pith. sign in

arxiv: 2510.23873 · v3 · submitted 2025-10-27 · 📡 eess.SY · cs.SY

A Spatio-Temporal Graph Learning Approach to Real-Time Economic Dispatch with Multi-Transmission-Node DER Aggregation

Zhentong Shao , Jingtao Qin , Xianbang Chen , Nanpeng Yu This is my paper

Pith reviewed 2026-05-18 02:45 UTC · model grok-4.3

classification 📡 eess.SY cs.SY
keywords real-time economic dispatchdistributed energy resourcesspatio-temporal graph convolutional networktransmission-node aggregationsecurity-constrained optimizationmarket clearing
0
0 comments X

The pith

A spatio-temporal graph network predicts distribution factors to speed real-time economic dispatch for aggregated DERs at multiple transmission nodes.

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

The paper develops a real-time economic dispatch framework that lets regional operators include aggregated distributed energy resources connected at transmission nodes without breaking existing market rules. It uses a spatio-temporal graph convolutional network to forecast how each aggregator affects power flows across the wider grid. An added iterative step quickly flags which transmission limits matter most at each interval. Together these steps cut computation time while still keeping the network inside safe operating bounds, as shown on test systems up to three thousand buses with actual demand records.

Core claim

By training a spatio-temporal graph convolutional network on historical data to forecast time-varying distribution factors for each transmission-node DER aggregator, and pairing it with an iterative constraint-screening routine, the method reduces the size of the security-constrained optimization solved at each real-time interval while preserving transmission feasibility on large networks.

What carries the argument

The spatio-temporal graph convolutional network that learns and predicts the dynamic influence of individual T-DER aggregators on transmission-line flows.

If this is right

  • Market-clearing software can solve the real-time problem faster because many transmission inequalities are screened out before the optimizer runs.
  • Aggregators of distribution-level resources can submit offers at transmission nodes and still have their impact on the bulk grid accounted for without manual distribution-factor updates.
  • Operational cost savings appear on the tested 118-, 2383-, and 3012-bus systems when the method is compared with static or non-graph baselines.
  • The framework stays compatible with current RTO market structures because it works inside the existing security-constrained economic dispatch formulation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same prediction-plus-screening pattern could be reused for other time-critical network-constrained optimizations such as contingency analysis or unit commitment.
  • If the graph model is retrained periodically on fresh telemetry, the approach may remain accurate even as the mix of solar, storage, and demand response changes over months or years.
  • Extending the graph to include weather or price signals as extra node features might further reduce forecast error for distribution factors under high renewable penetration.

Load-bearing premise

The graph network can forecast each aggregator's effect on the transmission grid accurately enough in rolling operation that the resulting dispatch stays feasible and reliable.

What would settle it

Run the full rolling dispatch loop on a 3000-bus model for several hours of real load data; compare the ST-GCN-predicted distribution factors against exact AC power-flow solutions and check whether any transmission line or voltage limit is actually violated in the cleared schedule.

Figures

Figures reproduced from arXiv: 2510.23873 by Jingtao Qin, Nanpeng Yu, Xianbang Chen, Zhentong Shao.

Figure 2
Figure 2. Figure 2: The timeline of RTED and DERA’s self-dispatch. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 1
Figure 1. Figure 1: Illustration of multi-transmission-node T-DER aggregation. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Graph-based representation of system physical profiles [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Architecture of the proposed PI-GCN framework [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Timeline of the proposed enhanced DF-based RTED. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: As shown in the figure, the testing loss trajectories 0 25 50 75 100 125 150 175 200 Training epoch 0.0 0.1 0.2 0.3 0.4 0.5 Loss Testing loss (a) 2383-bus system 0 25 50 75 100 125 150 175 200 Training epoch 0.0 0.1 0.2 0.3 0.4 0.5 Loss Testing loss (b) 3012-bus system [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 6
Figure 6. Figure 6: The normalized load curve over 8 days. The following strategies are considered for comparison: • BILEVEL: The bilevel model [27], which serves as the ideal benchmark for evaluation. • CONST: The fixed DFs, i.e., Fe = 1/|E(a)|, ∀e ∈ E(a); • MER: DFs are set via the most recent measurement data [2]; • KNN: The K-Nearest Neighbors (KNN)-based DF updating strategy in [27]. • ST-GCN: The proposed ST-GCN method … view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of smoothed rolling RTED operational costs for the 2383- [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
read the original abstract

The integration of distributed energy resources (DERs) into wholesale electricity markets, as mandated by FERC Order 2222, imposes new challenges on system operations. To remain consistent with existing market structures, regional transmission organizations (RTOs) have advanced the aggregation of transmission-node-level DERs (T-DERs), where a nodal virtual power plant (VPP) represents the mapping of all distribution-level DERs to their respective transmission nodes. This paper develops a real-time economic dispatch (RTED) framework that enables multi-transmission-node DER aggregation while addressing computational efficiency. To this end, we introduce a spatio-temporal graph convolutional network (ST-GCN) for adaptive prediction of distribution factors (DFs), thereby capturing the dynamic influence of individual T-DERs across the transmission system. Furthermore, an iterative constraint identification strategy is incorporated to alleviate transmission security constraints without compromising system reliability. Together, these innovations accelerate the market clearing process and support the effective participation of T-DER aggregators under current market paradigms. The proposed approach is validated on large-scale test systems, including modified 118-, 2383-, and 3012-bus networks under a rolling RTED setting with real demand data. Numerical results demonstrate significant improvements in reducing operational costs and maintaining transmission network feasibility, underscoring the scalability and practicality of the proposed framework.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes a real-time economic dispatch (RTED) framework for multi-transmission-node DER aggregation that uses a spatio-temporal graph convolutional network (ST-GCN) to adaptively predict distribution factors (DFs) capturing T-DER influence on the transmission network, combined with an iterative constraint identification strategy to reduce the number of security constraints while preserving reliability. The approach is claimed to accelerate market clearing and is validated on modified 118-, 2383-, and 3012-bus systems under rolling RTED with real demand data, with reported improvements in operational costs and maintained transmission feasibility.

Significance. If the quantitative validation holds, the work addresses a timely operational challenge arising from FERC Order 2222 by enabling scalable RTED with nodal VPP aggregations. The combination of graph-based learning for dynamic sensitivities and iterative constraint pruning targets computational bottlenecks in large-scale systems; the choice of large test networks and rolling-horizon setting with real data is a positive aspect that could support practical adoption if prediction accuracy and feasibility are rigorously demonstrated.

major comments (2)
  1. [Abstract] Abstract: the claim of 'significant improvements in reducing operational costs and maintaining transmission network feasibility' on the 118-/2383-/3012-bus cases is asserted without any numerical values, baseline comparisons, or error statistics. This absence directly weakens the ability to evaluate the central performance and reliability claims.
  2. [Results / Validation] Validation and results sections: the feasibility guarantee rests on the ST-GCN producing sufficiently accurate real-time DF predictions so that the subsequent dispatch plus iterative constraint identification keeps all transmission limits satisfied. Because DFs are linear sensitivities, even modest per-line prediction errors can propagate to violations when aggregated across T-DERs and rolled forward; the manuscript supplies no max/95-percentile DF error per line, no worst-case propagation analysis, and no robust or chance-constrained wrapper, leaving open the possibility that observed feasibility is an artifact of the particular training/test split rather than a general property.
minor comments (2)
  1. Provide the precise ST-GCN architecture (number of layers, temporal kernel sizes, graph construction for the transmission network) and the training loss used for DF regression.
  2. Clarify how the iterative constraint identification strategy selects which constraints to retain or drop at each RTED interval and whether any formal bound on the resulting optimality gap is derived.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough and constructive review of our manuscript. We have carefully addressed each major comment below and revised the manuscript to strengthen the presentation of results and validation.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim of 'significant improvements in reducing operational costs and maintaining transmission network feasibility' on the 118-/2383-/3012-bus cases is asserted without any numerical values, baseline comparisons, or error statistics. This absence directly weakens the ability to evaluate the central performance and reliability claims.

    Authors: We agree that the abstract would benefit from greater specificity to allow readers to immediately assess the performance claims. In the revised manuscript, we have updated the abstract to include key quantitative results from our experiments on the 118-, 2383-, and 3012-bus systems, including reported cost reductions relative to baselines and feasibility outcomes. revision: yes

  2. Referee: [Results / Validation] Validation and results sections: the feasibility guarantee rests on the ST-GCN producing sufficiently accurate real-time DF predictions so that the subsequent dispatch plus iterative constraint identification keeps all transmission limits satisfied. Because DFs are linear sensitivities, even modest per-line prediction errors can propagate to violations when aggregated across T-DERs and rolled forward; the manuscript supplies no max/95-percentile DF error per line, no worst-case propagation analysis, and no robust or chance-constrained wrapper, leaving open the possibility that observed feasibility is an artifact of the particular training/test split rather than a general property.

    Authors: We appreciate the referee's emphasis on rigorously quantifying prediction errors and their potential propagation. The original manuscript reports overall feasibility maintenance across rolling-horizon simulations with real data on three large test systems, but we acknowledge the value of more granular error statistics. In the revised version, we have added maximum and 95-percentile per-line DF prediction errors as well as a sensitivity analysis of error propagation under the rolling RTED setting. The iterative constraint identification strategy is intended to provide a practical safeguard; we have expanded the discussion to clarify how it mitigates the risk of violations. We have not added a chance-constrained wrapper, as that would represent a substantial change in problem formulation beyond the scope of the current learning-based approach. revision: partial

Circularity Check

0 steps flagged

No significant circularity: data-driven ST-GCN prediction of DFs is independent of central dispatch equations

full rationale

The paper proposes training an ST-GCN on external data to predict distribution factors for T-DERs, then feeding those predictions into a standard RTED formulation augmented by an iterative constraint identification loop. This structure does not reduce any claimed result to a self-definition, a fitted parameter renamed as prediction, or a self-citation chain; the DF predictions are learned from spatio-temporal inputs and validated on separate large-scale test cases (modified 118-, 2383-, and 3012-bus networks) with real demand data under rolling RTED. No uniqueness theorems or ansatzes from prior author work are invoked to force the architecture, and the feasibility claims rest on reported numerical outcomes rather than algebraic equivalence to the model's own inputs. The derivation chain therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The framework rests on standard power-system modeling assumptions for DER-to-transmission mapping and data-driven training of the neural network; no new physical entities are introduced.

free parameters (1)
  • ST-GCN model weights and hyperparameters
    Learned from training data to enable accurate prediction of distribution factors.
axioms (1)
  • domain assumption Distribution factors provide a sufficiently accurate linear mapping from distribution-level DER injections to transmission-node impacts.
    Invoked to justify the nodal VPP representation and aggregation approach.

pith-pipeline@v0.9.0 · 5784 in / 1286 out tokens · 45391 ms · 2026-05-18T02:45:29.169575+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    Participation of distributed energy resource aggregations in markets operated by regional transmission organizations and independent system operators,

    FERC, “Participation of distributed energy resource aggregations in markets operated by regional transmission organizations and independent system operators,” 2020. [Online]. Available: https: //www.ferc.gov/sites/default/files/2020-09/E-1 0.pdf

    work page 2020

  2. [2]

    Analyzing real-time market oscillation of demand response resource aggregation—a MISO case,

    Y . Liu et al., “Analyzing real-time market oscillation of demand response resource aggregation—a MISO case,”IEEE Trans. Energy Markets, Policy and Regulation, vol. 1, no. 4, pp. 455–467, 2023

    work page 2023

  3. [3]

    Wholesale market participation of DERAs: DSO-DERA- ISO coordination,

    C. Chen et al., “Wholesale market participation of DERAs: DSO-DERA- ISO coordination,”IEEE Trans. Power Syst., pp. 1–12, 2024

    work page 2024

  4. [4]

    Virtual power plant models and electricity markets-a review,

    N. Naval and J. M. Yusta, “Virtual power plant models and electricity markets-a review,”Renew. Sustain. Energy Rev., vol. 149, p. 111393, 2021

    work page 2021

  5. [5]

    Business practices manual: Markets and operations,

    MISO, “Business practices manual: Markets and operations,” 2022. [On- line]. Available: https://www.misoenergy.org/markets-and-operations/ markets-and-operations

    work page 2022

  6. [6]

    Integrating high DER-penetrated distribution systems into ISO energy market clearing: A feasible region projection approach,

    Y . Liu et al., “Integrating high DER-penetrated distribution systems into ISO energy market clearing: A feasible region projection approach,” IEEE Trans. Power Syst., vol. 36, no. 3, pp. 2262–2272, 2021

    work page 2021

  7. [7]

    Future resource studies in the MISO system,

    L. Wu et al., “Future resource studies in the MISO system,” 2020. [Online]. Available: https://cdn.misoenergy.org/Future%20Resource% 20Studies%20in%20the%20MISO%20System539337.pdf. 10 TABLE V HYPERPARAMETERS OFST-GCN ARCHITECTURE FORDF PREDICTION Module Layer type Parameter Value in_channel size 64 out_channel size 32 kernel size 12 in_channel size 32 out...

    work page 2020

  8. [8]

    Real-time joint regulations of frequency and voltage for tso-dso coordination: A deep reinforcement learning-based approach,

    R. Wang, S. Bu, and C. Y . Chung, “Real-time joint regulations of frequency and voltage for tso-dso coordination: A deep reinforcement learning-based approach,”IEEE Trans. Smart Grid, vol. 15, no. 2, pp. 2294–2308, 2024

    work page 2024

  9. [9]

    Aggregate flexibility of virtual power plants with temporal coupling constraints,

    S. Wang and W. Wu, “Aggregate flexibility of virtual power plants with temporal coupling constraints,”IEEE Trans. Smart Grid, vol. 12, no. 6, pp. 5043–5051, 2021

    work page 2021

  10. [10]

    Integrating high DER-penetrated distribution systems into ISO energy market clearing: A feasible region projection approach,

    Y . Liu, L. Wu, Y . Chen, and J. Li, “Integrating high DER-penetrated distribution systems into ISO energy market clearing: A feasible region projection approach,”IEEE Trans. Power Syst., vol. 36, no. 3, pp. 2262– 2272, 2021

    work page 2021

  11. [11]

    Strategic bidding for a virtual power plant in the day-ahead and real-time markets: A price-taker robust optimization approach,

    M. Rahimiyan and L. Baringo, “Strategic bidding for a virtual power plant in the day-ahead and real-time markets: A price-taker robust optimization approach,”IEEE Trans. Power Syst., vol. 31, no. 4, pp. 2676–2687, 2015

    work page 2015

  12. [12]

    Optimal offering strategy of a virtual power plant: A stochastic bi-level approach,

    E. Kardakos et al., “Optimal offering strategy of a virtual power plant: A stochastic bi-level approach,”IEEE Trans. Smart Grid, vol. 7, no. 2, pp. 794–806, 2016

    work page 2016

  13. [13]

    Optimal bidding strategy of DER aggregator considering dual uncertainty via information gap decision theory,

    X. Lu et al., “Optimal bidding strategy of DER aggregator considering dual uncertainty via information gap decision theory,”IEEE Trans. Ind. Appl., vol. 57, no. 1, pp. 158–169, 2021

    work page 2021

  14. [14]

    On the efficiency of connection charges—part I: A stochastic framework,

    D. Mu ˜noz- ´Alvarez, “On the efficiency of connection charges—part I: A stochastic framework,”IEEE Trans. Power Syst., vol. 33, no. 4, pp. 3822–3833, 2018

    work page 2018

  15. [15]

    On the efficiency of connection charges—part II: Integration of distributed energy resources,

    D. Mu ˜noz- ´Alvarez et al., “On the efficiency of connection charges—part II: Integration of distributed energy resources,”IEEE Trans. Power Syst., vol. 33, no. 4, pp. 3834–3844, 2018

    work page 2018

  16. [16]

    A network-aware distributed energy resource aggrega- tion framework for flexible, cost-optimal, and resilient operation,

    K. Utkarsh et al., “A network-aware distributed energy resource aggrega- tion framework for flexible, cost-optimal, and resilient operation,”IEEE Trans. Smart Grid, vol. 13, no. 2, pp. 1213–1224, 2022

    work page 2022

  17. [17]

    Grid-aware aggregation and realtime disaggregation of distributed energy resources in radial networks,

    N. Nazir et al., “Grid-aware aggregation and realtime disaggregation of distributed energy resources in radial networks,”IEEE Trans. Power Syst., vol. 37, no. 3, pp. 1706–1717, 2022

    work page 2022

  18. [18]

    Aggregate operation model for numerous small-capacity distributed energy resources considering uncertainty,

    Z. Yi et al., “Aggregate operation model for numerous small-capacity distributed energy resources considering uncertainty,”IEEE Trans. Smart Grid, vol. 12, no. 5, pp. 4208–4224, 2021

    work page 2021

  19. [19]

    Aggregation and disaggregation of energetic flexibility from distributed energy resources,

    F. M ¨uller et al., “Aggregation and disaggregation of energetic flexibility from distributed energy resources,”IEEE Trans. Smart Grid, vol. 10, no. 2, pp. 1205–1214, 2019

    work page 2019

  20. [20]

    Towards optimality preserving aggregation for scheduling distributed energy resources,

    R. R. Appino et al., “Towards optimality preserving aggregation for scheduling distributed energy resources,”IEEE Trans. Control Netw. Syst., vol. 8, no. 3, pp. 1477–1488, 2021

    work page 2021

  21. [21]

    An analysis of zonal electricity pricing from a long-term perspective,

    Q. L ´et´e, Y . Smeers, and A. Papavasiliou, “An analysis of zonal electricity pricing from a long-term perspective,”Energy Econ., vol. 107, p. 105853, 2022

    work page 2022

  22. [22]

    On the effects of storage facilities on optimal zonal pricing in electricity markets,

    M. Weibelzahl and A. M ¨artz, “On the effects of storage facilities on optimal zonal pricing in electricity markets,”Energy Policy, vol. 113, pp. 778–794, 2018

    work page 2018

  23. [23]

    Zonal price analysis of the italian whole- sale electricity market,

    A. Gianfreda and L. Grossi, “Zonal price analysis of the italian whole- sale electricity market,” in2009 6th International Conference on the European Energy Market, 2009, pp. 1–6

    work page 2009

  24. [24]

    Grid-aware flexibility aggregation for zonal balancing markets,

    E. Karangelos and A. Papavasiliou, “Grid-aware flexibility aggregation for zonal balancing markets,”Electr. Power Syst. Res., vol. 234, p. 110715, 2024

    work page 2024

  25. [25]

    Impact of FERC Order 2222 on DER participation rules in US electricity markets,

    B. C. Eldridge and A. Somani, “Impact of FERC Order 2222 on DER participation rules in US electricity markets,” Pacific Northwest National Laboratory, Richland, W A, Tech. Rep., 2022

    work page 2022

  26. [26]

    Real-time market-to-market oscillation management,

    Y . Yang et al., “Real-time market-to-market oscillation management,” in 2021 IEEE PES General Meeting, 2021, pp. 1–5

    work page 2021

  27. [27]

    Real-time economic dispatch approach for wholesale energy market with multi- transmission-node DER aggregation,

    Z. Shao, W. Wang, B. Eldridge, A. Somani, and L. Wu, “Real-time economic dispatch approach for wholesale energy market with multi- transmission-node DER aggregation,”IEEE Trans. Power Syst., vol. 40, no. 4, pp. 3185–3197, 2025

    work page 2025

  28. [28]

    Multi-transmission node der aggregation: Chance-constrained unit commitment with bounded hetero-dimensional mixture model for uncertain distribution factors,

    W. Wang, Z. Shao, Y . Liu, B. Eldridge, A. Somani, J. T. Holzer, and L. Wu, “Multi-transmission node der aggregation: Chance-constrained unit commitment with bounded hetero-dimensional mixture model for uncertain distribution factors,”IEEE Trans. Power Syst., pp. 1–13, 2025

    work page 2025

  29. [29]

    Participation of DERs at transmission level: FERC Order No.2222 and TSO-DSO coordination,

    R. Trivedi et al., “Participation of DERs at transmission level: FERC Order No.2222 and TSO-DSO coordination,” in2023 IEEE PES ISGT Conference - Middle East, 2023, pp. 1–5

    work page 2023

  30. [30]

    Piecewise linear approxima- tion of generators cost functions using max-affine functions,

    H. Ahmadi, J. R. Mart ´ı, and A. Moshref, “Piecewise linear approxima- tion of generators cost functions using max-affine functions,” in2013 IEEE Power & Energy Society General Meeting, 2013, pp. 1–5

    work page 2013

  31. [31]

    Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting,

    B. Yu, H. Yin, and Z. Zhu, “Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting,” inProc. 27th Int. Joint Conf. Artif. Intell.AAAI Press, 2018, p. 3634–3640

    work page 2018

  32. [32]

    Semi-supervised learning with graph learning-convolutional networks,

    B. Jiang, Z. Zhang, D. Lin, J. Tang, and B. Luo, “Semi-supervised learning with graph learning-convolutional networks,” inProceedings of the IEEE CVF conference on computer vision and pattern recognition, 2019, pp. 11 313–11 320

    work page 2019

  33. [33]

    Convolutional neural networks on graphs with fast localized spectral filtering,

    M. Defferrard, X. Bresson, and P. Vandergheynst, “Convolutional neural networks on graphs with fast localized spectral filtering,”Advances in neural information processing systems, vol. 29, 2016

    work page 2016

  34. [34]

    Transmission constraint filtering in large-scale security-constrained unit commitment,

    A. S. Xavier, F. Qiu, F. Wang, and P. R. Thimmapuram, “Transmission constraint filtering in large-scale security-constrained unit commitment,” IEEE Trans. Power Syst., vol. 34, no. 3, pp. 2457–2460, 2019

    work page 2019

  35. [35]

    MATPOWER: Steady-state operations, plan- ning, and analysis tools for power systems research and education,

    R. D. Zimmerman et al., “MATPOWER: Steady-state operations, plan- ning, and analysis tools for power systems research and education,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 12–19, 2010

    work page 2010

  36. [36]

    Open Power System Data Platform - European Power System Data,

    work page

  37. [37]

    Available: https://data.open-power-system-data.org/

    [Online]. Available: https://data.open-power-system-data.org/

    work page