【作者】 宋恒力；

【导师】 岳超源；

【作者基本信息】 华中科技大学 ， 控制科学与工程， 2013， 博士

【摘要】 近年来，我国加快了流域水电能源开发的进程，随着电力市场化改革逐步深入，流域梯级电站参与电力市场是必然的趋势，因此流域开发公司需要根据市场规则，制定市场环境下流域梯级水电站的最优竞价策略，这也是优化流域水资源能源结构、实现流域可持续发展的需要。流域梯级水电站具有特殊的技术、经济特性，在参与市场竞争的同时还需要统筹协调水资源以实现社会效益，在市场竞争环境下如何保证水电资源的最优配置成为近期研究的热点之一。本文运用联盟博弈理论与方法，结合我国电力市场规则，在借鉴国内外梯级水电站参与上网侧电力市场竞争经验的基础上，研究了流域梯级水电站的联盟策略、联盟形式与结盟条件等问题，为梯级电站联盟上网竞争提供理论依据，同时为以“同网同质同价”为目标的电价机制改革提供参考。流域梯级电站如果隶属于同一投资主体，则具备天然的联盟条件，在竞价时必然充分考虑电站间水力联系，以实现梯级总收益的最大化；而不同投资主体下的水电站，由于电站间的水资源联系和提高自身市场竞争力的需要，在一定条件下也具备联盟的可能性。本文为了探讨流域梯级水电站联盟的存在条件和内部形成动因，做了以下工作：首先建立了静态的博弈模型，该模型是在详细分析了电力市场技术经济特点和交易规则的基础上，结合水电站自身特征和梯级电站之间的水力联系现实依据建立的。模型对水电成本函数、需求函数及物理约束等进行了量化，并给出了合理的假设条件，模型通过增广拉格朗日法求解，比较在单次竞价中，流域“一站一价”和“统一电价”两种结算方式对电站收益的影响，分析了电站联盟存在的外部条件。其次，在联盟的形成动因方面，论文做了理论上的探究。着重讨论联盟博弈的解概念，包括稳定集、核心和Shapley值等，这些静态的解概念不能有效解决梯级电站联盟的动态性问题。为此，论文对基于远见性的联盟博弈理论做了探讨，在最大一致集概念的基础上，针对梯级电站联盟策略的历史依赖性，以完全无覆盖路径作为描述梯级电站联盟形成过程的方法和判定稳定性的理论依据。再次，在静态模型和理论研究的基础上，论文建立了完全信息条件下的动态联盟博弈模型，以分析流域梯级电站在多次竞价中，处于不同市场环境下的联盟形成过程。模型假设梯级电站三种潜在的联盟形式，联盟作为整体参与市场竞价，上报相同电价和联盟总发电量，中标后联盟内部充分考虑上下游电站水力联系协调发电，实现联盟总利润的最大化。在算例研究中，利用粒子群算法求解，分析了不同弹性需求条件下，流域梯级电站联盟形式的变化路径和最终稳定状态。最后，为了使模型更加符合电力市场实际，论文进行了不完全信息下的梯级电站联盟博弈分析。应用预测对手报价的方法解决参与者的不完全信息问题。为提高预测准确度，对灰色预测模型GM(1,1)做出修正，并用NORD-POOL电力市场的实际数据验证了预测模型的可行性，进而能够将其应用到博弈模型中。通过算例得出，在不完全信息下，电站的利润发生变化，由此决定电站对联盟形式的偏好顺序相应变化，这将会影响梯级电站联盟的稳定性。更多还原

【Abstract】 In recent years, China is gradually accelerating the development process of hydropower.Along with the reform of the power market, it is becoming inevitable for cascadedhydropower stations to participant in electricity market. Thus the basin developmentcompanies need to optimalize their bidding strategies in competitive electricity market accordingto market rules, which is of great practial significance for upgrading the water resourceenergy construction and realizing sustainable development of river basins. Cascadedhydropower stations have special technical and economic characteristics. They have to becoordinate as a whole to schedule water resource while competing with each other. How tooptimize the water and power resources by bidding in market has always been the forefrontand hotspot of academic research. This thesis focuses on cascaded hydropower station’salliance strategy, coalition formation and stability referring to successful cases of foreignpower market with participation of hydropower and combining the national conditions ofChina. The conclusion provides a theory reference for cascaded hydropower stations’bidding strategy and can be considered as a police suggest on pricing reforming which aimsto equal price for the same electricity quality on the same power network.Hydropower stations which are belonged to the same investment entity tend tocoordinate naturally. They would rather bid cooperatively than individually on the target ofmaximize the profit of the coalition considering adequately water and power resourcerelations among stations. The coalition could exist as hydropower stations need unifiedmanagement even if hydropower stations belong to different investment entities. To explorethe outer condition and inner motivation of hydropower stations forming coalition, astatic game model is proposed firstly basing on the technological and economic charactersof electricity market place considering intricate relationships among hydropower stations.The model quantifies constraint of cost and demand functions. An extended Lagrangemultiplier method is adopted to resolve the model. The impacts of two clearing form,different and unified pricing, on profit of stations are compared by the simulation exampleas the outer condition of coalitions. Secondly, theory on the motivation of coalition forming is studied. The resolve conceptsincluding stable set, core and Shapley value are discussed mainly to analyze the stabilityand dynamics of coalitions. But those resolve concepts are all static and lack the effect onthe dynamics of coalitions. Coalitional game theory based on farsightedness is researchedon the base of the largest consistent set. The ultimated uncovered path is considered as thejudgment of coalition’s stability as cascaded hydropower stations’ alliance strategy relyingon history path.On the next section, the dynamic coalitional game model is proposed on the achievementof above static model and theory to analyze the coalition forming process in differentmarket places during repeatedly bidding. Three potential coalition contractures areconsidered. Once coalition forms, the stations bid as a whole, declaring unified price andtotal generate quality. The generate quality is dispatched cardinally in coalition to maximizethe coalition’s profit considering the relationship among stations. This model is solved bygenetic algorithm to provide the moving of cascaded hydropower stations coalition andultimate state.Finally, the dynamic model is updated by the method of forecasting opponent’s biddingprice, relaxing the assumption of complete information to be more practical. The classicalgray predict model GM(1,1) is improved to gain a more accuracy result. The predict modelis proved feasible on the test of NORD-POOL power market actual data and embeddedsuccessfully in the game model. The simulation result shows the varying of profit ofstations, which changes the reference of coalition structure in a incomplete informationsystem. So the cascaded hydropower station coalition stability is challenged.更多还原