【作者】 孙力勇；

【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， 2007， 博士

【摘要】 全世界的电力工业正在由传统垄断管制型向竞争监管型转变,打破垄断、解除管制、引入竞争机制已经成为各国电力工业发展的整体趋势,因而对电力市场相关的经济技术问题提出了许多新的要求。传统垄断管制模式下,调度机构从系统的安全性、经济性角度出发直接安排发电公司内部运行;而在电力市场环境下,作为独立经济实体的发电公司,要以利润最大化为目标进行公司内部的经济运行安排。尤其是在短期电力市场中,发电公司经济运行所涉及的发电计划安排、竞价决策模型的建立和求解、报价方案制定、报价决策的风险管理等问题,更是直接影响到发电公司的收益状况。本文致力于电力市场环境下发电公司短期经济运行问题的研究,并取得了一些创新性的研究成果,可概括为:(1)在发电计划优化算法方面,提出了采用实数矩阵表示发电计划的表达方式,从而避免了分层求解机组启停和经济负荷分配子问题,而可以直接采用遗传操作或粒子群寻优进行求解;通过对遗传个体或粒子调整的方式来处理约束条件,切实地保证了优化结果的可行性;根据遗传操作和粒子群算法的各自特点,分别采用窗口变异和贪婪搜索等措施来提高算法性能。此外,本文还进一步以矩阵实数编码遗传算法(MRCGA)为基础,提出了发电公司在主能量和辅助服务市场中联合竞价时,考虑自身发电计划和机组边际收益的报价方案制定方法。(2)在计及风险因素发电公司竞价策略的研究上,从行为金融学中SP/A理论的决策思想出发,用数学语言描述了按统一出清价结算时发电公司在报价方案的安全性和潜力性上的要求,并结合经济学中的机会成本理论分析、描述了发电公司对利润分布形式的要求,进而建立起以自身发电计划为基础的发电公司面对风险时的竞价决策模型,而后结合MRCGA算法进行了求解。针对在按报价结算时,发电公司因报价段的不同而具有不同的风险态度和利润追求目标的特点,建立了以行为金融学中安全优先和双重心理帐户资产组合理论为基础的风险决策模型,该模型不仅以自身发电计划为基础,而且综合考虑了发电公司对报价方案在安全性和潜力性上的双重要求。(3)在发电公司竞价决策的风险管理上,提出了一种基于改进风险价值法(VaR)进行发电报价方案风险度量的新方法,同时在报价方案的风险控制上,建立了基于改进VaR机会约束的发电公司竞价决策模型和报价方案调整模型,并简单地探讨了求解方法。所建立的发电竞价模型和报价方案调整模型中,由于采用基于改进VaR机会约束的形式区别对待不同报价段中发电公司的风险偏好,因此为计及风险偏好发电公司竞价策略的研究提出了一个新方向。更多还原

【Abstract】 The electric power industry has experienced or has been experiencing an unprecedented reformation of deregulation to traditional monopoly and introduction of competition in many countries all over the world. Many requirements about the economic and technological issues in power market are presented. In traditional power systems, power dispatch organization plan generation schedule of the generation company (GENCO) according to the security and economic of power systems, whereas in the power market the independent GENCO has to cope with the economic operation which includes planning the self-schedule, establishing the generation bidding model and solving, controlling the risk of bidding strategy and etc. Therefore, this dissertation devoted to the research on short-term economic operation of GENCO in power market. The major research work is outlines as follows:(1) In the aspect of generation schedule optimization, the real number matrix is used to represent generation schedule, and therefore genetic operation or particle swarm optimization can solve the unit commitment (UC) problem, avoiding the disposal of the suboptimal economic dispatch (ED) problem. The new repairing mechanism that is used to adjust the chromosome or particle, guarantees that the generation schedule satisfies system and unit constraints. The window mutation and the greedy searching based on Priority list (PL) are applied to improve the convergence of genetic algorithm and particle swarm optimization respectively. Moreover, a new approach for solving optimal bidding strategies based hourly self-scheduling in a GENCO that participates in energy and ancillary service markets is presented. The matrix real-coded gentic algorithm (MRCGA) is used to solve the price-based unit commitment problem. (2) In the aspect of GENCO bidding strategy study, a new risk decision-making model of GENCO’s bidding in System-Margin-Price power market based on the SP/A (Security, potential and aspiration) theory is built. In the model, the opportunity cost of economic theory is used to analyze and evaluate the potential of GENCO’s biding strategy, and the security and aspiration are also analyzed and described mathematically. Because of considering the decision psychology and the self-scheduling, the proposed model is close to the actual decision-making process of GENCO. Furthermore, GENCO has different risk attitude and profit purpose towards different bidding blocks during the course of bidding decision in Pay-as-Bid power market, and accordingly the safety-first and the two mental accounts portfolio theory are introduced to model the psychology of GENCO, and a new risk decision-making of GENCO’s bidding strategy is built also. This model is close to the actual decision-making process of GENCO in Pay-as-Bid power market too.(3) In the aspecnt of risk management of GENCO’s bidding, a new risk evaluation method which is based on value at risk (VaR) is presented. Moreover, stochastic decision-making model of GENCO’s bidding strategy and the model of bidding strategy adjustment are proposed also with using the presented method. Because of risk attitude of GENCO being expressed by chance constraints based on the method, a new aspect which research on GENCO’s bidding strategy considering market risk is advanced.更多还原