guarantee.

Watch

Extended reading notes

Core claim

The paper claims that a graph neural network can estimate the voltage magnitude and phase angle of every bus even when only a minority of buses carry phasor measurement units, by treating missing measurements as random variables modeled with Gaussian mixture distributions learned from historical data. Its proposed CGNN-SE computes the expected activation of a graph-convolution layer under those distributions, then passes the result through a multi-head graph attention layer, with all parameters trained end to end. In the paper's experiments on the IEEE 118-bus system, CGNN-SE with 11 PMU-equipped buses reaches 0.018% magnitude mean absolute percentage error and 0.027 degrees phase-angle mean absolute error, compared with 0.270% and 0.143 degrees for a linear least-squares estimator with 32 PMUs. On the 2000-bus Texas system it reports 0.093% magnitude error with 120 PMUs. The paper also derives an upper bound on the output difference after a topology change and presents this bound as a mathematical explanation of why retraining is unnecessary.

Load-bearing premise

The proof of topology-change stability assumes the outage's effect on graph signals stays local so the recursive error term can be dropped, and if the outage instead shifts the system-wide operating point, the advertised bound does not follow.

Editorial extensions

If this is right

  • PMU-only state estimation would no longer require full observability: the paper's 118-bus results use 11 PMUs where a fully observable least-squares estimator needs 32.
  • A trained estimator could keep operating without retraining after single and multiple line openings, including N-1 and N-2 contingencies.
  • Real-time PMU failures could be absorbed by substituting learned Gaussian-mixture parameters, with reported errors plateauing as more PMUs fail.
  • The estimator is reported to stay accurate under non-Gaussian measurement noise and, with a Wald-test preprocessing step, under bad data in up to 50% of input features.
  • The reported online computation time for the 2000-bus system is below the 33 ms PMU timescale, so the approach could run in real time.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural next experiment, suggested by the proof rather than run in the paper, would open lines far from the PMU locations that trigger large re-dispatch; if the actual output difference grows faster than the bound in (19), the localized-effects assumption is the part that gives way.
  • The same expected-activation layer could carry over to distribution systems, where switch-status changes play the role of line openings and PMU coverage is even sparser; the architecture itself does not rely on transmission-level assumptions.
  • Because missing features are modeled as distributions, the Gaussian-mixture covariances could be used to output a confidence interval with each state estimate, a capability the paper does not develop.
  • The bound in (19) offers a design rule: keeping the architecture shallow and narrow should improve worst-case stability after topology changes, which the paper uses to justify hyperparameter tuning beyond raw validation accuracy.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes CGNN-SE, a graph neural network state estimator for PMU-unobservable power systems. A customized GCN first layer replaces missing PMU features by expected activations under Gaussian mixture models, followed by a multi-head graph attention (MH-GAT) layer and a linear readout for voltage magnitude and phase angle. The paper compares CGNN-SE against SCADA-SE, LSE, SVR, DNN-SE, GNN-SE, and WLM-SE on the IEEE 118-bus, 2000-bus Texas, and 1354pegase systems, under topology changes, PMU failures, bad data, and non-Gaussian noise. It also states Theorem 3, an upper bound on the change in estimator output under topology perturbation.

Significance. The empirical core of the paper is valuable and internally consistent. CGNN-SE reports lower magnitude MAPE and phase-angle MAE than all baselines with fewer PMUs, and the ablation in Table III isolates the benefit of the MH-GAT layer, while Table IV documents head-count sensitivity. The robustness figures for PMU failures and bad data correction are informative, and the paper is transparent that training is offline and online inference is fast. However, the advertised mathematical guarantee, Theorem 3 and Eq. (19), is not established for the implemented architecture, and the empirical evaluation consists of point estimates from a single simulator-based pipeline with no error bars or released artifacts. The contribution is therefore defensible as an empirical method paper, but the theoretical backing stated in the abstract and Section III-C needs substantial correction.

major comments (5)
  1. [§III-C and Appendix B] The proof of Theorem 3 models layers of the form x^(l) = sigma(A x^(l-1) W^T) with a fixed adjacency A, but Algorithm 1's CGNN-SE uses Eq. (14), the GMM expected ReLU, for the first layer and Eq. (18), MH-GAT, for the final layer. In the MH-GAT layer, each attention matrix A_k is recomputed from the input features, so A_k is not fixed when X changes to X'. Consequently, Eq. (19) does not bound the output change of the implemented architecture. Please either prove the bound for the actual layer updates used in CGNN-SE or explicitly scope Theorem 3 to the plain GCN backbone.
  2. [Appendix B, Eq. (B.8) to Eq. (B.10)] The recursive term ||x^(l-1)_:g - x'^(l-1)_:g|| is dropped from the bound solely by citing 'localized effects' [48]. If that term is retained, the recursion scales as e_l <= C_sigma (delta epsilon n_{l-1} + B e_{l-1}), which can grow geometrically in the number of layers, so the epsilon-linear bound in Eq. (19) does not follow. Moreover, Section IV-C explicitly states that robustness depends on 'the resulting shift in the power system's overall operating point,' which is a global quantity and is in tension with the localized-effects premise. A quantitative localization lemma is needed before Theorem 3 can be accepted.
  3. [Theorem 3 and abstract] The abstract and Section III-C describe Eq. (19) as an upper bound on 'SE error,' but the proof bounds only ||Phi(A,X,W) - Phi(A',X',W)||, i.e., the change in the network output between two topologies. This quantity is not the estimation error with respect to the true system state, which is what the abstract advertises. Please either restate the guarantee as an output-sensitivity bound or extend the argument to bound the actual estimation error.
  4. [Appendix A, Eq. (A.2)] The variance propagation in Eq. (A.2) writes the variance of (AXW)_ij as the sum of a_in^2 s_nf w_fj^2 terms, which requires an independence or uncorrelatedness assumption across features and nodes. This assumption is not stated, and it is questionable for power-system state variables, where voltage magnitudes and phase angles at neighboring buses are strongly correlated. Because Eq. (14) is a load-bearing component of the proposed first layer, the assumption should be stated explicitly and justified, or the variance formula should be replaced with one that accounts for cross-node covariances.
  5. [Section IV, Tables I-VI and Figs. 5-10] All reported results are point estimates from a single simulator-based pipeline, with no error bars, confidence intervals, or multiple random seeds. The margin over GNN-SE in the 2000-bus phase-angle MAE is 0.064 versus 0.083 degrees, and the 1354pegase comparison in Table VI reports only a single sigma_y^2 value per method without stating the number of test scenarios or repeated runs. Please report means and standard deviations over independent training runs and test samples, and specify the exact test-set sizes, so that the claimed superiority can be assessed statistically.
minor comments (5)
  1. [Section I, last paragraph] There is a typographical error: 'CGGN-SE' should be 'CGNN-SE'.
  2. [Appendix B] The word 'Lipshitz' should be 'Lipschitz' in the footnote to Eq. (B.4).
  3. [Section IV-J and Table VI] The 1354pegase experiment is underdescribed: please specify the PMU placement, noise model, number of test samples, and the training details that are 'chosen based on [39]' so that the comparison in Table VI is reproducible.
  4. [Section III-D, Eq. (20)] The notation Q^{-1}(alpha/2) and the decision rule in Eq. (20) should be defined more carefully; in particular, the roles of the two hypotheses and the threshold should be stated explicitly.
  5. [Appendix B, reference [48]] Reference [48] concerns localization in power-system security analysis, not localization of graph-signal perturbations in GNNs; a more direct reference or a self-contained lemma would make the argument easier to verify.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the GMM prior is an input distribution, not a definitional copy of the output, and the topology-bound gap is a proof rigor issue rather than an input-output equivalence.

full rationale

The claimed derivation chain is not circular in the sense defined here. The first-layer GMM expected-activation formula (Theorem 2 / Eq. (14)) is derived from the stated distributional assumption on X; it is a valid mathematical transformation, and the output state estimator (Eq. (6)) is not set equal to that input by construction. The GMM parameters for PMU-less nodes are fitted to historical power-flow-solver outputs and the same solver outputs provide the supervised labels; this creates a self-consistency / potential leakage risk, but it is a statistical overlap of training data, not an equivalence between the derivation's premise and conclusion. The topology-robustness bound in Theorem 3 has a real proof gap: the step from (B.8) to (B.10) drops the recursive difference term by appealing to the unproved 'localized effects' assertion from [48], and the proof uses plain GCN layers rather than the GMM first layer and MH-GAT final layer of Algorithm 1. That is an unsupported mathematical guarantee and an architecture mismatch, but it is not a circular reduction: the theorem's bound is not assumed in the inputs, it merely fails to follow. Comparisons to LSE, DNN-SE, GNN-SE and WLM-SE are external baselines; self-citations [28] and [47] supply architecture and prior-work context but are not the load-bearing justification for the central empirical claim. No step reduces by definition or by fitted-parameter renaming to its own input, so the appropriate finding is no significant circularity.

Assumptions & free parameters 7 free parameters · 7 assumptions · 0 invented entities

The central claim rests on a synthetic data-generation pipeline, a GMM-based missing-feature model, a multi-head attention architecture, and a stability bound whose proof assumes localized topology effects. No new physical entities are introduced, but several hyperparameters are unreported and the simulation pipeline substitutes for field validation.

free parameters (7)
  • GNN weight matrices and linear-layer bias = learned, values not reported
    Optimized by gradient descent on generated training data; all reported accuracy depends on this fit.
  • GMM component weights, means, variances = initialized by EM, updated end-to-end; values not reported
    Define the distribution used for every PMU-less node's missing feature in Eq. (7).
  • Hidden feature count per layer, F = 50
    Architecture hyperparameter; Theorem 3's bound grows with F^(L-2).
  • Number of attention heads, K = 4
    Tuned on the 118-bus system; ablation in Table IV shows error changes from 0.018% to 0.131% across K.
  • Number of Gaussian components, C = not reported
    Input to Algorithm 1 but never specified in the simulation setup.
  • Wald test false-positive rate, alpha = not reported
    Controls bad-data detection threshold in Eq. (20); not specified.
  • Kernel density estimation bandwidth = not reported
    Used to learn load distributions from SCADA history; bandwidth affects sampled operating conditions.
assumptions (7)
  • domain assumption Power-flow-generated synthetic states accurately represent real grid states and remain representative at test time.
    Section IV-A generates training labels and GMM targets from a power-flow solver driven by KDE-sampled loads; no real voltage measurements are used in training or evaluation.
  • domain assumption Topology changes are known and supplied through an updated adjacency matrix during online operation.
    Section III-C models topology change as A to A' and X to X', but the paper does not describe how line openings are detected or communicated in real time.
  • ad hoc to paper Node features are independent enough that the variance of a weighted sum is the sum of per-feature variances.
    Appendix A derives V(AXW) using (A elementwise A)S(W elementwise W), which requires independence across the summed random variables.
  • ad hoc to paper Topology-change effects on intermediate graph signals are localized and the recursive error term can be neglected.
    Appendix B drops the term ||x(l-1)_:g - x'(l-1)_:g||; the stated bound depends on this unproved simplification.
  • standard math ReLU is Lipschitz continuous with constant C_sigma = 1.
    Used in Appendix B to bound layer-to-layer error propagation; this is standard for ReLU.
  • domain assumption Missing PMU features are representable by a Gaussian mixture distribution.
    Equation (7) assumes any node feature vector, observed or missing, follows a mixture of Gaussians.
  • domain assumption The simulated noise models match realistic PMU errors.
    Section IV-F uses a two-component GMM noise constrained by 1% TVE; Section IV-B uses 1% Gaussian TVE, but no field data validate these models.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Topology-Aware Graph Neural Network-based State Estimation for PMU-Unobservable Power Systems." pith.science (2026). https://pith.science/paper/ALXHNIJW

@misc{pith2026250603493,
  author       = {Pith},
  title        = {Pith review of: Topology-Aware Graph Neural Network-based State Estimation for PMU-Unobservable Power Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ALXHNIJW}},
  note         = {Machine review of arXiv:2506.03493}
}
read the original abstract

Traditional optimization-based techniques for time-synchronized state estimation (SE) often suffer from high online computational burden, limited phasor measurement unit (PMU) coverage, and presence of non-Gaussian measurement noise. Although conventional learning-based models have been developed to overcome these challenges, they are negatively impacted by topology changes and real-time data loss. This paper proposes a novel deep geometric learning approach based on graph neural networks (GNNs) to estimate the states of PMU-unobservable power systems. The proposed approach combines graph convolution and multi-head graph attention layers inside a customized end-to-end learning framework to handle topology changes and real-time data loss. An upper bound on SE error as a function of topology change is also derived. Experimental results for different test systems demonstrate superiority of the proposed customized GNN-SE (CGNN-SE) over traditional optimization-based techniques as well as conventional learning-based models in presence of topology changes, PMU failures, bad data, non-Gaussian measurement noise, and large system implementation.

Figures

Figures reproduced from arXiv: 2506.03493 by the authors.

Figure 1
Figure 1. Total number of branches in a utility power system located in the U.S. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. An illustration of the MH-GAT for node 1 at layer [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The proposed CGNN-SE framework showcases an end-to-end geometric learning process, where a power system graph with missing features is fed [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Overall implementation of the proposed CGNN-SE [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 6
Figure 6. Figure 6: Stability of the proposed CGNN-SE for IEEE 118-bus system under [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 5
Figure 5. Figure 5: Comparing magnitude MAPE and phase angle MAE of DNN-SE and [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 7
Figure 7. Figure 7: Robustness of CGNN-SE to topology changes including [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 9
Figure 9. Figure 9: Robustness of CGNN-SE to simultaneous topology change and [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 8
Figure 8. Figure 8: Robustness of CGNN-SE to PMU failures in IEEE 118-bus system [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 10
Figure 10. Figure 10: Performance of CGNN-SE without and with bad data correction in [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 42 canonical work pages

  1. [48]

    An adaptive localization method for real-time security analysis,

    G. Ejebe, R. Paliza, and W. Tinney, “An adaptive localization method for real-time security analysis,” IEEE Transactions on Power Systems , vol. 7, no. 2, pp. 777–783, 1992. Shiva Moshtagh (Student Member, IEEE) received the B.Sc. degree in electrical engineering from the Jundi-Shapur University of Technology, Dezful, Iran, in 2015, and the M.Sc. degree i...

    work page 1992

  2. [1]

    Mitigation of saturated cut-sets during multiple outages to enhance power system security,

    R. S. Biswas, A. Pal, T. Werho, and V . Vittal, “Mitigation of saturated cut-sets during multiple outages to enhance power system security,” IEEE Transactions on Power Systems , vol. 36, no. 6, pp. 5734–5745, 2021

    work page 2021

  3. [2]

    A high temporal-spatial resolution power system state estimation method for online DSA,

    J. Hu, Q. Wang, Y . Ye, Z. Wu, and Y . Tang, “A high temporal-spatial resolution power system state estimation method for online DSA,” IEEE Transactions on Power Systems , 2023

    work page 2023

  4. [3]

    Micro-flexibility: Challenges for power sys- tem modeling and control,

    S. Chatzivasileiadis et al., “Micro-flexibility: Challenges for power sys- tem modeling and control,” Electric Power Systems Research, vol. 216, p. 109002, 2023

    work page 2023

  5. [4]

    Real-time equality-constrained hybrid state estimation in complex variables,

    I. D ˇzafi´c and R. A. Jabr, “Real-time equality-constrained hybrid state estimation in complex variables,” International Journal of Electrical Power & Energy Systems , vol. 117, p. 105634, 2020

    work page 2020

  6. [5]

    Algebraic approach to PMU place- ment for minimum variance linear state estimation in power networks,

    A. Mishra and R. A. de Callafon, “Algebraic approach to PMU place- ment for minimum variance linear state estimation in power networks,” IEEE Transactions on Power Systems , 2022

    work page 2022

  7. [6]

    A survey of power system state estimation using multiple data sources: PMUs, SCADA, AMI, and beyond,

    G. Cheng, Y . Lin, A. Abur, A. G ´omez-Exp´osito, and W. Wu, “A survey of power system state estimation using multiple data sources: PMUs, SCADA, AMI, and beyond,” IEEE Transactions on Smart Grid , 2023

    2023

  8. [7]

    A robust state estimation framework considering measurement correlations and imperfect synchronization,

    J. Zhao, S. Wang, L. Mili, B. Amidan, R. Huang, and Z. Huang, “A robust state estimation framework considering measurement correlations and imperfect synchronization,” IEEE Transactions on Power Systems , vol. 33, no. 4, pp. 4604–4613, 2018

    work page 2018

Show all 48 references
  1. [8]

    A survey on hybrid SCADA/W AMS state estimation methodologies in electric power transmission systems,

    O. Darmis and G. Korres, “A survey on hybrid SCADA/W AMS state estimation methodologies in electric power transmission systems,” En- ergies, vol. 16, no. 2, p. 618, 2023

    2023

  2. [9]

    Power system state esti- mation using PMUs with imperfect synchronization,

    P. Yang, Z. Tan, A. Wiesel, and A. Nehorai, “Power system state esti- mation using PMUs with imperfect synchronization,” IEEE Transactions on Power Systems, vol. 28, no. 4, pp. 4162–4172, 2013

    2013

  3. [10]

    Power system real-time monitoring by using PMU-based robust state estimation method,

    J. Zhao et al., “Power system real-time monitoring by using PMU-based robust state estimation method,” IEEE Transactions on Smart Grid , vol. 7, no. 1, pp. 300–309, 2015

    2015

  4. [11]

    A hybrid power system state es- timator using synchronized and unsynchronized sensors,

    N. M. Manousakis and G. N. Korres, “A hybrid power system state es- timator using synchronized and unsynchronized sensors,” International Transactions on Electrical Energy Systems , vol. 28, no. 8, p. e2580, 2018

    2018

  5. [12]

    Analysis of hybrid state estimators: Accuracy and convergence of estimator formulations,

    Z. Jin, P. Wall, Y . Chen, J. Yu, S. Chakrabarti, and V . Terzija, “Analysis of hybrid state estimators: Accuracy and convergence of estimator formulations,” IEEE Transactions on Power Systems , vol. 34, no. 4, pp. 2565–2576, 2018

    2018

  6. [13]

    Optimal placement of phasor measurement unit in smart grids considering IEEE TRANSACTIONS ON POWER SYSTEMS 13 multiple constraints,

    T. Chen, H. Ren, Y . Sun, M. Kraft, and G. A. Amaratunga, “Optimal placement of phasor measurement unit in smart grids considering IEEE TRANSACTIONS ON POWER SYSTEMS 13 multiple constraints,” Journal of Modern Power Systems and Clean Energy, vol. 11, no. 2, pp. 479–488, 2022

    2022

  7. [14]

    Deep neural network-based state estimator for transmission system considering practical implementation challenges,

    A. C. Varghese, H. Shah, B. Azimian, A. Pal, and E. Farantatos, “Deep neural network-based state estimator for transmission system considering practical implementation challenges,” Journal of Modern Power Systems and Clean Energy , 2024

    2024

  8. [15]

    Neural-network-based power system state estimation with extended observability,

    G. Tian, Y . Gu, D. Shi, J. Fu, Z. Yu, and Q. Zhou, “Neural-network-based power system state estimation with extended observability,” Journal of Modern Power Systems and Clean Energy, vol. 9, no. 5, pp. 1043–1053, 2021

    2021

  9. [16]

    Time synchronized state estimation for incompletely observed distribution systems using deep learning considering realistic measurement noise,

    B. Azimian, R. S. Biswas, A. Pal, and L. Tong, “Time synchronized state estimation for incompletely observed distribution systems using deep learning considering realistic measurement noise,” in 2021 IEEE Power & Energy Society General Meeting (PESGM) , pp. 1–5, 2021

    2021

  10. [17]

    Learning the unobservable: High-resolution state estimation via deep learning,

    K. R. Mestav and L. Tong, “Learning the unobservable: High-resolution state estimation via deep learning,” in 2019 57th Annual Allerton Conference on Communication, Control, and Computing (Allerton) , pp. 171–176, IEEE, 2019

    2019

  11. [18]

    Dynamic detection of transmission line outages using hidden Markov models,

    Q. Huang, L. Shao, and N. Li, “Dynamic detection of transmission line outages using hidden Markov models,” IEEE Transactions on Power Systems, vol. 31, no. 3, pp. 2026–2033, 2015

    2026

  12. [19]

    Real-time contingency analysis with corrective transmission switching,

    X. Li, P. Balasubramanian, M. Sahraei-Ardakani, M. Abdi-Khorsand, K. W. Hedman, and R. Podmore, “Real-time contingency analysis with corrective transmission switching,” IEEE Transactions on Power Systems, vol. 32, no. 4, pp. 2604–2617, 2017

    2017

  13. [20]

    A deep neural network approach for online topology identification in state estimation,

    D. Gotti, H. Amaris, and P. L. Larrea, “A deep neural network approach for online topology identification in state estimation,” IEEE Transactions on Power Systems, vol. 36, no. 6, pp. 5824–5833, 2021

    2021

  14. [21]

    State and topology estimation for unobservable distri- bution systems using deep neural networks,

    B. Azimian, R. S. Biswas, S. Moshtagh, A. Pal, L. Tong, and G. Dasarathy, “State and topology estimation for unobservable distri- bution systems using deep neural networks,” IEEE Transactions on Instrumentation and Measurement , vol. 71, pp. 1–14, 2022

    2022

  15. [22]

    A review of graph neural networks and their applications in power systems,

    W. Liao, B. Bak-Jensen, J. R. Pillai, Y . Wang, and Y . Wang, “A review of graph neural networks and their applications in power systems,” Journal of Modern Power Systems and Clean Energy , vol. 10, no. 2, pp. 345– 360, 2021

    2021

  16. [23]

    Geometric deep learning: going beyond Euclidean data,

    M. M. Bronstein, J. Bruna, Y . LeCun, A. Szlam, and P. Vandergheynst, “Geometric deep learning: going beyond Euclidean data,” IEEE Signal Processing Magazine, vol. 34, no. 4, pp. 18–42, 2017

    2017

  17. [24]

    Data-driven priors for robust PSSE via Gauss-Newton unrolled neural networks,

    Q. Yang, A. Sadeghi, and G. Wang, “Data-driven priors for robust PSSE via Gauss-Newton unrolled neural networks,” IEEE Journal on Emerging and Selected Topics in Circuits and Systems , vol. 12, no. 1, pp. 172–181, 2022

    2022

  18. [25]

    Integrating model- driven and data-driven methods for fast state estimation,

    Z. Wu, Q. Wang, J. Hu, Y . Tang, and Y . Zhang, “Integrating model- driven and data-driven methods for fast state estimation,” International Journal of Electrical Power & Energy Systems , vol. 139, p. 107982, 2022

    2022

  19. [26]

    Physics-informed geometric deep learning for inference tasks in power systems,

    S. de Jongh, F. Gielnik, F. Mueller, L. Schmit, M. Suriyah, and T. Leibfried, “Physics-informed geometric deep learning for inference tasks in power systems,” Electric Power Systems Research , vol. 211, p. 108362, 2022

    2022

  20. [27]

    Graph neural networks on factor graphs for robust, fast, and scalable linear state estimation with PMUs,

    O. Kundacina, M. Cosovic, D. Miskovic, and D. Vukobratovic, “Graph neural networks on factor graphs for robust, fast, and scalable linear state estimation with PMUs,” Sustainable Energy, Grids and Networks , vol. 34, p. 101056, 2023

    2023

  21. [28]

    Time-synchronized state estimation using graph neural networks in presence of topology changes,

    S. Moshtagh, A. I. Sifat, B. Azimian, and A. Pal, “Time-synchronized state estimation using graph neural networks in presence of topology changes,” in 2023 North American Power Symposium , pp. 1–6, 2023

    2023

  22. [29]

    Learning on attribute-missing graphs,

    X. Chen, S. Chen, J. Yao, H. Zheng, Y . Zhang, and I. W. Tsang, “Learning on attribute-missing graphs,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 2, pp. 740–757, 2022

    2022

  23. [30]

    On the unreasonable effectiveness of feature propagation in learning on graphs with missing node features,

    E. Rossi, H. Kenlay, M. I. Gorinova, B. P. Chamberlain, X. Dong, and M. M. Bronstein, “On the unreasonable effectiveness of feature propagation in learning on graphs with missing node features,” in Learning on Graphs Conference , pp. 11–1, PMLR, 2022

    2022

  24. [31]

    Local augmentation for graph neural networks,

    S. Liu et al., “Local augmentation for graph neural networks,” in Inter- national Conference on Machine Learning , pp. 14054–14072, PMLR, 2022

    2022

  25. [32]

    Graph convolutional networks for graphs containing missing features,

    H. Taguchi, X. Liu, and T. Murata, “Graph convolutional networks for graphs containing missing features,” Future Generation Computer Systems, vol. 117, pp. 155–168, 2021

    2021

  26. [33]

    Graph attention networks,

    P. Veli ˇckovi´c, G. Cucurull, A. Casanova, A. Romero, P. Li `o, and Y . Bengio, “Graph attention networks,” in International Conference on Learning Representations, 2018

    2018

  27. [34]

    A topology adaptive high-speed transient stability assessment scheme based on multi-graph attention network with residual structure,

    J. Huang, L. Guan, Y . Su, H. Yao, M. Guo, and Z. Zhong, “A topology adaptive high-speed transient stability assessment scheme based on multi-graph attention network with residual structure,” International Journal of Electrical Power & Energy Systems , vol. 130, p. 106948, 2021

    2021

  28. [35]

    Processing of missing data by neural networks,

    M. ´Smieja, Ł. Struski, J. Tabor, B. Zieli ´nski, and P. Spurek, “Processing of missing data by neural networks,” Advances in Neural Information Processing Systems, vol. 31, 2018

    2018

  29. [36]

    Stability properties of graph neural networks,

    F. Gama, J. Bruna, and A. Ribeiro, “Stability properties of graph neural networks,” IEEE Transactions on Signal Processing , vol. 68, pp. 5680– 5695, 2020

    2020

  30. [37]

    Bayesian state estimation for unobservable distribution systems via deep learning,

    K. R. Mestav, J. Luengo-Rozas, and L. Tong, “Bayesian state estimation for unobservable distribution systems via deep learning,” IEEE Trans- actions on Power Systems , vol. 34, no. 6, pp. 4910–4920, 2019

    2019

  31. [38]

    Multichannel signal detection based on Wald test in subspace interference and Gaussian noise,

    W. Liu, J. Liu, H. Li, Q. Du, and Y .-L. Wang, “Multichannel signal detection based on Wald test in subspace interference and Gaussian noise,” IEEE Transactions on Aerospace and Electronic Systems, vol. 55, no. 3, pp. 1370–1381, 2018

    2018

  32. [39]

    Complex-variable weighted least modulus state estimation,

    R. A. Jabr, “Complex-variable weighted least modulus state estimation,” IEEE Transactions on Power Systems , vol. 39, no. 4, pp. 6026–6035, 2024

    2024

  33. [40]

    IEEE/IEC international standard - measuring relays and protection equipment - part 118-1: Synchrophasor for power systems - measure- ments,

    “IEEE/IEC international standard - measuring relays and protection equipment - part 118-1: Synchrophasor for power systems - measure- ments,” 2018

    2018

  34. [41]

    General optimal substation coverage algorithm for phasor measurement unit placement in practical systems,

    A. Pal, C. Mishra, A. K. S. Vullikanti, and S. Ravi, “General optimal substation coverage algorithm for phasor measurement unit placement in practical systems,” IET Generation, Transmission & Distribution , vol. 11, no. 2, pp. 347–353, 2017

    2017

  35. [42]

    A robust hybrid power system state estimator with unknown measurement noise,

    J. Zhao, L. Mili, and M. La Scala, “A robust hybrid power system state estimator with unknown measurement noise,” Advances in Electric Power and Energy: Static State Estimation , pp. 231–253, 2020

    2020

  36. [43]

    A PMU placement scheme ensuring real-time monitoring of critical buses of the network,

    A. Pal, G. A. Sanchez-Ayala, V . A. Centeno, and J. S. Thorp, “A PMU placement scheme ensuring real-time monitoring of critical buses of the network,” IEEE Transactions on Power Delivery, vol. 29, no. 2, pp. 510– 517, 2013

    2013

  37. [44]

    Sparse graph attention networks,

    Y . Ye and S. Ji, “Sparse graph attention networks,” IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 1, pp. 905–916, 2021

    2021

  38. [45]

    Multi-hop attention graph neural network,

    G. Wang, R. Ying, J. Huang, and J. Leskovec, “Multi-hop attention graph neural network,” arXiv preprint arXiv:2009.14332 , 2020

    2009 arXiv

  39. [46]

    Subgraph- based attention network for multi-hop question answering,

    C. Gong, Z. Wei, X. Wang, R. Wang, Y . Li, and P. Zhu, “Subgraph- based attention network for multi-hop question answering,” in 2024 International Joint Conference on Neural Networks (IJCNN) , pp. 1–9, IEEE, 2024

    2024

  40. [47]

    Analytical verification of performance of deep neural network based time-synchronized distri- bution system state estimation,

    B. Azimian, S. Moshtagh, A. Pal, and S. Ma, “Analytical verification of performance of deep neural network based time-synchronized distri- bution system state estimation,” Journal of Modern Power Systems and Clean Energy, vol. 12, no. 4, pp. 1126–1134, 2024

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.