【作者】 王金凤；

【导师】 李渝曾；

【作者基本信息】 上海大学 ， 控制理论与控制工程， 2012， 博士

【摘要】 电能不同于一般商品,不能大规模有效地储存,并且要求供需双方瞬时平衡,由于供需双方的突变不能通过储量来平衡,因而电力市场均衡价格呈现强烈的易变性。电力市场中电价的易变性使得各个市场参与者都面临巨大的利益损失风险,因此越来越多的市场参与者认识到电力市场中风险管理的重要性,并积极采用合适的风险管理工具和方法来回避或控制风险。本文重点介绍了新的风险测量因子——条件风险价值(Conditional Value-at–Risk,CVaR),并分别采用现代投资组合方法和机会约束规划方法,探讨了供电公司的风险测量与管理问题。基于pool交易模式,本文的主要研究工作及创新点包括以下五个方面:第一:引入新的风险测量和管理工具。CVaR是近期发展起来的一个新的风险测量工具,具有一致性测量和不依赖于投资回报符合正态分布的假设等良好的数学特性。由于电价具有均值回复、波动性、跳跃特性以及周期性等特点,将CVaR应用于电力市场风险评估与管理,具有重要的现实意义。第二:基于投资组合方法和CVaR,建立了供电公司在实时市场、日前市场和长期合同市场间购电决策模型,可以为供电公司在多个市场间购电比例优化分配和风险评估提供决策方法和依据。重点分析了各种风险条件下,各市场的购电份额、供电公司的收益以及相应的风险大小。该模型可以转化为线性规划求解,避免了其他模型复杂的求解过程以及可能不收敛的情况。算例的计算结果表明了模型的有效性:供电公司可以利用长期合同规避市场风险,同时期望收益也会下降;CVaR能较为真实地反映供电企业面临的风险大小,供电公司针对风险的态度(即置信水平)不同,收益也不同。第三:考虑期权交易,建立了供电公司的购电模型。电力期权合约比远期合同和电力期货交易具有更高的灵活性,但由于期权的购买费用关于日前现货价格是不连续的,因此用解析的方法求解最大化组合收益有一定的困难;引入风险测量因子CVaR可以将该非线性问题转化为线性规划问题求解。基于此,本文建立了供电公司在合同市场、现货市场和期权市场间的购电组合模型,重点分析了有无期权、期权价格、期权敲定价格等因素对购电组合收益和风险的影响。算例结果证明了电力期权的参与可以有效地降低购电组合的风险,并且期权价格和敲定价格对购电组合也有比较明显的影响。对期权的合理使用可以使供电公司在避免高电价风险的同时也能从低电价中获益。第四:机会约束规划主要针对约束条件中含有随机变量,且必须在观测到随机变量的实现之前做出决策的情况。与此相似,供电公司与用户签订可中断负荷合同时也无法观测到未来某一时刻负荷和电价的变化情况。本文首次利用机会约束规划方法,建立了供电公司实施可中断负荷时的风险分析定价模型,模型以期望收益最大为目标,考虑了风险约束、中断量约束、中断费用约束等,采用基于随机模拟的遗传算法求解。算例结果表明:随着回避风险程度的增加,供电公司从可中断负荷中的得到的收益也就越小,这与现实中高风险高回报的原则一致;同时用户的收益由于中断费用和中断量不同所得收益有很大的差别,但中断量越大收益越大。第五:可中断负荷在削峰填谷、备用市场以及阻塞管理等方面发挥越来越重要的作用。本文把不同类型的用户看作具有不同风险和收益特征的子市场,采用投资组合的方法建立了供电公司实施可中断负荷时的风险分析定价模型,该模型可以帮助供电公司进行可中断负荷方案的决策。与机会约束规划模型相比,本模型不但可以考虑市场条件下的多种不确定因素以及供电公司对风险的偏好程度,而且风险也能用VaR和CVaR明确表示出来,便于供电公司更加直观地比较风险的大小。更多还原

【Abstract】 The differences between electrical power and ordinary commodities are that it can not store effectively and need instantaneous balance between the supply and the demand. So the equilibrium prices in power market are strongly volatile. The volatility of price makes the participators face huge risks and so more and more participators recognize the importance of risk management. Various risk management tools and methods are adopted to avoid or control risk.In this paper, Conditional Value-at–Risk (CVaR), a new risk measure tool, is introduced. At the same time, modern portfolio theory and chance-constrained programming method are presented to research the risk management problem of power supplying company. Based on pool trade model, the main work and the key contributions of this dissertation are as follows:Firstly, a new risk measure and management tool is introduced. CVaR is a new risk measure tool and it has goods mathematics characteristics, such as consistent measure, not depending on the normal distribution of portfolio profit, et al. Due to the mean-profit, volatility, jumping and periodicity of power price, it is of momentous current significance to introduce CVaR to risk assessment and risk management of power market.Secondly, based on modern portfolio theory and CVaR, an energy purchasing model is built among real-time market, day-ahead market and contract market, which can supply a reference to power supplying company who purchase energy among different markets and assess purchasing risks. The research emphesis is focused on purchasing account in various markets, portfolio profits and risk values under different risk levels. This model can be converted to a linear programming problem to solve and so it can avoid the complex and non convergence of other models. An example is used to illustrate the validity of the proposed model. It shows that forward contracts can be used by power supplying company to hedge risks, at the same time the expected profit is reduced. It also shows that CVaR can reflect really the risk faced by power supplying company,and who has different profits with different confidence levels.Thirdly, accounting for options, a power purchasing model is introduced. Electric options have more flexibility than forward contracts and futures. Since the purchasing cost for options is not continuous with the day-ahead price, it is different to obtain a maximum profit of a portfolio analytically. But it can be changed to a linear model due to the introduction of CVaR. Based on this, an energy purchasing model is presented among contract market , day-ahead market and options market, and the research focus is put on the effects of options,option price and strike price on power portfolio profit and risk value. A numerical example has demonstrated that electrical options can reduce risks of portfolio effectively, and the price and strike price of options have obvious effects on the portfolio, too. It can not only avoid risks from high price but also get return from low price to use electrical options appropriately.Fourthly, Chance-constraint programming is aimed at the case which has a subject with stochastic variable, but the decision must be made before the stochastic variable can be observed. Similarly, power supplying company can not forecast the variation of electric load and energy price when it signs an interruptible load contract with customer. Based on chance-constraint programming, this dissertation presents a risk analyzing and pricing model of interruptible load for power supplying company. The model has a object of maximum expected profit and is subjected to risk, interruptible account and interruptible price, et, al. Genetic Algorithm based on Monte carlo random simulation is used to solve this problem. The result has shown that the less profit of power supplying company from interruptible load with the increasing of risk aversion level, and that agrees with the rules: higher profit accompanied by higher risk. At the same time, customers’profits have obvious difference due to variant interruptible price and interruptible amount, more profits with more interruptible amounts.Finally, Interruptible load plays a more and more important roles in peak load shaving, reserving market, congestion management and et, al. This dissertation looks different consumers as sub-markets with different risks and profits, and based on modern portfolio theory, a risk analyzing and pricing model of interruptible load for power supplying company is built, which can help power supplying company to choose interruptible scheme. Compared to chance-constraint programming model, this model can not only consider various indeterminacies under electric market and the preference of power supplying company, it can also give a specific risk value with VaR (Value at Risk) and CVaR. This helps power supplying company to compare visually risks of different decisive schemes.更多还原