基于模糊粗糙集的大型汽轮机组设备故障识别方法  

Method of Fault Identification for Large Steam Turbine Units Based on Fuzzy Rough Sets

作　　者：莫子孟 尹立平 MO Zimeng;YIN Liping(CHN Energy Ledong Power Generation Co.,Ltd.,Ledong,Hainan,573539)

机构地区：[1]国家能源集团乐东发电有限公司,海南乐东573539

出　　处：《能源科技》2024年第3期44-48,共5页Energy Science and Technology

摘　　要：针对大型汽轮机组设备故障种类多,提出基于模糊粗糙集的大型汽轮机组设备故障识别方法。首先,模糊化处理大型汽轮机组设备故障信息,将复杂的故障信息转化为简单的模糊编码后,使用故障类型-征兆特征决策表生成方法构建特征决策表,表中各行代表故障类型,各列代表故障征兆特征;将决策表数据输入基于改进可拓神经网络聚类的故障分类模型中,决策表的历史数据作为训练数据,当下机组设备运行状态数据作为测试数据,通过判断当下设备运行状态是否与某故障类型-征兆特征决策表的数据匹配,完成设备故障识别。实验中,此方法可有效识别16种汽轮机组设备故障。In response to the numerous types of faults in large steam turbine units,a method for fault identification based on fuzzy rough sets for large steam turbine units is proposed.Firstly,the fault information of large steam turbine units is fuzzified to transform complex fault information into simple fuzzy codes.Then,a feature decision table is constructed using the fault type-sign feature decision table generation method,where each row represents a fault type and each column represents a fault sign feature.Subsequently,the decision table data is input into a fault classification model based on an improved extensible neural network clustering method.The historical data of the decision table is used as training data,while the current operating status data of the unit is used as testing data.By determining whether the current equipment operating status matches the data in a certain fault type-sign feature decision table,equipment fault recognition is achieved.In the experiment,this method can effectively identify 16 kinds of turbine unit faults.

关 键 词：模糊粗糙集 大型汽轮机组 设备故障 决策表 可拓神经网络聚类 故障分类识别 

分 类 号：TM621[电气工程—电力系统及自动化]