改进的Adaboost方法及其在水电站设备故障检测中的应用  

Improved Adaboost Method and Its Application in Equipment Fault Detection of Hydropower Station

作　　者：陈涛 张华飞 衣传宝 孙成勋 高阳 徐华雷 

机构地区：[1]国网吉林省电力有限公司电力科学研究院,吉林长春130021 [2]国网新源控股有限公司,北京100761

出　　处：《水力发电》2018年第3期62-65,共4页Water Power

摘　　要：针对水电站运行人员巡检时间过长,检查设备故障效率过低等问题,设计了水电站故障检测方案。根据改进的Adaboost方法对不同工况下机器作用所产生的噪声值进行训练,并建立一个分类器模型,将其应用到水电站设备故障检测方案当中。通过仿真实验,结果表明改进的Adaboost分类器正确率很高,达到89.1%。此方案可以提高水电站设备故障的检测效率,加强了工作人员的安全保障。In view of longer operation personnel inspection time and lower equipment tault check efficiency in hydropower station, a fault detection scheme is designed, in which, the noises generated by the operation of machines in different operation conditions are trained according to improved Adaboost method and a classifier model is set up. The model is applied to equipment fault detection scheme of hydropower station. The simulation experiment results show that the improved Adaboost classifier has a high correct rate of 89.1% . The scheme can improve the detection efficiency of equipment fault of hydropower station and improve the security of staffs.

关 键 词：故障检测 ADABOOST 熵权法 分类器模型 

分 类 号：TM312[电气工程—电机]