【作者】 祝全乐；

【导师】 苑津莎； 刘建新； 马新顺；

【作者基本信息】 华北电力大学 ， 电工理论与新技术， 2012， 博士

【摘要】 竞争电力市场环境中,发电公司和用户逐渐成为电力市场参与者的主角,其市场行为也受到越来越多的关注。在单一购买联营体市场环境下,日前市场中的发电公司最优竞价策略问题受到发电公司本身和市场监管部门两方面的重视,一直以来都是电力市场领域一个研究的热点。另外随着用户侧购电选择权的逐步放开,大用户可以在不同售电商中自由选择购电,但是同时市场中的不确定因素也会给大用户购电带来不同程度的风险。如何面对这些不确定因素,在控制风险的前提下构造合理的购电组合策略成为大用户面临的关键问题之一。本文基于可信性理论,综合考虑市场中的随机和模糊这两类不确定信息,对联营体市场模式下日前市场中发电公司的最优竞价策略问题和大用户直购电市场环境下大用户的最优购电策略问题分别进行了研究。主要研究成果如下：(1)针对以往在研究发电公司最优竞价策略时仅能单独考虑市场中的随机或者模糊不确定信息的问题,基于可信性理论,综合考虑这两类不确定信息,建立了由调度中心优化问题和发电公司最优竞价策略问题所组成两层不确定规划模型,其中为了与实际竞价环境更为贴近,在模型中考虑了网络中的输电容量约束。针对新模型的求解,由于新模型为同时含有随机变量和模糊变量的两层模型,很难通过将其转化为确定性等价模型求解,因此提出了一种将互补直接优化、不确定模拟、神经网络和遗传算法结合在一起的混合求解算法。并分别在3节点系统,11节点系统和30节点系统上对所提出的模型和求解算法进行了计算分析验证。(2)针对基本遗传算法在求解发电公司最优竞价策略时可能会遇到的过早收敛于局部极值的问题(即早熟问题),通过引入不可行种群,并改进评价选择、交叉和变异策略,提出了一种具有较好抗早熟能力且适用于求解带约束规划问题的改进遗传算法。不可行种群的引入和新的评价选择策略能够简化约束处理问题,新的交叉和变异策略能够在保证种群中优良模式得以延续的前提下尽可能地增加种群的多样性,扩大搜索空间,以防止早熟现象的发生。并通过本文算例和经典测试函数验证了新遗传算法的改进效果。(3)基于可信性理论对大用户最优购电策略问题进行了初步研究。综合考虑大用户在主能量市场和备用市场上的购电交易,依据风险价值思想,基于本原机会约束提出了一种计及风险的大用户最优购电策略模型,模型中综合考虑了随机和模糊这两类不确定信息以及这些不确定信息所带来的风险。采用基于随机模糊模拟、神经网络和遗传算的混合算法对模型进行求解。并通过具体算例说明了所提出的风险管理方法和模型的可行性。更多还原

【Abstract】 In competitive electricity market environment, generation companies and consumers are becoming the protagonists of the electricity market participants, and their market behaviors have also attracted increasing attention. Under the single-purchaser pool market model, the problem of the optimal bidding strategies for generation companies in day-ahead market is subjected to attention by both generation companies and market regulators, it has been a research focus in the field of electricity market. In addition, as the electricity purchasing option for the consumer side to gradually open up, large consumers could purchase electricity freely from different power providers. But at the same time the uncertainties in the market will bring different levels of risk to large consumers in their electricity purchasing process. How to face these uncertainties and build reasonable electricity purchasing strategy under the premise of controlling risk become one of the key issues facing large consumers.This dissertation, based on credibility theory and considering the random and fuzzy uncertain information in the market, focuses on researching the optimal bidding strategies for generation companies in day-ahead market under the pool model and the optimal electricity purchasing strategies for large consumers under direct electricity-purchase model for large consumers. Main achievements are presented as follows:(1) According to the problem that random or fuzzy uncertainty information in the market can only be considered separately in the past optimal bidding strategy research, based on credibility theory, comprehensively considering random and fuzzy uncertainty information, builded two level uncertain programming model which is composed of independent system operator optimal model and generation company optimal bidding strategy model. The two level model takes transmission capacity constraints into account for closing the actual auction environment. To solve the new model, proposed a hybrid solution algorithm combined with complementary direct optimization, uncertainty simulation, neural network and genetic algorithm, because the new model is a two level model that contains random variables and fuzzy variables synchronously, it is difficult to be solved by translating the model into certainty equivalent model. Used3-bus system,11-bus system and30-bus system to demonstrate the proposed model and solution algorithm.(2) According to the precocious problem encountered in basic genetic algorithm for solving the generation companies optimal bidding strategies, proposed an improved genetic algorithm based on introducing infeasible population and improving selection, crossover and mutation strategy. The introduction of the infeasible population and the new evaluation and selection strategy can simplify the problem of constraints handling. Under the premise of ensuring the continuity of the population fine mode, the new crossover and mutation strategy can increase the diversity of population, expand the search space and prevent the occurrence of precocious problem. The numerical examples in this dissertation and the classic test functions are used to verify the improved effect of the new genetic algorithm.(3) Based on the credibility theory, researched the problem of optimal electricity purchasing strategy for large consumers preliminary. comprehensively considering electricity purchasing transactions of large consumers in the primary energy market and reserve market, according to the value at risk thought, based on the primitive chance constraint, proposed an optimal electricity purchasing strategy model for large consumers taking risk into account. The random and fuzzy uncertain information, as well as the risks arising from these uncertain information are synchronously considered in the model. The hybrid algorithm based on random fuzzy simulation, neural networks and genetic algorithm is employed to solve the model. Numerical examples illustrate the feasibility of the proposed risk management method and model.更多还原