基于关联规则的火电厂传感器故障检测  

SENSOR FAULT DETECTION IN THERMAL POWER PLANTS BASED ON ASSOCIATION RULE

作　　者：鲍文[1] 于达仁[1] 王伟[1] 徐志强[1] 

机构地区：[1]哈尔滨工业大学,黑龙江哈尔滨150001

出　　处：《中国电机工程学报》2003年第12期170-174,共5页Proceedings of the CSEE

基　　金：国家自然科学基金(59976008)~~

摘　　要：为了克服建模误差对诊断结果的影响,充分利用火电厂的运行数据,提出一种基于关联规则的传感器故障检测的新方法。针对火电厂多传感器系统的特点和火电厂不同的运行工况,充分利用火电厂运行过程中存在的大量的冗余信息,对数据挖掘中的关联规则算法进行了改进。通过对生成的关联规则的评估和仿真,表明此关联规则能够很好地用于传感器故障检测,而且提高了计算速度。A new fault detection approach for sensor in thermal power plants is proposed, which is based on the association rule algorithm of the data mining in order to overcome the bad effect of the diagnosis result as modeling error and make full use of the running data. The improved association rule algorithm of data mining takes full advantage of large redundant information in the process of thermal power plants according to the multi-sensor character and the different running phases. The result demonstrates that the developed algorithm can be used to fault detection of different kinds of sensors and improves the calculation speed with association rules' evaluations and simulation.

关 键 词：火电厂 传感器 故障检测 关联规则 数据挖掘 人工神经网络 

分 类 号：TM621.7[电气工程—电力系统及自动化] TP212[自动化与计算机技术—检测技术与自动化装置]