摘要
为了避免电厂设备在运行期间出现异常状态直接或间接导致机组停机增加维护成本,提出了一种基于大数据技术的设备状态评估和预警方法。多元状态估计技术是该方法实现设备故障诊断和健康管理(PHM)的可行技术之一,它的实现依赖海量健康数据的训练学习。基于大数据技术对历史状态数据离线学习并训练健康状态评估模型,针对目标设备实时分析相关参数的残差值变化,通过滑动窗口残差统计法自动检测偏差情况,实现目标设备异常状态的在线监测。以某电厂火电机组的制粉系统为例进行状态评估和健康诊断研究,引入参数贡献率来表征引起异常的强弱因素,进一步推进了对设备状态和故障问题的分析,试验结果表明该方法能够有效地进行电厂设备状态评估和设备故障预警。
In order to avoid the increase of maintenance cost in power plants directly or indirectly resulting from the abnormal conditions and shutdown of equipment during operation,a state assessment and early warning application for power plant equipment based on big data technology is proposed.Multivariate state assessment is one of the feasible technologies to realize equipment Prognostic and Health Management(PHM),and its implementation relies on training and learning of massive health data.Offline training of historical status data is made to establish health state assessment model based on big data technology.Making real-time analysis on the changes of related parameter residual values of the targeted equipment and taking automatic detection by sliding window residual statistics method can realize online monitoring on abnormal status of targeted equipment.Taking the state assessment and health diagnosis of the pulverizing system in a thermal power plant as an example,the parameter contribution rate is introduced to characterize the strength and weakness of factors leading to the anomalies,which is helpful in making further analysis on equipment status and fault.The experimental results show that this method can effectively evaluate the state and make fault warning for power plant equipment.
作者
胡杰
唐静
谢仕义
王云
HU Jie;TANG Jing;XIE Shiyi;WANG Yun(College of Mathematics and Computer Science,Guangdong Ocean University,Zhanjiang 524088,China;College of Information Engineering,Beijing Institute of Petrochemical Technology,Beijing 102617,China;Yuanguang Software Company Limited,Zhuhai 519085,China;Huadian Environmental Protection System Engineering Company Limited,Beijing 100070,China)
出处
《华电技术》
CAS
2020年第2期1-6,11,共7页
HUADIAN TECHNOLOGY
基金
广东省教育厅创新强校项目(Q18286)
中海油能源发展股份有限公司科技项目(CY-2J-19-2C-005)。
关键词
多元状态评估
大数据技术
故障诊断
贡献率分析
制粉系统
multivariate state estimation
big data technology
fault diagnosis
contribution rate analysis
pulverizer system
作者简介
胡杰(1993-),女,河南信阳人,在读硕士研究生,从事大数据技术等方面的研究工作(E-mail:1793961720@qq.com);唐静(1984-),女,湖北武汉人,副教授,工学博士,从事信息科学和人工智能等方面的研究工作(E-mail:tangjing1205@126.com);谢仕义(1963-),男,四川巴中人,教授,工学硕士,从事数字海洋、物联网与大数据技术等方面的研究工作(E-mail:shiyixie@126.com)。
