导　　师： 陈皓勇

学科专业： 080802

授予学位： 博士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 电力市场化改革是世界性的难题。国外现实经验表明，电力市场建设是一个研究、实践、评估与改进反复交替不断发展的过程。我国电力市场改革遇到的问题和挫折反映了我国当前在市场结构、交易机制、输电服务等综合研究与实践上仍不成熟，在目前条件下开展电力市场综合实践风险较大。当前阶段如何高效率、低风险、贴近电力工业实情地进行电力市场综合研究与设计具有重要意义。 电力市场模拟实验是近年来由国外学者提出的一种研究方法和技术手段，研究者可以脱离实际市场复杂性，通过可控实验环境，实现电力市场方案验证、问题发掘、培训支持和评价分析。本文研究目的是贴近电力工业实情设计开发一套融合实验经济学和基于代理计算经济学方法、软切换市场模型、可用于广域网的先进电力市场综合模拟系统，并对涉及的系统架构、博弈实验模型、博弈实验设计和市场出清方法等问题进行探讨，其主要研究内容可概述如下： 1.在综合国内外代表性的实验方法的内涵、共性和差异、优势和劣势，以及模拟实验平台的方法和技术特征基础上，本文提出并开发了融合实验经济学和ACE方法、可软切换市场模型、可用于广域网的电力市场综合模拟系统。首先从技术趋势、技术难点和需求分析3个角度，分析了电力市场综合模拟系统的可行性。其次对模拟系统的关键架构内容逻辑模型与角色、数据库、功能、模拟实验流程作了设计阐述。对该系统的可行性、软件架构设计作详细讨论。本文构建的模拟系统以可用于实际交易为根本目标，在可模拟行为博弈和机器博弈、多市场模式、跨网络交易上有重要创新成果。 2.借助本文构建的模拟系统，就模拟实验设计的问题展开讨论。首先构建了模拟实验市场模型；其次设计了①实验经济学行为博弈、②基于代理计算经济学机器博弈、③实验经济学与基于代理计算经济学的混合博弈的实验流程；最后结合我国某区域电力市场数据对系统进行了模拟实验研究。模拟实验结果表明，实验经济学行为博弈和基于代理计算经济学（ACE）机器博弈模拟的过程和结果具有可比性，两种方法的模拟过程相同，实验经济学行为博弈中参与人的决策行为是非常复杂的，并不符合经济理论简单的理性假设；两种实验方法也具有互相参考性：两种实验方法的对比研究能论证ACE机器博弈模拟的可靠性，以及更复杂市场模式下的性能，ACE机器博弈对于电力经济系统这一高度非线性的系统博弈均衡的求解，能验证实验经济学行为博弈结论的有效性。 3.考虑风电接入是目前电力市场模拟实验和我国区域电力市场的新研究课题。构建了一种基于日前市场多风电场功率预测偏差的极限场景方法，建立了考虑火电、水电和燃气发电的考虑安全约束机组组合模型，采用混合整数规划法进行了求解，最后通过118节点系统算例进行了详细分析，验证了所提极限场景方法的有效性，主要结论有：极限场景方法是一种具有高适应性的模拟方法，对于高渗透率情形同样可求得适应性好的机组组合方案；当研究更大规模风电场数目情形时，场景如何选取还有待进一步讨论；火电、水电、燃气发电联合系统对多风电场接入具有更好的适应力。 4.提出了考虑多风电场接入的日内多时间尺度滚动调度策略和模型，与日前市场机组组合相结合，构成了4h时间尺度的滚动调度策略，采用内点法进行了求解，最后采用118节点系统算例和某省级电网进行了详细分析，验证了所提极限场景方法及其调度策略的可行性。主要结论有：与日前计划相比，日内滚动计划的预测精度高，误差场景个数少，松弛了约束条件，使调度的经济性得到改善；验证了滚动调度方式较传统调度方式更能适应风电场出力波动，得出了滚动调度方式是以损失经济性为代价换取对风电场出力不确定性的适应能力的结论。 The electricity market reform is a worldwide issue. Practical experience shows thatelectricity market construction is a repeated development process of research, practice,evaluation and improvement. During China’ electricity market reform, the power industryencountered a series of complex theoretical and practical issues, which reflect that the currentcomprehensive study and practice of the electricity market structure, trading mechanismdesign and electricity transmission services is still immature. Under the current circumstances,there are certain risks for China to carry out integrative practice of the electricity market.Therefore, a comprehensive study and design of the electricity market with high efficiencyand low risk based on the actual conditions of the power industry are of great significance. Electricity market simulation is a research method and technology which is presented byforeignal scholars in recent year. This method and technology allows the researchers to verifythe market programs, discover potential problems, offer training support, analyze and evaluatethe power market through well-controalled environment which avoids the complexity ofactual market. This paper aims to design and develop a set of electricity market simulationsystem combining experimental economics and agent-based computational economics method,soft switch market model, which can be used in wide area networks, and discuss the relatedproblems such as the system architecture, game experiment model, game experiment designand market clearing method. The main research contents can be summarized as follows: 1. Basing on summary of the contents, similarities and differences, advantages anddisadvantages of the representative experiment methods at home and abroad, and thesimulation platform method and technical features, this paper presents an electricity marketsimulation system combining experimental economics and ACE method, soft switch marketmodel, which can be used in wide area networks. Firstly, this paper analyzes the feasibility ofthe electricity market simulation system from the three angles of technology trends, technicaldifficulties and demand analysis. Secondly, this paper illustrates the design of key architecturecontents of simulation system: logical model and its role，database，functions, simulationprocess. The feasibility and software architecture design of this system will be discussed indetails in this paper. The proposed simulation system can be used in the actual transaction as a fundamental objective, and bring important innovation in the simulative behavior game,machine game, multi-market model and the cross-network transactions. 2. Discuss the problem of simulation experimental design with the analog systemconstructed in this paper. Firstly, the paper builds simulation market model; Secondly, wedesign①experimental economics game,②machine game based on Agent ComputationalEconomics,③experiment processes combining the experimental economics and parallelgame based on Agent Computational Economics Game Theory; Finally, this paper conductedsimulation experiment studies on the system using the electricity market data of one region inChina. Simulation results showed that the process and results of the experimental economicsgame and computer game simulation based on agent computational economics /(ACE/) arecomparable. The simulation process of these two methods is the same. The participantdecision-making behavior under experimental economics game environment is very complexwhich does not meet the simple rational assumption of economic theory. The twoexperimental methods have certain mutual reference: a comparative study of the twoexperimental methods can demonstrates the credibility of ACE machine game simulation, andits function under more complex market model. Meanwhile, using the ACE machine game asa solution of system game equilibrium for power economic system, a highly non-linearsystem, can verify the conclusion of experimental economics game. 3. Associated wind power integration is the new research subject for the currentelectricity market simulation experiment and the regional electricity market of China. Thispaper establishes an extreme scenario method based on the day-ahead market wind powerpredicted deviations, builds a unit combination model of thermal power, hydropower and gasgeneration with security constraints, adopts a mixed integer programming method to solve theproblem, uses the118node system account case for detailed analysis at the end, and verifiesthe extreme scenario method mentioned. The main conclusions are as follows: the extremescenario method is a simulation method with high adaptability, which can obtain the unitcombination schemes with good adaptability even under the circumstance of high wind farmpenetration; for study of larger-scale wind power, the selection of scenario needs furtherdiscussion; the integrate power system with thermal power, hydropower and gas generation ishighly adaptable to associated wind power farms. 4. Propose a real-time balancing market multi-time rolling dispatch strategy and modelof associated wind power integration combined with the day-ahead market unit Commitmentcombination to form a rolling dispatch method with the time scale of4hours. Adopts theinner point solutions to solve the problem, uses the118node system account case and oneprovincial power grid for detailed analysis at the end, and verifies the extreme scenariomethod mentioned and the feasibility of its dispatch strategy. The main conclusions are asfollows: compared with the day-ahead market scheduling, the prediction precision ofreal-time balancing market rolling schedulingis higher; the number of deviation scenario isless; the constraint condition is reduced; the economic efficiency of dispatch is improved; itverifies that the rolling dispatch method is more adaptable to wind power fluctuationcompared with conventional dispatch method. The conclusion is that the adaptability ofrolling dispatch method to the uncertainty of wind power is gained at the cost of economicefficiency.

关 键 词： 电力市场 综合模拟 实验经济学 行为博弈 计算经济学 机器博弈 代理 混合博弈 博弈实验 日前市场 机组组合 出清算法 平衡市场 滚动调度

分 类 号： [F426.61 F224.32]

领　　域： [经济管理] [经济管理]