【作者】 吴超；

【导师】 王培红；

【作者基本信息】 东南大学 ， 动力机械及工程， 2006， 硕士

【摘要】 电力市场是现代电力改革的重要课题之一。随着“厂网分开,竞价上网”,电力市场的逐步建立,独立发电企业如何在电力市场中充分运用市场规则安排自己的发电计划,从而获得最大的经济效益,目前还是一个全新的领域,无论在理论上,还是实践中都具有重要意义。本文结合国家自然科学基金《热力设备实时数据中知识学习与挖掘算法研究》编号(50376011)和国电南自股份有限公司与东南大学动力系合作开发《发电公司竞价决策系统》两个项目,对发电侧竞价方案的生成及优化进行了研究。论文的主要内容、方法及创新点如下:1、根据《华东电力市场运营规则》对目前华东电力市场的交易规则、月度竞价市场、日前竞价市场作了详细的分析。2、作为国家自然科学基金研究的一部分,本文重点比较和改进了神经网络和模糊神经网络的预测方法和模型,并用于预测各时段系统边际电价,模拟结果表明改进后的模糊神经网络模型可以获得准确且稳定的预测结果。3、为了兼顾预测数值与预测趋势的准确性,本文在模拟试验的基础上,进一步建立了大样本(7天历史数据)预测和小样本(3天历史数据)预测的加权平均模型,有效地改善了边际电价的预测精度。4、本文发现了竞价利润曲线随边际电价变化的规律;建立了机组启停组合优化的数学模型;设计了机组启停组合样本生成规则;改进了遗传优化算法;提出了基于边际电价的全天高利润目标优化算法。并据此生成了全天96点边际电价-最优负荷方案。模拟实验表明此方法能比较快速的找到最优解,且具有良好的稳定性和收敛性。5、在上述的边际电价预测和优化计算基础上,本文根据华东电力市场规则生成了日前电力市场的申报方案(将上述单价格段方案扩展至报价需要的十价格段方案),并利用编程工具开发了优化方案生成软件。文中还给出了软件的主要界面并对软件的使用方法进行了详细的介绍。本文结合华东电力市场规则,对包括电价预测模型分析、预测电价数值处理、日前竞价方案优化、日前申报方案生成的具体方法进行了研究,为解决独立发电公司生产经营中的实际问题提供了一种可行的研究思路,希望能够为进一步的理论与实践的研究提供帮助。更多还原

【Abstract】 The electric market is one of the important projects in the reform of the modem electrical system. With "power plants separated form electrical network, power plants competing electric price", and the electric power market gradually being built-up, it is still a new field that an independent power corporation how to make use of market rules to arrange its own generating scheme in order to achieve maximum economic benefits , which has the significance both in theory and practice.This thesis studies the quote project and optimization of power plant basing following two programs: the National Natural Science Foundation of China (KDD and data mining research in thermal equipment real-time database) with the serial number 50376011, power plant quoting price decision taken charge by Nanjing Automation Co., Ltd and The Department of Power Engineering of southeast university.The main contents, method and innovation are as follows:1. This thesis analyzes electric power market bargain rule, monthly quote market, and day quote market of East China according to east china electric power market rule.2. Being a part research of the National Natural Science Foundation of China, this thesis comparison and improve the prediction method and model of the artificial network and artificial fuzzy network. The result of imitating shows that the improved artificial fuzzy network can acquire an accurate and stable estimate result.3. For the accuracy both of prediction-data and prediction-trend ,this thesis built up the combination average model of great sample (7-day history data) prediction and small sample (3- day history data) prediction on the foundation of imitating test. This improved the estimate accuracy of the SMP availably.4. This thesis summarized the change regulation of the profits curve with system marginal price; built up the mathematical model of unit stop and run; designed the rule of build sample based on unit stop and run; improved genetic algorithm; Put forward the optimize algorithm according to the whole day highest profit target on the base of the system marginal price. Imitate experiment expresses this method can find out the superior solution quickly, and have a good stability and the astringency.5. Basing on the foundation of SMP prediction and project optimization, and according to the east china rules, this thesis born day quote project (according to the single price segment expanded to ten price segment). Programmed tool is utilized to develop day quote assistance decision system in the thesis. In addition, we introduce the main interface and the operation method of this software.Combining with east china electric power market rules, this thesis carry on a research on SMP forecast, SMP process, project optimize, quote project boring. On the other side the thesis can serve as the essential and firm basis of further theoretical research about resolving the actual problems of independently plant, and I hope this thesis provide a help for further theories and practice researching.更多还原