東華大學圖書館 |

Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems./
Author:	Zhao, Tianyu.
Description:	1 online resource (266 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 84-01, Section: B.
Contained By:	Dissertations Abstracts International84-01B.
Subject:	Energy. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29282520click for full text (PQDT)
ISBN:	9798802756478

Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems.
Zhao, Tianyu.

Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems. - 1 online resource (266 pages)

Source: Dissertations Abstracts International, Volume: 84-01, Section: B.

Thesis (Ph.D.)--The Chinese University of Hong Kong (Hong Kong), 2022.

Includes bibliographical references

In this thesis, we study two important problems for the efficient and optimal operation in the deregulated electricity supply chain and power system: (i) what is the outcome of the bidding game among demand-side consumers in the deregulated electricity market and how to design the desirable market mechanism to improve market efficiency? and (ii) how to design a tractable approach to solve the optimal power flow problem efficiently for the cost-effectively and trustworthy operation of power system? Specially, we investigate the two problems via game theoretical and deep learning approaches.Day-ahead Electricity Market Equilibrium with Load Uncertainty: We consider the scenario where N utilities strategically bid for electricity in the day-ahead market and balance the mismatch between the committed supply and actual demand in the real-time market, with uncertainty in demand and local renewable generation in consideration. We model the interactions among utilities as a non-cooperative game, in which each utility aims at minimizing its per-unit electricity cost. We investigate utilities' optimal bidding strategies and show that all utilities bidding according to (net load) prediction is a unique pure strategy Nash equilibrium with three salient properties. First, it incurs no loss of efficiency; hence, competition among utilities does not i increase the social cost. Second, it is robust and (0, N − 1) fault immune. That is, faulty behaviors of irrational utilities only help to reduce other rational utilities' costs. Third, it is resilient against coalitions in the sense of super strong with arbitrary monetary transfer schemes, i.e., no coalition of utilities can deviate from the equilibrium cooperatively so that all members' costs are not increased, and at least one member's cost is strictly reduced. The expected market supply demand mismatch is minimized simultaneously, which improves the planning and supply-and-demand matching efficiency of the electricity supply chain. We prove the results hold under the settings of correlated prediction errors and a general class of real-time spot pricing models, which capture the relationship between the spot price, the day-ahead clearing price, and the market-level mismatch. Simulations based on real-world traces corroborate our theoretical findings. Our study adds new insights to market mechanism design. In particular, we derive a set of fairly general sufficient conditions for the market operator to design real-time pricing schemes so that the interactions among utilities admit the desired equilibrium.Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems: We propose preventive learning as the first framework to solve constrained optimization problems with convex constraints without post-processing and study its application to DC optimal power flow problems. Ensuring solution feasibility is a key challenge in developing Deep Neural Network (DNN) schemes for solving constrained optimization problems, due to inherent DNN prediction errors. In this thesis, we propose a "preventive learning" framework to systematically guarantee DNN solution feasibility for problems with convex constraints and general objective functions. We first apply a predict-and-reconstruct design to not only guarantee equality constraints but also exploit them to reduce the number of variables to be predicted by DNN. Then, as a key methodological contribution, we systematically calibrate inequality constraints used in DNN training, thereby anticipating prediction ii errors and ensuring the resulting solutions remain feasible. We characterize the calibration rate and the critical DNN size sufficient for ensuring universal feasibility, based on which a universal solution feasibility guaranteed DNN can be directly constructed without training. We further propose a new Adversary-Sample Aware training algorithm to improve DNN's optimality performance without sacrificing feasibility guarantee. Overall, the framework provides two DNNs. The first one constructed from the step of determining the sufficient DNN size can guarantee universal feasibility, while the other DNN obtained from the proposed Adversary-Sample Aware training algorithm further improves optimality and maintains DNN's universal feasibility simultaneously. We apply the preventive learning framework to develop DeepOPF+ for solving the essential DC optimal power flow problem in grid operation. It outperforms existing DNN-based schemes in ensuring feasibility and attaining consistent desirable speedup performance in both light-load and heavy-load regimes. Simulation results over IEEE test cases show that DeepOPF+ generates 100% feasible solutions with < 0.87% optimality loss and up to two orders of magnitude computational speedup, as compared to a state-of-the-art iterative solver.Overall, we provide game-theoretical and deep learning approaches for the optimal and efficient operation in the power system in both day-ahead and real-time scales. We believe our work provides useful insights with economic and practical implications in modern power systems.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2023

Mode of access: World Wide Web

ISBN: 9798802756478Subjects--Topical Terms:

876794
Energy.
Subjects--Index Terms:

Electricity marketIndex Terms--Genre/Form:

542853
Electronic books.

Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems.
LDR:11422nmm a2200433K 4500 001 2362336
005 20231027104005.5
006 m o d
007 cr mn ---uuuuu
008 241011s2022 xx obm 000 0 eng d
020 $a 9798802756478
035 $a (MiAaPQ)AAI29282520
035 $a AAI29282520
040 $a MiAaPQ $b eng $c MiAaPQ $d NTU
100 1 $a Zhao, Tianyu. $3 3543216
245 1 0 $a Day-ahead Electricity Market Equilibrium with Load Uncertainty and Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems.
264 0 $c 2022
300 $a 1 online resource (266 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 84-01, Section: B.
500 $a Advisor: Chen, Minghua;Zhang, Angela Yingjun.
502 $a Thesis (Ph.D.)--The Chinese University of Hong Kong (Hong Kong), 2022.
504 $a Includes bibliographical references
520 $a In this thesis, we study two important problems for the efficient and optimal operation in the deregulated electricity supply chain and power system: (i) what is the outcome of the bidding game among demand-side consumers in the deregulated electricity market and how to design the desirable market mechanism to improve market efficiency? and (ii) how to design a tractable approach to solve the optimal power flow problem efficiently for the cost-effectively and trustworthy operation of power system? Specially, we investigate the two problems via game theoretical and deep learning approaches.Day-ahead Electricity Market Equilibrium with Load Uncertainty: We consider the scenario where N utilities strategically bid for electricity in the day-ahead market and balance the mismatch between the committed supply and actual demand in the real-time market, with uncertainty in demand and local renewable generation in consideration. We model the interactions among utilities as a non-cooperative game, in which each utility aims at minimizing its per-unit electricity cost. We investigate utilities' optimal bidding strategies and show that all utilities bidding according to (net load) prediction is a unique pure strategy Nash equilibrium with three salient properties. First, it incurs no loss of efficiency; hence, competition among utilities does not i increase the social cost. Second, it is robust and (0, N − 1) fault immune. That is, faulty behaviors of irrational utilities only help to reduce other rational utilities' costs. Third, it is resilient against coalitions in the sense of super strong with arbitrary monetary transfer schemes, i.e., no coalition of utilities can deviate from the equilibrium cooperatively so that all members' costs are not increased, and at least one member's cost is strictly reduced. The expected market supply demand mismatch is minimized simultaneously, which improves the planning and supply-and-demand matching efficiency of the electricity supply chain. We prove the results hold under the settings of correlated prediction errors and a general class of real-time spot pricing models, which capture the relationship between the spot price, the day-ahead clearing price, and the market-level mismatch. Simulations based on real-world traces corroborate our theoretical findings. Our study adds new insights to market mechanism design. In particular, we derive a set of fairly general sufficient conditions for the market operator to design real-time pricing schemes so that the interactions among utilities admit the desired equilibrium.Preventive Learning for Ensuring DNN Solution Feasibility with Application to DC Optimal Power Flow Problems: We propose preventive learning as the first framework to solve constrained optimization problems with convex constraints without post-processing and study its application to DC optimal power flow problems. Ensuring solution feasibility is a key challenge in developing Deep Neural Network (DNN) schemes for solving constrained optimization problems, due to inherent DNN prediction errors. In this thesis, we propose a "preventive learning" framework to systematically guarantee DNN solution feasibility for problems with convex constraints and general objective functions. We first apply a predict-and-reconstruct design to not only guarantee equality constraints but also exploit them to reduce the number of variables to be predicted by DNN. Then, as a key methodological contribution, we systematically calibrate inequality constraints used in DNN training, thereby anticipating prediction ii errors and ensuring the resulting solutions remain feasible. We characterize the calibration rate and the critical DNN size sufficient for ensuring universal feasibility, based on which a universal solution feasibility guaranteed DNN can be directly constructed without training. We further propose a new Adversary-Sample Aware training algorithm to improve DNN's optimality performance without sacrificing feasibility guarantee. Overall, the framework provides two DNNs. The first one constructed from the step of determining the sufficient DNN size can guarantee universal feasibility, while the other DNN obtained from the proposed Adversary-Sample Aware training algorithm further improves optimality and maintains DNN's universal feasibility simultaneously. We apply the preventive learning framework to develop DeepOPF+ for solving the essential DC optimal power flow problem in grid operation. It outperforms existing DNN-based schemes in ensuring feasibility and attaining consistent desirable speedup performance in both light-load and heavy-load regimes. Simulation results over IEEE test cases show that DeepOPF+ generates 100% feasible solutions with < 0.87% optimality loss and up to two orders of magnitude computational speedup, as compared to a state-of-the-art iterative solver.Overall, we provide game-theoretical and deep learning approaches for the optimal and efficient operation in the power system in both day-ahead and real-time scales. We believe our work provides useful insights with economic and practical implications in modern power systems.
520 $a 在本文中，我们研究了两个重要问题，以实现放松管制的电力供应链和电力系统的高效及优化运行：（i）在放松管制的电力市场中，需求侧用户之间的竞价博弈结果如何，以及如何设计理想的市场机制以提高市场效率？以及（ii）如何设计一种易于使用的方法来有效地解决最优潮流问题，从而实现电力系统的经济高效和可靠运行？在文中，我们特别通过博弈论和深度学习的方法来研究这两个问题。具有负载不确定性的日前电力市场的均衡：我们考虑这样一种情况，即有 N 个公用事业公司在日前市场战略性地竞标电力，并在实时市场中平衡已确定的供应和实际需求之间的不匹配，同时亦考虑到需求和本地的可再生能源发电的不确定性。我们将公用事业公司之间的相互作用建模为非合作博弈，其中每个公用事业公司以最小化其单位电力成本为目标。我们研究了公用事业公司的最优投标策略，并表明所有公用事业公司根据（净负荷）的预测值投标是唯一的纯策略纳什均衡，其具有两个显着属性。首先，它不会导致效率损失，因此公用事业公司之间的竞争不会增加社会成本。其次，它是鲁棒的并且具有 (0, N − 1) 错误免疫能力。也就是说，非理性的公用事业公司的错误行为只会降低其他理性公用事业公司的成本。第三，它在具有任意货币转移方案的超强纳什均衡的意义上对联盟是具有弹性的，即任何公用事业联盟都不能协同偏离均衡，从而不会增加所有成员的成本，并且严格降低至少一个成员的成本。在该均衡下，市场预期的供需失配也被最小化，从而提高了电力供应链的规划和供需匹配效率。我们证明了在相关预测误差和在一类通行的现货定价模型下，该结果仍然成立。该类通行的现货定价模型描述了现货价格、日前结算价格和市场层面错配之间的关系。基于真实世界数据的模拟证实了我们的理论发现。我们的研究为市场机制设计增加了新的见解。特别是，我们推导出了一组相当普遍的充分条件以供市场运营商设计实时定价方案，从而公用事业公司之间的相互作用可以实现所需的均衡。用于确保深度神经网络解的可行性的预防性学习，并应用于直流最优潮流问题：我们提出预防性学习作为第一个解决具有凸约束的约束优化问题而无需后处理的框架，并研究其在直流最优潮流问题中的应用。由于深度神经网络固有的预测错误，确保解的可行性是开发用 iv 于解决约束优化问题的深度神经网络 (DNN) 方案的关键挑战。在本文中，我们提出了一种"预防性学习"框架，以系统地保证 DNN 对于具有凸约束和一般目标函数的问题的解可行性。我们首先应用预测和重构设计来保证等式约束，同时还可以利用它们来减少 DNN 预测的变量数量。然后，作为一个关键的方法论贡献，我们系统地校准了 DNN 训练中使用的不等式约束，从而在即使有预测误差的情况下也可以确保得到的解仍然可行。我们确定了为确保普遍可行性所需的校准比率和足够的 DNN 大小，基于该步骤一个无需训练即可以保证解的通用可行性的 DNN 可以被直接构建出来。我们进一步提出了一种新的对抗样本感知的训练算法，以在不牺牲可行性保证的情况下进而提高 DNN 的最优性的性能。总体而言，该框架提供了两个 DNN。第一个构建自确定足够的 DNN 大小的步骤的 DNN 可以保证解的通用可行性，而从所提出的对抗样本感知的训练算法获得的另一个 DNN 进一步提高了最优性并同时保持了 DNN 的通用可行性。我们应用预防性学习框架来开发 DeepOPF+，以解决电网运行中至关重要的直流最优潮流问题。它改进了现有的基于 DNN 的方案，以确保可行性并在轻载和重载状态下获得一致的理想加速性能。基于 IEEE 测试用例的模拟结果表明，DeepOPF+ 生成 100% 的可行解，与 < 0.87% 最优性损失和高达两个数量级的计算速度加速相比与最先进的迭代求解器。总体而言，我们提供了博弈论和深度学习方法，以实现电力系统在日前和实时时间尺度上的优化和高效运行。我们相信我们的工作为现代电力系统提供了具有经济和实际意义的有用见解。
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2023
538 $a Mode of access: World Wide Web
650 4 $a Energy. $3 876794
650 4 $a Electrical engineering. $3 649834
650 4 $a Computer science. $3 523869
650 4 $a Information science. $3 554358
653 $a Electricity market
653 $a Load uncertainty
653 $a Preventive learning
653 $a DNN Solution
653 $a Power flow
655 7 $a Electronic books. $2 lcsh $3 542853
690 $a 0544
690 $a 0505
690 $a 0791
690 $a 0984
690 $a 0723
710 2 $a ProQuest Information and Learning Co. $3 783688
710 2 $a The Chinese University of Hong Kong (Hong Kong). $3 1017547
773 0 $t Dissertations Abstracts International $g 84-01B.
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29282520 $z click for full text (PQDT)