基于改进主成分分析法的火电机组能耗特征识别方法  

An Improved Principal Component Analysis Based Recognition Method for Energy Consumption Characteristics of Thermal Generation Unit

作　　者：马瑞[1] 康仁[2] 罗斌[1] 徐慧明[3] 何进[4] 

机构地区：[1]智能电网运行与控制湖南省重点实验室(长沙理工大学),湖南省长沙市410004 [2]甘肃省电力公司天水供电公司,甘肃省天水市741000 [3]国网信息通信有限公司,北京市宣武区100761 [4]华润电力(常熟)有限公司,江苏省常熟市215536

出　　处：《电网技术》2013年第5期1196-1201,共6页Power System Technology

基　　金：国家自然科学基金项目(51277015)~~

摘　　要：火电机组的能耗特性(发电煤耗率与各能耗参数的特性关系)反映着机组的能耗特征,它是机组节能研究的重要组成内容。机组的能耗特性等重要知识蕴含于机组的运行数据中,理论上机组的能耗特性可以从其历史数据中回归挖掘出来,然而机组的能耗指标不仅数量多,而且相互之间具有多重共线性特点,这给回归分析带来了困难。运用改进主成分分析法的特征提取和降维的功能,以损失少量信息为代价,解决了能耗变量维数过高及其多重相关性给能耗特性方程的回归带来的困难。然后,从能耗特性方程出发,辨识出了各重要能耗指标的灵敏系数。最后,通过对国内某火电厂的600 MW机组进行能耗特征分析,验证了该方法的有效性。The energy consumption characteristics of thermal generation unit, namely the characteristic relation among coal consumption rate for power generation and energy consumption parameters, reflects energy consumption feature of the unit, so it is an important part in the research on energy-saving of generation unit. Important information such as the energy consumption characteristics of generation unit is implied in operational data, thus theoretically the energy consumption characteristics of the unit can be attained by regression in data mining （DM） technique. However, various energy consumption indices of a unit and the multiple collinearities among them make the regression analysis difficult. A method to solve this problem is proposed： firstly, utilizing the functions of feature extraction and dimension reduction in principal component analysis （PCA） and at the cost of losing bits of information, the difficulty in the regression of energy consumption characteristic equation due to too high dimension of energy consumption variables and multiple correlativity among them is solved; then based on the energy consumption characteristic equation, sensitivity coefficients of important energy consumption indices are recognized; finally, the effectiveness of the proposed method is verified by analysis on energy consumption characteristics of an actual 600MW unit of a certain thermal power plant in China.

关 键 词：低碳电力 特征提取 能耗特征 改进主成分分析法 敏感性分析 

分 类 号：TM611[电气工程—电力系统及自动化]