【作者】 郑西西；

【导师】 谷俊杰；

【作者基本信息】 华北电力大学 ， 热能工程， 2011， 硕士

【摘要】 火电机组运行优化目标值是关系到机组经济性的重要因素,它提供反映机组当前最佳运行状态的运行参数和性能指标,为运行优化操作指导提供了基础和依据。传统的优化目标值确定方法如设计值、变工况热力计算往往不能很好反映机组的实际运行状态。随着火电厂自动化程度的提高,其生产过程中数据采集和存储技术的有了长足发展,机组运行积累数据的规模和质量进一步提高,这些生产数据中蕴含着机组优化所需的大量状态信息。本文将关联规则挖掘应用于火电厂优化目标值的确定,在经典算法基础上,针对火电厂生产数据的特点,提出了基于哈希散列和数据表消减的改进算法,该算法利用哈希表的快速查找特性,通过直接扫描数据表生成候选项集,将项集及其支持频次存入哈希表,同时删除表中不需要的记录完成数据表的初步消减,扫描完成后移除哈希表中不满足最小支持频次的记录以产生频繁项集,并依据该轮次生成的频繁项集来完成对数据表的再次消减,从而减少后续的计算量,提高定量关联规则挖掘的效率。另外,本文给出了改进算法软件实现的总体结构框架以及采用的主要数据结构,最后对某1000MW机组的历史生产数据挖掘得出100%负荷下各参数的优化目标值,并对结果进行分析。结果表明,该方法获得的机组优化目标值与机理分析相一致,可以用来指导优化运行。更多还原

【Abstract】 Optimization target is an important factor related to economic of thermal power unit. It provides operation parameters and performance indexes which reflect the best running of unit, and also provides the foundation and basis of optimization operation guide. Traditional determination of optimization target, such as design value, varying condition calculation are often can’t reflect the actual states of unit. With the continuous development of data acquisition and storage technology, the scale and quality of production data of thermal power plant are significantly improved, and the production data contains a large number of information required by operation optimization. This paper applies association rules to determine the optimization target of power plant, and proposed an improved algorithm based on hash table and data reduction according to the features of the production data of thermal power plant. The improved algorithm generates candidate itemsets by scanning a data table directly, and then pushes the itemsets and their support to a hash table, completes the first round data reduction by deleting needless records in the table simultaneously. After the scan is complete, the algorithm generate frequent itemsets by removing the records that don’t meet the minimum support from the hash table, and completes the second round data reduction according to the frequent itemsets for reducing the follow-up computation and improving the efficiency. In addition, this paper presents an overall framework and main data structure of a software implementation of the improved algorithm. As a practical example, optimal values for 100% load of a certain 1000MW power set are found with the software. The analysis of the results show it’s consistent with mechanism analysis, and can raise the power set’s efficiency.更多还原