【作者】 喻菁；

【导师】 周建中；

【作者基本信息】 华中科技大学 ， 系统分析与集成， 2005， 博士

【摘要】 电力市场的改革在世界范围内蓬勃兴起,为社会带来巨大经济效益的同时,也促使电力系统研究领域产生了众多新的研究课题。探索中国电力市场改革的发展方向,并预先对适合中国国情的电力市场模式进行理论分析和模型研究具有重大的理论和现实意义。为了探讨适合中国的电力市场改革发展模式,必需从国外借鉴改革经验。在分析国外已建立的电力市场模式,并综合国际上对电力市场模式的研究成果后,可将电力市场划分为:垂直垄断、发电竞争、输电开放和输配电开放四种类型。每种类型的电力市场对电源分布情况、电力系统的基础建设状况、市场需求等条件均有不同要求。因为中国幅员辽阔,经济处于高速发展阶段,各地区发展程度很不均衡,所以导致理论上划分的任意一种电力市场模式都不能完全满足要求。根据中国的电源、电网、市场等必要因素的发展现状,不难得出中国电力市场将在一段时期内处于多种模式并存状态的结论,即在条件成熟的部分地区率先建立输电开放或输配电开放模式的区域电力市场,而在其它大部分地区进行垂直垄断向发电竞争模式的转化和改革。既然多种模式并存是发展必须经历的过程,那么在经济高度发达、电网建设完善,以及其它必备条件均成熟的地区建立输配电开放模式的电力市场,即电力零售市场必定成为发展的方向。由于电力市场一直被国内外学者视为复杂性科学交叉发展的前沿问题之一,电力市场构成的复杂系统交织着极为复杂的各种物质流与信息流的映射关系,加上人们对电力市场演化复杂本质认识的局限性,使电力市场的模型研究成为一项复杂但非常具有挑战性的工作。很多传统的研究方法在研究及其成果的应用中受到限制,不再完全适应新的研究任务的要求,需要从新兴的领域和交叉学科探求新的研究方法和手段。随着分布式人工智能的兴起,Multi-Agent 系统(MAS)理论诞生并得到迅速发展,其灵活、智能、自适应、自学习等特点,为电力市场建模研究提供了一种有效的方法。因此,作者引入MAS 体系建立了一种新型的电力零售市场模型,该市场模型中包括电力交换公司Agent(PXA)、独立电厂Agent(IPPA)、电力零售商Agent(RCA)和电力用户Agent(PUA)四类Agent,通过Agent 之间的相互协调、合作、竞争,完成市场的交易过程。尤其是在建模过程中创新使用有色Petri 网描述并分析了Agent 之间的更多还原

【Abstract】 The reform of the electricity market rises vigorously worldwide, bringing the enormous economic benefits to the society, and impels the research field of the power system to produce the numerous new subjects for researchers. It has significant and realistic meanings to explore the developing direction of the reform of electric market in China, to carry on theory analysis and model the electricity market suitable for China. To probe into a suitable electricity market mode for China’s reform development, we must draw lessons from other countries’reform experience. After analyzing the modes of electricity market that have already been set up abroad, combined with research results of the modes of electric market in the world, the electricity market could be categorized as: vertical monopoly, competitive generation, transmitting open, transmitting and distribution open electricity market. Each type of electricity market has different demands for power resource distribution situation, power grids construction state and power consumption demands. Because China is vast in territory, rapid in its economic development and unbalanced in the development degree in different areas, no modes of electricity market theoretically defined can completely meet the demands of China electricity market reform. According to the current developing situation of China electricity market, including the power resource, power grids and market demand in China, etc., it is easy to draw the conclusion that the China electricity market will, within a certain period of time, be in a state where all modes of market coexist. That is to say, some areas with satisfied conditions may take the leads in setting up a regional electricity market, which would be characterized as the transmission open or even transmission and distribution open mode in the electricity market, while in other most areas a transition or reform will be carried out from vertical monopoly mode to competitive generation open mode. Since the coexisting state of many kinds is a necessary step during development, setting up the electricity market of the transmission and distribution open mode, namely retail electricity market, in the area where economy is highly developed and power grids are perfect in construction and other essential requirements are met, must become the direction of development. The electricity market has always been regarded domestically and internationally as one of the pioneering issues in the science of complicacy and cross discipline, because of the complicated system composed of electricity market interweaves with various kinds of extremely complicated reflection between physical and information flows; in addition, due to limitation on people’s understanding of complicated essence of electricity market, the modeling research of the electricity market has become a very complicated and challenging work. A lot of traditional research approaches used in past has now limited further research in laboratory and achievement application in market, so they no longer meet the needs of new research projects completely; therefore, it is necessary to seek new research approaches and means from new developing field and cross discipline. With a rise in distributed artificial intelligence, the theory of Multi-Agent system (MAS) was born and developed rapidly, which has offered a kind of effective method for the research in modeling electricity market because of its flexibility, intelligence, self-adaptation and self-learning, etc. Thus, the author in this study has introduced MAS and developed a new model for retail electricity market including four kinds of agents, which are the power exchanges company agent (PXA), independent power plant agent (IPPA), retail company agent (RCA) and power user agent (PUA). The trade course of this modeled market is completed through their coordination, cooperation and competition among the agents. We for the first time use the colored Petri-net in the present study to describe and analyze communication, coordination and competition as well as cooperation among the agents, which has effectively expressed their complicated coupling relations and interaction mechanism among the agents in market members and solved the problem concerning the view of MAS relations that is difficult to complete for general flow chart. This model of retail electricity market, which has utilized the agents abundantly to model the characteristics of market member, can accurately reflect the complex environments in the electricity market through colored Petri net and avoid the difficulty occurring often in developing accurate mathematics model by using classical research approach. This model also has considered the mutual opening and privacy during their communication among market members by designed colored token. In the meantime, this model can control the market scale by setting the quantity of the agents in the market member and provide theoretical foundation and artificial analytical method for an ideal market looking for"effective competition ". This model also possesses a very important value for their application in industry, because of the merging of MAS mechanism existed in colored Petri-net, which results in development of distributed computer system easily In addition, the study of short-term generation has been being a focus in academy and engineering since it can bring remarkable economic benefits to power industry. However, new environment in electricity market provides a new thought for solving the problem on short-term generation, which includes: 1). The electricity is no longer optimized and distributed by central dispatcher, but is the goods traded in the electricity market; 2). The traded contents not only include the generated electricity, but also the reserve capacity that will be a part of the trade; 3).the price of electricity is no longer unalterable, which will be affected by electricity demand in market, and the power plants can adjust their generation schedules according to the changed price of electricity. This new thought can be called profit-based short-term generation schedules (PSGS). Under PSGS, market operating system will force market members to participate in the trade competition actively and self-consciously, which will result in a variable capacity of generation and reservation and thus increase unexpected factors in short-term generation problem greatly. For better solving PSGS problem, adapting to new environment in retail market of electricity and distributing traditional centralized optimization and concentrated dispatching problem to the units at the basic level, in this study we have developed a profit-base short-term generation schedules Multi-Agent system (PSGS-MAS) based on artificial intelligence by using MAS to solve PSGS problem. This system can automatically complete formulation, decision and assessment on generation plan through self-adaptation, self-education and cooperation of the agents. The agent can learn experience from historical operations and reduce the unnecessary visit route, thus using this system can shorten calculating time and improve the efficiency on systematic optimization decision. This method has provided a brand-new research approach and implementing means for the similar problem occurred in generation schedules. During developing PSGS-MAS, the decision made by the agent through a core of decision-maker, organization of Agent –Mobile Agent (MA), is even more similar to the human thinking mode, which may make possible to avoid spoiling the ship for a halfpennyworth of tar. The instance analysis has demonstrated that the method based on therule is an effective means to improve Agent intelligence, which also will enhance the operation performance of the whole system. In short-term generation schedule problem, large-scale unit commitment is a focal and difficult point. The problem is because of logic disturbances from complex constraints, which always results in an unexpected or controversial outcome. Therefore, by adjusting our research work starting from reconstructing high dimensions to low dimensions, we have proposed a method named multilayer controllable colored Petri-net (MCCP) and offered an effective optimize method and means for the similar commitment problem in power system generation and distribution by utilizing MCCP to model various kinds of complicate constraints in unit commitment problem and to describe intricate restraint terms in ocular chart way. To best our knowledge, most of the theories and methods are for the first time applied in this research work. Theoretical analysis with simulation results has demonstrated that the theory of Multi-Agent and Petri-net possesses wide prospects in the research fields relevant electricity market and will bring remarkable economic benefits.更多还原