摘要
将基于监督学习的方法引入到对设备的早期故障预测之中,针对现有预警技术的不足,完善了基于监督学习的预警技术,并提出了快变报警、缓变报警、趋势报警和轴心轨迹报警的报警方式。通过自学习阈值以及针对性的报警策略可以准确地预测报警的发生,并且可以进行有效的故障诊断。在此基础上开发了预警管理系统,该系统能够较好地捕捉到设备的早期故障,并可以对设备的早期故障进行准确的数据分析,对预防和诊断设备故障具有重要的意义。
The method based on supervised learning is introduced into the forecast of early fault on devices.Aiming at the lack of existing early warning technology based on supervised learning is improved,and fast-changing alarm,slow-changing alarm,trend alarm and axis track alarm are proposed.The method can predict occurrence of alarm accurately through self-learning,and diagnosis fault effectively.The early warning management system is developed based on this method,the system can capture the early fault on the device effectively,and analyze data for the early fault accurately,which has great significance for the prevention and diagnosis of device＇s fault.
出处
《电子设计工程》
2011年第12期8-10,14,共4页
Electronic Design Engineering
基金
国家高技术研究发展计划(863计划)专题课题(2007AA04Z433)
国家自然科学基金重点资助项目(50635010)
关键词
状态监测
预警
监督学习
早期故障
condition monitoring
early warning
supervised learning
early fault
