振动趋势判别云模型的故障诊断方法
Fault Diagnosis Method Based on Vibration Trend to Discriminate Cloud Model
张栋良 1洪勤勤 2汪刘峰 2张凯文2
作者信息
- 1. 上海市电站自动化技术重点实验室,上海 200090;上海电力大学自动化工程学院,上海 200090
- 2. 上海电力大学自动化工程学院,上海 200090
- 折叠
摘要
在汽轮机故障诊断领域,序列数据的变化趋势能够反映振动过程中的运行状态和发展态势,是专家在诊断时经常使用的特征依据.由于汽轮机组本体因结构、工况等导致的故障样本多样性和稀缺性以及专家经验和定性描述相对丰富的诊断现状,提出了一种基于云模型的汽轮机振动时间序列趋势判别方法.通过总结专家经验和故障案例,结合不确定性云模型生成定性趋势的云参数评估模型;利用样本数据通过逆向云得到的云参数生成大量云滴,代入云参数评估模型计算趋势等级确定度;引入趋势判别决策树得到序列数据的定性描述.最后以某亚临界双排汽凝气式汽轮机为研究对象,验证了该方法的可行性和有效性.
Abstract
In the field of turbine fault diagnosis,the variation trend of series data can reflect the operating state and development trend in the vibration process,which is the characteristic basis often used by experts in diagnosis.Due to the diversity and scarcity of fault samples caused by the structure and working condition of the turbine unit,as well as the relatively rich experience of ex-perts and qualitative description of the diagnosis status,a cloud model-based method for identifying the trend of turbine vibra-tion time series is proposed.By summarizing expert experience and fault cases,the cloud parameter evaluation model of qualita-tive trend is generated in combination with the uncertain cloud model.A large number of cloud drops are generated by cloud pa-rameters obtained by reverse cloud based on sample data,and the trend level determination is calculated by substituting the cloud parameter evaluation model.The trend discriminant decision tree is introduced to obtain the qualitative description of series data.Finally,the feasibility and effectiveness of the method are verified by taking a subcritical double-exhaust condensing steam tur-bine as the research object.
关键词
定性趋势分析/云模型/趋势判别/故障诊断Key words
Qualitative Trend Analysis/Cloud Model/Trend Discrimination/Fault Diagnosis引用本文复制引用
基金项目
上海市自然科学基金(15ZR1418300)
出版年
2023
NETL
NSTL
维普
万方数据