【作者】 李建强；

【导师】 刘吉臻；

【作者基本信息】 华北电力大学（河北） ， 热能工程， 2006， 博士

【摘要】 数据挖掘技术是当今智能系统理论的重要研究内容,它综合运用人工智能、计算智能、模式识别、数理统计等先进技术从大量数据中挖掘发现有价值和隐含的知识,因而近年来得到国内外极大的重视。数据挖掘把人们对数据的应用,从低层次的末端查询操作,提高到为经营决策提供决策支持。电站运行优化涉及多个领域的问题,是一项复杂的系统工程,是保证机组安全经济运行的重要手段。目前电站运行优化中的关键和难点问题如优化目标值确定、运行优化操作指导等还存在很多不完善之处。随着信息技术的发展,电力企业积累了大量的数据,从热力系统自身的运行数据中找到改善系统运行的知识和手段,是很有现实意义和研究价值的问题。但由于电力数据自身的多元性、动态性与交连性,给人工分析和处理带来困难,为此本文将数据挖掘理论与应用同电站运行优化相结合,将数据挖掘技术引入电站运行优化过程,提出了基于数据驱动的电站运行优化新方法。本论文围绕数据挖掘技术在电站运行优化中的应用主题,对基于数据挖掘的电站运行优化目标值的确定、电站运行优化操作指导、基于数据挖掘的电站运行优化系统的构建等几个方面进行了深入研究。1.对电站运行优化问题,特别是运行优化目标值确定问题进行了深入分析,详细讨论了当前电站运行优化目标值的确定方法,指出当前方法的缺点和不足,提出了基于模糊关联规则挖掘的电站运行优化目标值确定的新方法,解决了传统优化目标值确定中对机组实际状态考虑不足而失去指导意义的问题。2.深入探讨了数据挖掘技术应用于运行优化的关键技术难题,指出了电站运行参数间的关联特性。提出将数据可视化技术引入数据挖掘过程,用于指导有效的数据选择,避免了挖掘的盲目性,为定量关联规则挖掘指明了方向。3.研究了数据挖掘技术的最新发展,对数值型关联规则挖掘进行了深入分析,提出了改进的模糊关联规则挖掘算法并应用于电站优化目标值的确定。4.对确定电站运行参数优化目标值这类复杂的半结构化问题,提出了人机一体化的电站运行优化目标值解决方案,初步构建了基于偏差机理分析的运行优化操作指导系统。5.将数据挖掘与运行优化相结合,提出了基于数据挖掘的电站运行优化系统开发方案,对面向厂级监控系统(SIS)的运行优化系统结构进行了设计,为基于数据挖掘的电站运行优化扩展到整个SIS系统提供了强有力的支持。更多还原

【Abstract】 Data mining is one of important parts of intelligence system. For its mighty function in finding valuable and potential knowledge among large amount of data by means of many advanced technologies such as artificial intelligence, pattern recognition and mathematical statistics, it has been widely developed in many fields. Data mining improves the application of data from primary searches to advanced decision making.Power plant operation optimization involves many fields and it is an important method to keep the unit operating safely and economically. But up to now, many noduses have been left unsolved such as decision of optimal target value and optimization operation guide.With the development of power plant information technology, more and more field data were saved to MIS or SIS. Power enterprise gathered enough data and abundant knowledge was contained in these data. It will be valuable and significative if we can find knowledge from operation data to improve the operation level of the system. While power plant operation data is complex, dynamic and interactive, it is difficult to analyze data artificially. So this paper proposed a new method based on data mining to optimizing operation. The new method combined pivotal data mining theory and power plant operation optimization and induced data mining technology to operation optimization process. Aiming at the application of data mining in power plant operation optimization, this paper mainly studied the following three problems: the decision of optimal target value based on data mining, operation guide based on variance analysis, the construction of power plant operation optimization system.1. Reviewed and analyzed power plant operation optimization deeply, especially the decision of optimal target value. This paper discussed current methods to decide optimal value and analyzed their disadvantage. Then proposed a new method based on fuzzy association rule to decide optimal target value. This method took full considerations of virtual condition, so it could be used to instruct operation. And because the value came from real operation data, it was realizable in operation.2. Analyzed the problems of operation optimization based on data mining in detail. Indicated the importance of data selecting and in order to avoid blindness in更多还原