电站关联规则的主元分析挖掘方法及传感器故障检测  

Association Rules Mining Based on Principal Component Analysis and Sensor Fault Detection of Power Plant

作　　者：邱凤翔[1] 司风琪[1] 徐治皋[1] 

机构地区：[1]东南大学能源与环境学院,江苏省南京市210096

出　　处：《中国电机工程学报》2009年第5期97-102,共6页Proceedings of the CSEE

摘　　要：提出了基于主成分分析的相似关联规则的数据挖掘方法,并利用最小二乘支持向量回归方法对传感器进行故障检测。通过主成分分析寻找具有相似关联规则的参数,利用参数间的相似关联关系,建立最小二乘支持向量回归模型,通过该模型生成残差对传感器进行状态监测和故障定位,并对故障数据进行重构,代替故障数据。通过某300MW机组数据实例分析,表明该方法能准确快速地寻找具有较高相似关联规则的参数,并能给出可信的重构数据,具有一定的实用性。Similarity association rules data mining method based on principal component analysis （PCA） was proposed. Similarity association rules parameters was found through PCA and a least squares support vector regression （LS-SVR） model that detects sensor fault was built. Then the sensor fault location was implemented on the base of the reconstruction residuals from the LS-SVR model, which using the relationship of these similarity association rules parameters. Data reconstruction was implemented by the LS-SVR model instead of fault data. Data from a 300 MW unit were validated by the proposed method. The result reveals that the method can find high similarity association rules parameters fast and effectively. The LS-SVR Model can locate the sensor fault and get credible reconstruction data by using of the relationship of these similarity association rules parameters.

关 键 词：主成分分析 关联规则 最小二乘支持向量回归火电厂 传感器 故障检测 

分 类 号：TM621[电气工程—电力系统及自动化]