Pith. sign in

REVIEW 4 major objections 7 minor 41 references

Coordinated Power Smoothing Control for Wind Storage Integrated System with Physics-informed Deep Reinforcement Learning

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A hierarchical wind-storage controller claims an 11% profit gain and a 19% cut in grid power fluctuation.

desk verdict A sensible bi-level MARL framework for wind-storage smoothing, but the PINN part is under-specified and one update equation is wrong as written, so the headline gains are not yet credible. read the letter →

arxiv 2412.17838 v1 pith:F4TLUJG2 submitted 2024-12-17 eess.SY cs.AIcs.SY

Shuyi Wang , Huan Zhao , Yuji Cao , Zibin Pan , Guolong Liu , Gaoqi Liang , Junhua Zhao This is my paper · ORCID
classification eess.SYcs.AIcs.SY
keywords windstorageintegratedsystemspowersmoothingcontrolmulti-agentdeepreinforcementlearningphysics-informedneuralnetworkbi-levelMarkovdecisionprocessbatterydegradationcostwakeeffectWFSimsimulation
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 show that power smoothing for a wind farm with battery storage is best handled by splitting the controller into two levels that act at different frequencies: an upper agent adjusts turbine axial induction factors every f minutes, and a lower agent commands battery charge and discharge every minute. The authors argue that this bi-level Markov decision process, trained with multi-agent deep deterministic policy gradient and a physics-informed loss term for power fluctuation, outperforms a model predictive control baseline in a wind farm simulator. If the results hold, the same stored energy can simultaneously earn about 11% more total profit and feed the grid with about 19% smaller power fluctuations, which matters because grid operators cap or penalize minute-scale wind power swings.

What carries the argument

The carrying object is PAMA-DDPG, a multi-agent deep deterministic policy gradient algorithm organized as a bi-level Markov decision process. The upper-level agent outputs axial induction factors for the turbines on a slow time scale, and the lower-level agent outputs battery power every minute; both actors share critics during centralized training and execute with local observations. The distinctive mechanism is a physics-informed loss term derived from the ideal-smoothing differential equation $\partial P_G^t / \partial t = 0$, rewritten through the power-balance equation $P_W^t - P_B^t = P_G^t$ as $\partial P_B^t / \partial t + N(P_B^t) = 0$, so the actor's gradient includes a weighted penalty on the squared residual of that equation. This constrains the feasible policy space and is what the paper credits for faster convergence and better final performance, alongside the Jensen Park wake model and the linear degradation cost $K_{\text{deg}}$ in the reward.

What would settle it

Run the trained PAMA-DDPG policy on wind sequences the agent never saw in training, or in a higher-fidelity large-eddy simulation of the same wind farm, and compare grid-power fluctuation with the MPC baseline; if the roughly 19% reduction does not appear, the central claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that including both the wind wake effect and the battery degradation cost in the environment model, and then embedding the power-smoothing dynamics directly into the policy network's loss, yields a controller that beats traditional methods on both economics and grid reliability. On average over four one-day wind scenarios in WFSim, the proposed PAMA-DDPG algorithm raises total profit by roughly 11%, lowers the fluctuation severity FS by about 19%, and reduces fluctuation-threshold violations by about 22% relative to the MPC baseline. The paper also claims that removing either the wake model or the degradation model degrades all three metrics, and that the physics-informed loss accelerates convergence from roughly 5e4 to 4e4 iterations while reaching a higher converged reward.

Load-bearing premise

The load-bearing premise is that the WFSim environment, with its Jensen Park wake model and linear battery degradation parameters, behaves enough like a real wind-storage system that the simulated 11% profit gain and 19% fluctuation reduction will carry over to physical deployment.

Editorial extensions

If this is right

  • A wind-storage controller can be trained to act at two different frequencies, turbines every f minutes and battery every minute, without losing either economic or smoothing performance.
  • Adding wake and battery-degradation physics to the training environment is worth its complexity: omitting either model drops profit by 9 to 15 percent and increases fluctuation severity by 7 to 35 percent in the paper's tests.
  • The physics-informed loss can shorten training: convergence drops from about 5e4 to 4e4 iterations while the converged reward and profit improve.
  • The improvement is not limited to one wind regime: the four test scenarios cover low, moderate, and high wind, and the relative gains over MPC persist across them.

Reading between the lines

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

  • Our inference: the ideal-smoothing constraint $\partial P_G^t / \partial t = 0$ makes the lower-level agent chase a flat power trajectory, so a natural extension is to replace it with a time-varying reference slope, which could reduce battery cycling during wind ramps while still respecting the fluctuation threshold.
  • Our inference: the bi-level formulation should transfer to other hybrid systems with mismatched response rates, such as photovoltaic plants paired with storage or wind paired with flywheels, because the slow-fast actor split is independent of the specific wake model.
  • Our inference: since the degradation cost is linear and calibrated to a single real-world battery application, the charge-discharge trade-off might shift under nonlinear degradation or aging-aware cost models, and testing that variant would directly probe the framework's limits.
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

4 major / 7 minor

Summary. This paper proposes a bi-level coordinated control scheme for a wind-storage integrated system (WSIS). The upper-level controller sets the axial induction factors of the wind turbines once per f-minute interval under a Jensen Park wake model, while the lower-level controller sets BESS charging/discharging power every minute under a battery energy and degradation cost model. The problem is cast as a bi-level MDP and solved with a multi-agent DDPG algorithm, augmented by a physics-informed loss term derived from the condition ∂P_G/∂t = 0 (the PAMA-DDPG algorithm). Experiments in WFSim with real wind data across four one-day scenarios report roughly 11% higher total profit, 19% lower fluctuation severity (FS), and 22% fewer threshold violations (VO) relative to an MPC baseline, as well as faster convergence than MA-DDPG (4e4 vs 5e4 iterations) on a fixed five-minute wind sequence.

Significance. If the reported results are reproducible, the paper would make a useful practical contribution: the bi-level treatment of the different control frequencies of turbines and batteries is well motivated, the integration of wake and battery-degradation models into one RL framework goes beyond earlier works such as refs. [9], [11], and [19], and the four-scenario ablations in Section 4.3 directly test the value of model completeness. The headline percentages are internally consistent with the reported base values (Table 2, β1=5: $933.01 vs $840.55, 15.39 MW vs 19.01 MW, 18 vs 23). The robustness sweeps over β1 and η in Tables 2-3 are a useful step, though limited. However, the defining algorithmic novelty—the physics-informed loss—is under-specified and, as written in Eq. (40), not implementable; the 'physics' prior in Eq. (35) duplicates the smoothing objective; and the main quantitative claims carry no uncertainty quantification. These issues are load-bearing for the paper's central narrative, so the result is promising but not yet established.

major comments (4)
  1. [3.4, Eqs. (35)-(37)] The operator N in Eq. (36) is never defined, so the residual f(t, s_L) in Eq. (37) and the loss L_f in Eq. (39) cannot be computed from the text. Combining Eq. (17) (P_G = P_W − P_B) with Eq. (35) (∂P_G/∂t = 0) gives ∂P_B/∂t = ∂P_W/∂t, which forces N(P_B) = −∂P_W/∂t as the only consistent choice. This is a substantive gap rather than a typo: as the paper itself notes in Section 3.2.2, P_W varies with wind speed even when the induction factors are held fixed, so the missing N changes the value of the loss. Because the PINN loss is the paper's third contribution and the basis of the Section 4.4 acceleration claim, the definition of N and the computation of ∂P_B/∂t must be stated explicitly.
  2. [3.4, Eq. (40)] Eq. (40) adds a scalar ω_f·|f(t,s)|² to the gradient vector ω_u·∇_a Q ∇_θμ μ inside the expectation, which is a type error; the correct term is ω_f·∇_θμ |f|², requiring the gradient of the physics residual with respect to the actor parameters. The paper also does not specify how ∂P_B/∂t is discretized (e.g., finite difference over which consecutive transitions) or how mini-batch sampling provides those consecutive transitions. As written, the update cannot be executed, and the claimed 4e4-iteration convergence of PAMA-DDPG cannot be attributed to the proposed physics-informed mechanism.
  3. [3.4, Eq. (35) and the paragraph after Eq. (40)] The 'physics' prior is not an independent physical law: Eq. (35) (∂P_G/∂t = 0) is the zero-violation limit of the smoothing constraint (19), i.e., the control objective itself. This is confirmed by the paper's own statement that no separate computation of P_B or N(P_B) is needed because P_G is already computed: the residual f reduces to a function of P_G. The physics-informed framing is therefore overstated, and the faster-convergence evidence in Section 4.4 cannot distinguish the PINN mechanism from simply regularizing the actor update toward the smoothing objective. Please either reframe the term as a regularizer on the control objective, or add an experiment that compares the PINN loss against an equivalent reward-based shaping term.
  4. [4.2 and 4.4] The headline quantitative claims (+11% profit, −19% FS, −22% VO) are point estimates from four deterministic scenarios with no multiple-seed runs, error bars, or statistical tests. The only variance reporting in the paper is in Section 4.4, and it is restricted to a single fixed five-minute wind sequence, where the PAMA-DDPG advantage over MA-DDPG is small (reward 15.57 vs 14.69; profit $8.18 vs $7.49). Please report means and standard deviations over several independent training runs for the main four-scenario comparison before the improvements over the MPC baseline are claimed.
minor comments (7)
  1. [Algorithm 1] Line 18 of Algorithm 1 updates 'θ_L of the lower level' in the upper-level branch; this should be θ_U. Line 12 also refers to a reward r_H^t that is not defined anywhere (presumably r_U^t).
  2. [4.4] The sentence 'The average time for the two algorithms to choose an action and update the policy is 1e-8 and 1e-2 seconds, respectively' is implausible for a deep-network forward pass (1e-8 s is ten nanoseconds), and the mapping from the two times to MA-DDPG/PAMA-DDPG is ambiguous. Please correct the orders of magnitude and clarify which quantity refers to which algorithm.
  3. [Table 1] The PINN loss weights ω_u and ω_f in Eq. (38), and the penalty coefficients β and ν in Eqs. (22)-(23), do not appear in the hyperparameter table. Without these values the reported results are not reproducible.
  4. [3.4] The statement that 'the policy network μ_L and physics-informed network f(t, s_L) have the same parameters, while the activation functions are different due to the operator N' is unclear: a differential operator does not change activation functions, and the shared-parameter architecture requires a precise description or a diagram.
  5. [3.2.2 / 4.1] The electricity price Pr_t is a constant in all experiments (Section 4.1) yet appears in the state of both agents; either add price-varying scenarios or justify retaining this state dimension.
  6. [4.6] The sentence 'it is more likely to charge than to discharge, as charging behavior incurs degradation costs' contradicts Eq. (13), in which the degradation cost is incurred on discharging (via max(−P_B, 0)). Please correct the direction of the argument.
  7. [4.1 / Fig. 6] The caption of Fig. 6 uses 'Senario' instead of 'Scenario'; Eq. (26) uses P_i^{t′} for turbine power while the variable was defined as P_turbine,i in Eqs. (1) and (16); and Section 4.1 should state explicitly how WFSim, the Jensen Park model, and the ADM interact in the simulation environment.

Circularity Check

1 steps flagged · score 4.0 of 10

The 'physics' in PINN is the smoothing objective: Eq. (35) sets dP_G/dt=0, exactly the quantity Eqs. (18) and (22)-(24) penalize; the 11%/19% MPC comparisons are otherwise independent.

  1. self definitional [Section 3.4, Eqs. (35)-(40)]
    "Assume that the power transferred to the grid is invariant to simulate the ideal dynamic system. Thus, the changes in power PtG remain zero over time t. The invariable physical equation about the optimal power smoothing objective is derived as: ∂PtG/∂t = 0"

    Equation (35) is the smoothing objective itself, not an independent physical law: the lower-level reward (28) and penalty (22)-(24) are built on P_FG=|P_G^{t-1}-P_G^t|, so ∂P_G/∂t=0 says exactly that grid power should be constant. The reformulation (36) defines N(P_B) only implicitly; since P_G=P_W-P_B (Eq. 17), encoding ∂P_G/∂t=0 through f in (37) requires N(P_B)=-∂P_W/∂t, which makes f(t,s_L)=-∂P_G/∂t. Therefore L_f in (39) is a mean-square penalty on the quantity the controller is already rewarded to minimize, and the PINN speedup claimed in Sec. 4.4 is objective injection/reward shaping rather than a first-principles constraint discovered from the system.

full rationale

The main numerical claims (about 11% higher profit and 19% lower fluctuation vs MPC) are not circular: they are computed by running PAMA-DDPG, MA-DDPG, single-agent DDPG, and MPC on the WFSim simulator in Section 4.2, with wake and battery-degradation parameters taken from external sources (Jensen Park [31], Rwanda battery data [34], WFSim [38]). No parameter is fitted to the reported 11%/19% numbers, and no load-bearing uniqueness theorem or self-citation chain is used; self-citations [17] and [33] appear only as background. The circularity is confined to the PINN component of Section 3.4. Equation (35) imposes ∂P_G/∂t=0, which is not a physical invariant but the literal power-smoothing objective already encoded in Eq. (18) and the lower-level reward. The operator N in (36) is never specified, but to make (36) equivalent to (35) one must set N(P_B)=-∂P_W/∂t, so f=-∂P_G/∂t and L_f in (39) penalizes the objective itself. Consequently, the faster convergence attributed to 'physics information' in Section 4.4 is a reward-shaping effect, not a first-principles prior. Separately, Eq. (40) is a correctness defect rather than circularity: it adds the scalar ω_f|f|^2 to a gradient vector, so the actual ∇_{θμ}|f|^2 term is missing. These issues weaken the distinctiveness of the PINN contribution, but they do not make the headline MPC comparison circular.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new physical entities. The main imported assumptions are the wake model, the actuator disk model, the battery degradation model, and WFSim. The most paper-specific assumption is the invariance condition dP_G/dt=0 used for the PINN loss, which is an ad hoc restatement of the smoothing objective. The free parameters include several hand-chosen penalty coefficients and the unreported PINN loss weights.

free parameters (3)
  • Penalty coefficients beta1, beta2, kappa = beta1=5, beta2=10, kappa=10
    Hand-chosen weights in the reward functions, Eqs. (26) and (28), listed in Table 1. Only beta1 is subject to a robustness test; the others are fixed without sensitivity analysis.
  • PINN loss weights omega_u, omega_f = not reported
    Weights that balance the data and physics loss terms in Eq. (38) are never given a value in the paper, yet the PINN acceleration claim depends on them.
  • Violation penalty coefficients beta, nu = not reported
    Introduced in Eqs. (22)-(23) for the lower-level objective, these coefficients are not specified in Table 1 or the text.
assumptions (6)
  • domain assumption The Jensen Park wake model accurately describes inter-turbine wakes in the wind farm.
    Used in Eqs. (5)-(6) and embedded in WFSim; the wake model directly affects the computed wind power and therefore the claimed profit improvement.
  • domain assumption The Actuator Disk Model Cp(alpha) = 4 alpha (cos phi_y - alpha)^2 is a valid turbine power model.
    Eq. (3) is taken from standard wind turbine literature and is not re-validated in this paper.
  • domain assumption The linear battery degradation model from reference [34] correctly estimates degradation costs.
    Eqs. (10)-(13) are imported from a prior paper; the numerical results depend on the degradation cost term Kdeg.
  • ad hoc to paper The optimal power smoothing policy can be represented by the invariant condition dP_G/dt = 0.
    Section 3.4 introduces this condition as the physics prior, but it is actually the control objective, not an independently established physical law.
  • ad hoc to paper Wind turbines and the BESS operate at control frequencies of 1/f and 1 per minute, respectively.
    Section 3.1 sets this timescale separation as the basis of the bi-level formulation; it is a modeling choice rather than an observed constraint.
  • standard math The decision process satisfies the Markov property and can be formulated as an MDP.
    Section 3.2 reformulates the problem as a Markov decision process; standard assumption for reinforcement learning, though partially violated by the bi-level dependence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Coordinated Power Smoothing Control for Wind Storage Integrated System with Physics-informed Deep Reinforcement Learning." pith.science (2026). https://pith.science/paper/F4TLUJG2

@misc{pith2026241217838,
  author       = {Pith},
  title        = {Pith review of: Coordinated Power Smoothing Control for Wind Storage Integrated System with Physics-informed Deep Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F4TLUJG2}},
  note         = {Machine review of arXiv:2412.17838}
}
read the original abstract

The Wind Storage Integrated System with Power Smoothing Control (PSC) has emerged as a promising solution to ensure both efficient and reliable wind energy generation. However, existing PSC strategies overlook the intricate interplay and distinct control frequencies between batteries and wind turbines, and lack consideration of wake effect and battery degradation cost. In this paper, a novel coordinated control framework with hierarchical levels is devised to address these challenges effectively, which integrates the wake model and battery degradation model. In addition, after reformulating the problem as a Markov decision process, the multi-agent reinforcement learning method is introduced to overcome the bi-level characteristic of the problem. Moreover, a Physics-informed Neural Network-assisted Multi-agent Deep Deterministic Policy Gradient (PAMA-DDPG) algorithm is proposed to incorporate the power fluctuation differential equation and expedite the learning process. The effectiveness of the proposed methodology is evaluated through simulations conducted in four distinct scenarios using WindFarmSimulator (WFSim). The results demonstrate that the proposed algorithm facilitates approximately an 11% increase in total profit and a 19% decrease in power fluctuation compared to the traditional methods, thereby addressing the dual objectives of economic efficiency and grid-connected energy reliability.

Figures

Figures reproduced from arXiv: 2412.17838 by the authors.

Figure 1
Figure 1. The structure of the wind storage integrated system. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The coordinated bi-level control framework. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. The workflow of the MA-DDPG. dynamics of power fluctuation for optimal policy, the PINN-based method is introduced to enhance the efficiency of the training process of the multi-agent DRL. For the physical models of a general dynamic system, PINN incorporates its physical constraints into the neural network via differential equations involving time and other physical inputs. The general form [37] can be expressed as… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The wind turbine structural system. computational complexity. In practical applications, the wind turbine with three rotor blades is typically positioned upwind of the tower and the nacelle, as depicted in [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: One-day wind speed profile for 4 testing scenarios. [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Relative metrics of 4 methods in various scenarios. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Relative metrics of different environment model completeness in various scenarios. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Learning process of the proposed approach. [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: BESS control results obtained by the proposed PAMA-DDPG over 100 minutes. [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 35 canonical work pages

  1. [9]

    X. Wang, J. Zhou, B. Qin, L. Guo, Coordinated power smoothing control strategy of multi-wind turbines and energy storage systems in wind farm based on madrl, IEEE Transactions on Sustainable Energy (2023)

    work page 2023

  2. [11]

    X. Wang, J. Zhou, B. Qin, L. Guo, Coordinated control of wind turbine and hybrid energy storage system based on multi-agent deep reinforcement learning for wind power smoothing, Journal of Energy Storage 57 (2023) 106297

    work page 2023

  3. [19]

    J. Yang, M. Yang, M. Wang, P. Du, Y . Yu, A deep reinforcement learning method for managing wind farm uncertainties through energy storage system control and external reserve purchasing, International Journal of Electrical Power & Energy Systems 119 (2020) 105928

    work page 2020

  4. [1]

    L. Yin, W. Ding, Multi-objective high-dimensional multi-fractional-order optimization algorithm for multi-objective high-dimensional multi- fractional-order optimization controller parameters of doubly-fed induction generator-based wind turbines, Engineering Applications of Ar- tificial Intelligence 126 (2023) 106929

    work page 2023

  5. [2]

    Shivashankar, S

    S. Shivashankar, S. Mekhilef, H. Mokhlis, M. Karimi, Mitigating methods of power fluctuation of photovoltaic (pv) sources–a review, Renewable and Sustainable Energy Reviews 59 (2016) 1170–1184

    work page 2016

  6. [3]

    X. Tang, M. Yin, C. Shen, Y . Xu, Z. Y . Dong, Y . Zou, Active power control of wind turbine generators via coordinated rotor speed and pitch angle regulation, IEEE Transactions on Sustainable Energy 10 (2) (2018) 822–832

    work page 2018

  7. [4]

    Y . Kim, M. Kang, E. Muljadi, J.-W. Park, Y . C. Kang, Power smoothing of a variable-speed wind turbine generator in association with the rotor-speed-dependent gain, IEEE Transactions on Sustainable Energy 8 (3) (2016) 990–999

    work page 2016

  8. [5]

    Uehara, A

    A. Uehara, A. Pratap, T. Goya, T. Senjyu, A. Yona, N. Urasaki, T. Funabashi, A coordinated control method to smooth wind power fluctuations of a pmsg-based wecs, IEEE Transactions on energy conversion 26 (2) (2011) 550–558

    work page 2011

Show all 41 references
  1. [6]

    Barra, W

    P. Barra, W. De Carvalho, T. Menezes, R. Fernandes, D. Coury, A review on wind power smoothing using high-power energy storage systems, Renewable and Sustainable Energy Reviews 137 (2021) 110455

    2021

  2. [7]

    P. Lu, N. Zhang, L. Ye, E. Du, C. Kang, Advances in model predictive control for large-scale wind power integration in power systems: A comprehensive review, Advances in Applied Energy (2024) 100177

    2024

  3. [8]

    Zhang, J

    H. Zhang, J. Yang, X. Ren, Q. Wu, D. Zhou, E. Elahi, How to accommodate curtailed wind power: A comparative analysis between the us, germany, india and china, Energy Strategy Reviews 32 (2020) 100538. 18

    2020

  4. [10]

    Huang, Q

    S. Huang, Q. Wu, Y . Guo, F. Rong, Hierarchical active power control of dfig-based wind farm with distributed energy storage systems based on admm, IEEE Transactions on Sustainable Energy 11 (3) (2019) 1528–1538

    2019

  5. [12]

    Xiong, S

    L. Xiong, S. Yang, S. Huang, D. He, P. Li, M. W. Khan, J. Wang, Optimal allocation of energy storage system in dfig wind farms for frequency support considering wake effect, IEEE Transactions on Power Systems 37 (3) (2021) 2097–2112

    2021

  6. [13]

    M. A. Ahmad, M. R. Hao, R. M. T. R. Ismail, A. N. K. Nasir, Model-free wind farm control based on random search, in: 2016 IEEE international conference on automatic control and intelligent systems (I2CACIS), IEEE, 2016, pp. 131–134

    2016

  7. [14]

    H. Dong, M. Edrah, X. Zhao, M. Collu, X. Xu, K. Abhinav, Z. Lin, Model-free semi-active structural control of floating wind turbines, in: 2020 Chinese Automation Congress (CAC), IEEE, 2020, pp. 4216–4220

    2020

  8. [15]

    R. S. Sutton, A. G. Barto, Reinforcement learning: An introduction, MIT press, 2018

    2018

  9. [16]

    Y . Cao, H. Zhao, Y . Cheng, T. Shu, G. Liu, G. Liang, J. Zhao, Y . Li, Survey on large language model-enhanced reinforcement learning: Concept, taxonomy, and methods, arXiv preprint arXiv:2404.00282 (2024)

    2024 arXiv

  10. [17]

    H. Zhao, J. Zhao, J. Qiu, G. Liang, Z. Y . Dong, Cooperative wind farm control with deep reinforcement learning and knowledge-assisted learning, IEEE Transactions on Industrial Informatics 16 (11) (2020) 6912–6921

    2020

  11. [18]

    J. Zhu, W. Hu, X. Xu, H. Liu, L. Pan, H. Fan, Z. Zhang, Z. Chen, Optimal scheduling of a wind energy dominated distribution network via a deep reinforcement learning approach, Renewable Energy 201 (2022) 792–801

    2022

  12. [20]

    Bus ¸oniu, R

    L. Bus ¸oniu, R. Babu ˇska, B. De Schutter, Multi-agent reinforcement learning: An overview, Innovations in multi-agent systems and applications-1 (2010) 183–221

    2010

  13. [21]

    L. Feng, Y . Xie, B. Liu, S. Wang, Multi-level credit assignment for cooperative multi-agent reinforcement learning, Applied Sciences 12 (14) (2022) 6938

    2022

  14. [22]

    T. I. Dada, P. Thodoroff, N. D. Lawrence, Application of multi-agent reinforcement learning for battery management in renewable mini-grids, in: AAAI-22 Workshop on Machine Learning for Operations Research (ML4OR), 2021

    2021

  15. [23]

    G. E. Karniadakis, I. G. Kevrekidis, L. Lu, P. Perdikaris, S. Wang, L. Yang, Physics-informed machine learning, Nature Reviews Physics 3 (6) (2021) 422–440

    2021

  16. [24]

    Huang, J

    B. Huang, J. Wang, Applications of physics-informed neural networks in power systems-a review, IEEE Transactions on Power Systems 38 (1) (2022) 572–588

    2022

  17. [25]

    Stiasny, S

    J. Stiasny, S. Chevalier, S. Chatzivasileiadis, Learning without data: Physics-informed neural networks for fast time-domain simulation, in: 2021 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm), IEEE, 2021, p...

    2021

  18. [26]

    Chakraborty, Transfer learning based multi-fidelity physics informed deep neural network, Journal of Computational Physics 426 (2021) 109942

    S. Chakraborty, Transfer learning based multi-fidelity physics informed deep neural network, Journal of Computational Physics 426 (2021) 109942

    2021

  19. [27]

    Banerjee, K

    C. Banerjee, K. Nguyen, C. Fookes, M. Raissi, A survey on physics informed reinforcement learning: Review and open problems, arXiv preprint arXiv:2309.01909 (2023)

    2023 arXiv

  20. [28]

    J. Gao, S. Chen, X. Li, J. Zhang, Transient voltage control based on physics-informed reinforcement learning, IEEE Journal of Radio Frequency Identification 6 (2022) 905–910

    2022

  21. [29]

    Mikkelsen, et al., Actuator disc methods applied to wind turbines, Ph.D

    R. Mikkelsen, et al., Actuator disc methods applied to wind turbines, Ph.D. thesis, PhD thesis, Technical University of Denmark (2003)

    2003

  22. [30]

    Annoni, P

    J. Annoni, P. M. Gebraad, A. K. Scholbrock, P. A. Fleming, J.-W. v. Wingerden, Analysis of axial-induction-based wind plant control using an engineering and a high-order wind plant model, Wind Energy 19 (6) (2016) 1135–1150

    2016

  23. [31]

    Katic, J

    I. Katic, J. Højstrup, N. O. Jensen, A simple model for cluster e fficiency, in: European wind energy association conference and exhibition, V ol. 1, A. Raguzzi Rome, Italy, 1986, pp. 407–410

    1986

  24. [32]

    H. Li, Z. Wan, H. He, Constrained ev charging scheduling based on safe deep reinforcement learning, IEEE Transactions on Smart Grid 11 (3) (2019) 2427–2439

    2019

  25. [33]

    H. Zhao, Z. Liu, X. Mai, J. Zhao, J. Qiu, G. Liu, Z. Y . Dong, A. M. Ghias, Mobile battery energy storage system control with knowledge- assisted deep reinforcement learning, Energy Conversion and Economics 3 (6) (2022) 381–391

    2022

  26. [34]

    Bordin, H

    C. Bordin, H. O. Anuta, A. Crossland, I. L. Gutierrez, C. J. Dent, D. Vigo, A linear programming approach for battery degradation analysis and optimization in offgrid power systems with solar energy integration, Renewable Energy 101 (2017) 417–430

    2017

  27. [35]

    Cortina, V

    G. Cortina, V . Sharma, M. Calaf, Investigation of the incoming wind vector for improved wind turbine yaw-adjustment under di fferent atmospheric and wind farm conditions, Renewable Energy 101 (2017) 376–386

    2017

  28. [36]

    Van Hasselt, A

    H. Van Hasselt, A. Guez, D. Silver, Deep reinforcement learning with double q-learning, in: Proceedings of the AAAI conference on artificial intelligence, V ol. 30, 2016

    2016

  29. [37]

    Raissi, P

    M. Raissi, P. Perdikaris, G. E. Karniadakis, Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, Journal of Computational physics 378 (2019) 686–707

    2019

  30. [38]

    Boersma, P

    S. Boersma, P. Gebraad, M. Vali, B. Doekemeijer, J. Van Wingerden, A control-oriented dynamic wind farm flow model:“wfsim”, in: Journal of Physics: Conference Series, V ol. 753, IOP Publishing, 2016, p. 032005

    2016

  31. [39]

    URL https://www.aedb.org/wind-mast-data

    Wind mast data, alternative Energy Development Board. URL https://www.aedb.org/wind-mast-data

  32. [40]

    Ma, On-grid electricity tari ffs in china: Development, reform and prospects, Energy policy 39 (5) (2011) 2633–2645

    J. Ma, On-grid electricity tari ffs in china: Development, reform and prospects, Energy policy 39 (5) (2011) 2633–2645

    2011

  33. [41]

    rep., National Energy Administration (2017)

    Technical specifications for active power regulation and control of wind farm, Tech. rep., National Energy Administration (2017). 19

    2017

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.