【作者】 刘红岭；

【导师】 张焰；

【作者基本信息】 上海交通大学 ， 电力系统及其自动化， 2009， 博士

【摘要】 电力工业的市场化改革在世界范围内不断推进,打破了传统的垂直垄断管理模式,发电公司、供电公司及电力用户将以独立的市场参与者成为电力市场中的主体。在市场环境下,水电系统作为独立的发电商,其优化调度管理模式发生了本质性的转变。水电厂商可参与多种交易途径以获得收益,其将以市场交易为中心,以电价为决策导向,以追求自身发电收益最大化为目标,这使得水电系统的优化调度问题面临着新的内容与挑战。同时,在市场环境下,水电系统所面临的径流和电价的随机性,也使得水电厂商在市场交易中面临着很大的收益风险,迫切需要进行风险管理以保障其在市场交易中的利益。本文结合国家自然科学基金重点项目“市场条件下流域梯级水电能源联合优化运行和管理的先进理论与方法”,重点研究充分考虑径流和电价随机性的水电系统的优化调度问题及相应的市场交易策略,融入风险管理方法,建立相应的考虑风险的优化调度模型,以实现收益最大化与风险最小化间的平衡。主要研究内容如下:(1)详细介绍随机线性规划的模型建立、求解及模型优越性评估指标的计算。以场景树模型作为随机线性规划模型的输入,提出并具体实现基于时间序列模型及启发式方法的场景构建方法。针对随机线性规划模型求解困难的问题,以概率距离度量场景缩减前后随机数据过程的近似程度,具体分析快速后向场景缩减技术的设计原理与实现,在尽量减少场景数量的同时,保持随机变量场景树模型的重要特征,进一步扩展了随机线性规划方法的应用范围。(2)提出一种新型的基于随机线性规划的水电站中长期合约电量决策模型,充分考虑径流和电价的随机性对收益的显著影响。该模型以基于不同场景构建方法得到的场景树模型表示径流和电价的随机性,将远期合约决策与日前市场交易决策视为随机规划框架下的不同阶段决策。通过与预测值模型的比较分析,随机线性规划模型由于充分考虑了随机性的影响保证了其收益的优越性;通过与不同决策模式下的随机规划模型比较分析,两种模型对水电站远期合约决策及收益影响的相似性进一步验证了随机线性规划模型的有效性。(3)提出一种新型的基于随机线性规划的水电站组合交易决策模型,解决如何利用有限的发电资源合理参与多个市场的交易以获得考虑随机性影响的期望收益最大化的问题。基于组合交易策略与发电调度策略的紧密联系,将组合交易决策及发电补偿决策视为随机规划框架下的不同阶段决策。通过构建不同的场景树模型来表示径流和电价的随机性,验证不同的场景树模型输入虽然对组合交易决策有影响,但并不影响整个随机线性规划模型的有效性。对组合交易决策的VaR风险评估表明,灵活的组合交易决策能够带来更多收益,同时,也面临收益风险,进一步进行风险管理具有重要意义。(4)对电力市场环境下的风险管理方法及其应用进行总结分析,进而提出在确定性框架下把对电价风险的考虑融入水电站短期优化调度问题中进行求解,分别提出基于期望收益-VaR风险效用模型以及基于场景收益风险惩罚或风险约束的水电站优化调度模型,对不同风险管理方法下的期望收益与风险间的平衡问题进行分析。针对模型的求解,在遗传算法框架下融入改进的快速进化算法的进化机制及竞争选择机制,在进化过程中结合惩罚机制与修复机制,提出基于改进的快速进化算法和遗传算法的IFEP-GA混合优化算法作为上述两种模型的求解方法。并对电价风险的考虑对进化算法的影响加以分析,为水电站在市场竞争中选择合理的风险水平及相应的水电站短期优化调度提供决策支持。(5)基于以VaR作为风险测度与日前市场交易紧密相关的特性,提出将VaR风险惩罚项、组合交易决策及发电补偿决策作为随机规划框架下不同阶段的决策,建立随机规划框架下的期望收益- VaR风险效用模型。通过不同风险因子下模型的求解,得到期望收益与风险间的有效前沿曲线,对应得到不同风险水平下的水电站组合交易决策;通过对组合交易决策以不同的风险测度方法进行风险分析,为水电站进一步选择其合理的风险水平提供决策依据。(6)基于随机规划框架下的风险测度与随机性建模及市场交易策略的紧密相关性,提出以Semi-Variance风险测度,将场景树模型中各场景电价与期望电价间的差额在日前市场交易中反映为相应的风险,进而在随机规划框架下建立考虑风险的水电站组合交易策略模型。模型将随机性建模与风险建模相结合,能够充分反映随机性的影响,实现了对风险的有效管理,进而能够得到不同风险水平下的水电站组合交易决策。通过对水电系统优化调度问题及其市场交易优化问题的应用研究,验证了本文所提出的优化方法和所建立的数学模型的有效性,为水电站在市场环境下进行合理的市场交易提供决策依据,并为其所面临的收益最大化和风险最小化间的平衡问题提供有效的解决途径。更多还原

【Abstract】 The electricity industry throughout the world, which has long been dominated by vertically integrated utilities, is underlying enormous changes. Restructuring has necessitated the decomposition of the three components of electric power industry: generation, transmission and distribution. In deregulated markets, hydropower producers are regarded as independent generating companies, and the management model of optimal scheduling has been greatly changed. The diversity of trading types provides hydropower producer more options to sell their generation products among multiple markets. With the sole objective of maximizing revenues, hydropower producers concern the power exchange in electricity market, and trading decisions are made with the center of market price. All these changes bring new field and challenge to hydroelectric scheduling. Furthermore, a major concern for hydropower producers in restructured market is the profit uncertainty caused by uncertainty in inflows and market prices, and introducing risk management to guarantee the market trade revenue has become an urgent need for hydropower producers.With the support of NSFC project“advanced theory and methodology of river basin cascade hydroelectric energy joint optimal operation and management in electricity market”, this paper concerns hydroelectric scheduling problem and corresponding market trade strategy considering the uncertainty in inflows and market prices. Furthermore, risk management methods is incorporated, and risk-constrained optimal scheduling model is formulated to obtain the tradeoff between maximum of revenue and minimum of risk. The main research contents of this paper are listed as the following:(1) Stochastic linear programming model is introduced in detail, including the construction of model, solution method and the calculation of advantage measure. Scenario tree model is recognized as input of stochastic linear programming model, and its constructing techniques based on time series method and heuristic method are proposed and realized. Concerning the solution difficulty of stochastic linear programming model, a probability metric is used to control the goodness-of-fit of the approximations of the random data process. The design principle and technical realization of fast backward scenario reduction is further illustrated. The number of scenarios is endeavored to be reduced while still retaining the essential features of the scenario tree, which extends the application of stochastic linear programming.(2) A new model for medium term forward contracting determination is proposed based on stochastic linear programming, which considers the uncertainty in inflows and market prices simultaneously with scenario tree model based on different constructing methods. Forward contracting decisions and day-ahead market trading decisions are recognized as different decisions of different stages in a stochastic programming framework. Through the comparison with expected value model, the advantage of higher revenue is guaranteed for considering the influence of uncertainty; through the comparison with a different stochastic programming model, the similarity of influence on forward contract decisions and revenue further verifies the availability of the proposed stochastic linear programming model.(3) A new model for portfolio decisions is proposed based on stochastic linear programming, which considers hydropower scheduling and multi-market trading decisions under uncertainty in inflows and market prices. Portfolio decisions and recourse decisions of generation scheduling are recognized as different decisions of different stages in a stochastic programming framework, and uncertainty is modeled with different scenario tree model, which will result in different portfolio decisions, but would have no influence on entire stochastic linear programming model. The risk assessment on portfolio decisions shows that more revenue is guaranteed through the flexible portfolio decisions. However, portfolio decisions also face great risk, which demonstrates the necessary of risk management.(4) Based on the summary on methods and application of risk management in deregulated market, the deterministic models incorporating price risk into short-term hydropower scheduling problem are proposed, which are expected revenue-VaR risk utility model and integrating scenario risk penalty or risk constraint model. To solve the above models, IFEP-GA hybrid optimization algorithm is proposed, which integrates evolutionary mechanism and competition selection mechanism of IFEP algorithm in GA framework, and penalty mechanism and repair mechanism are combined in the evolutionary process. The considering of price risk on evolution algorithm is analyzed, which provide hydropower producer proper risk level and corresponding short-term optimal scheduling.(5) For the close relation between VaR risk measure and day-ahead spot market trade, a new expected revenue-VaR risk utility model based on stochastic programming is proposed. The solution of different risk factors provides the efficient frontier between the expected revenue and risk. Furthermore, risk will be different under different risk measures, which further provide the decision base for proper risk level.(6) For the close relation between risk measures, model of uncertainty and market trading strategy based on stochastic programming, Semi-Variance risk measure is proposed, which reflects the risk from the difference between price scenario and expected price in a scenario tree model, and a new risk-constrained portfolio management model is constructed, in which uncertainty model and risk model is combined. The influence of uncertainty is reflected and risk is effectively managed, and corresponding flexible portfolio decisions under different risk levels can be obtained.With the application of hydropower scheduling problem and optimization of multi-market trade strategy, the proposed optimization algorithm and mathematical models are verified effectively, which provide decision base for proper trade strategy in deregulated market and effective solution for tradeoff between maximum revenue and minimum risk.更多还原