基于声誉的激励性电力竞价机制研究

Frequently happened "Price Spikes", caused by uncertain market factors, exacerbate the contradiction between the generators' personal interest and the overall objective of the electricity market, and accelerate the requirement of improving current bidding mechanism. Based on the theory of mechanism design, supplemented by the method of robust control, this project aims to designing a new reputation-based incentive bidding mechanism, which, along with its long-term and short-term incentive effects, could guide the generators to behave consistently with the long-term goal of the market, so that the disturbing effects of the uncertain factors could be suppressed in a controllable range. Firstly, we will study on evaluating the reputaion of the generators via methods of literature review, expert interview, and field investigation, and then a preliminary design of mechanism will be made. Secondly, a reputation-based incentive model for electricity bidding will be proposed to improve the mechanism design. Further, symbiotic evolution of uncertain factors, generators' strategies and incentive mechanism will be investigated, and the method of robust control is applied to optimize the evolutionary scheme of the incentive mechanism. Lastly, a multi-agent system will be built and applied to test and modify the research results. The work of this project could not only enrich theory of auction mechanism design, but also provide decision references for managers of the electricity market, and hence contributes to future development of Chinese electricity market.

不确定市场因素引发的频繁"价格钉"现象，加剧了发电商个体利益和电力市场整体目标之间的矛盾，加速了对现行竞价机制进行改进和完善的需求，该需求转化形成研究新型电力竞价机制的推动力。本项目基于机制设计理论，辅以鲁棒控制方法，设计一种基于声誉的新型激励性电力竞价机制，结合该机制的长、短期激励效应，引导发电商个体行为快速趋向市场长期目标，将不确定因素的扰动影响抑制在可控范围内。结合文献查阅、专家访谈与实地调研，研究发电商声誉度量方法，探讨激励机制的机理设定，进而构建基于声誉奖励的电力竞价激励模型，借助模型优化促成激励机制的机理设计；并进一步研究不确定因素、发电商策略、激励机制的共生演化，应用鲁棒控制方法促进激励机制的自适应演化；最后构建多Agent仿真系统，检验和修正研究结论。研究成果既可在理论上丰富和发展拍卖机制设计理论，且可在应用上为电力竞价市场的管理者提供参考，推动我国电力市场的进一步发展。

不确定市场因素引发的频繁"价格钉"现象，加剧了发电商个体利益和电力市场整体目标之间的矛盾，加速了对现行竞价机制进行改进和完善的需求，该需求转化形成研究新型电力竞价机制的推动力。本项目基于机制设计理论，以长、短期激励性契约为媒介，设计了一种基于声誉的新型激励性电力竞价机制，引导发电商个体行为快速趋向市场长期目标，将不确定因素的扰动影响抑制在可控范围内。结合文献查阅、专家访谈与实地调研，研究发电商声誉度量方法，探讨激励机制的机理设定，进而构建基于声誉奖励的电力竞价激励模型，借助模型优化促成激励机制的机理设计；并进一步研究不确定因素、发电商策略、激励机制的共生演化，最后构建多Agent 仿真系统，应用基于Stackelberg博奕的Q学习方法促进激励机制的自适应演化；通过仿真检验和修正研究结论。研究成果既可在理论上丰富和发展拍卖机制设计理论，且可在应用上为电力竞价市场的管理者提供参考，推动我国电力市场的进一步发展。