| 摘要: |
| 在设备状态监测过程中引入数据自组织挖掘思想,建立一种设备状态退化预警方法。采用隐马尔科夫模型(HMM)对设备的早期退化状态进行准确辨识和评估,并进一步建立设备退化过程的自组织预测模型。案例分析中将该方法应用到旋转机械轴承运行状态退化的预警过程中。结果表明,基于自组织数据挖掘的设备状态退化趋势预测方法预测效果准确、客观性强,预测值与实际值的拟合程度高,相对误差仅为3.1％。新方法能够预测设备未来时间段的退化状态及其发展趋势,提前给出预警信息,有效地制定预知维修计划,及时采取预防措施,防止因设备突发失效引起非计划停机造成生产和经济损失。 |
| 关键词: 数据自组织挖掘 隐马尔科夫模型 数据分组处理方法 状态退化预警 |
| DOI：10.3969/j.issn.1673-5005.2014.03.023 |
| 分类号:TH 17 〖HTH〗 |
| 基金项目:国家自然科学基金项目(51104168); 教育部新世纪优秀人才支持计划项目(NCET-12-0972); 北京市自然科学基金项目(3132027); 中国石油大学(北京)科研基金项目(YJRC-2013-35); 北京市优秀博士学位论文指导教师科技项目(YB20111141401);中国石油天然气集团公司科学研究与技术开发项目(2012B-3407) |
|
| An early warning method of degradation for mechanical facilities based on data self-organization mining technology |
|
HU Jin-qiu, ZHANG Lai-bin, HU Chun-yan, LI Wen-qiang
|
|
(Faculty of Mechanical and Oil-Gas-Storage and Transportation Engineering in China University of Petroleum, Beijing 102249, China)
|
| Abstract: |
| Data self-organization mining technology was introduced during facility condition monitoring process, and an early warning method of degradation for facilities was developed. Hidden Markov model (HMM) was used to identify and assess the early degradation state of the facility, and the predictive model was further developed to predict the future degradation trend. In the case study, the proposed method was applied to bearings in the rotating machinery. The results show that the effectiveness, objectivity and accuracy of this method are validated by the test results. The predictive states are consistent with the actual situation, and the relative error is only 3.1%. In this way, the early warning of the degradation states can be given to make engineer carry out appropriate maintenance strategies effectively and timely, which can avoid production and economic losses due to unplanned shutdown of machine. |
| Key words: data self-organization mining hidden Markov model (HMM) group method of data handling (GMDH) early warning of degradation |
