基于数据挖掘的煤电机组能效特征指标及其基准值的研究  

Research on Indexes of Energy Efficiency and Its Reference-value for Coal-fired Power Units Based on Data-mining

作　　者：徐婧[1] 顾煜炯[1] 王仲[1] 李倩倩[1] 杨楠[1] XU Jing GU Yujiong WANG Zhong LI Qianqian YANG Nan(National Thermal Power Engineering ＆ Technology Research Center, North China Electric Power University, Changping District, Beijing 102206, China)

机构地区：[1]华北电力大学国家火力发电工程技术研究中心,北京市昌平区102206

出　　处：《中国电机工程学报》2017年第7期2009-2015,共7页Proceedings of the CSEE

基　　金：中央高校基本科研业务费专项资金项(2014XS32;2015XS88)~~

摘　　要：煤电机组的优化运行对于节能降耗有重要的意义,如何合理地选择能效特征指标并确定其基准值是机组优化运行的关键。基于数据挖掘算法,对机组历史运行数据分析:利用灰色关联度分析法,选取对供电煤耗产生主要影响的能效特征指标;采用K均值算法对多特征参数进行聚类划分并确定其基准值;结合广义回归神经网络,预测供电煤耗在基准工况下的目标值;对离散基准值样本点进行样条插值,建立机组全工况下的动态基准值工况库。最后,选取某超超临界百万湿冷机组运行数据分析与验证,分析结果表明提出的能效指标基准值研究方法,可以为煤电机组的优化运行提供调整方向。The operation optimization of the coal-fired power is of great significance to energy saving and consumption reduction. And how to choose the indexes of energy efficiency and determine its target-value is the key point to the optimized operation of the units. Based on the data mining algorithm, the historical operation data was analyzed： The energy efficiency features related to coal consumption of electricity supply was selected by grey correlation degree analysis algorithm; The target-value of the multi feature parameters in typical operating conditions was determined by K-means clustering algorithm; Combined with generalized regression neural network, the target value of the coal consumption of electricity supply under the reference condition was predicted; And the unit dynamic reference condition database under all conditions was established by spline interpolation for the sample points of the given target-value. Historical data of an ultra-supercritical water cooling unit with 1000MW capacity was adopted as an instance to analyze, and results show that the proposed system can provide the adjustment direction for optimizing operation of the coal-fired power unit.

关 键 词：煤电机组 优化运行 数据挖掘 能效特征指标 基准值 

分 类 号：TK01[动力工程及工程热物理]