| 引用本文: | 董洁,游培航,彭开香.基于动态内部主元分析和隐马尔科夫模型的动态过程故障检测与分类方法[J].控制理论与应用,2020,37(10):2073~2080.[点击复制] |
| Peng Kaixiang,Dong Jie,You Peihang.Dynamic processes fault detection and classification based on dynamic-inner principal component analysis and hidden Markov model[J].Control Theory and Technology,2020,37(10):2073~2080.[点击复制] |
|
| 基于动态内部主元分析和隐马尔科夫模型的动态过程故障检测与分类方法 |
| Dynamic processes fault detection and classification based on dynamic-inner principal component analysis and hidden Markov model |
| 摘要点击 2758 全文点击 1107 投稿时间：2020-01-04 修订日期：2020-03-05 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2020.00007 |
| 2020,37(10):2073-2080 |
| 中文关键词 工业动态过程 故障检测 故障分类 DiPCA HMM |
| 英文关键词 industrial dynamic process fault detection fault classification DiPCA HMM |
| 基金项目 国家自然科学基金 |
|
| 中文摘要 |
| 随着工业生产过程的扩大, 保证生产过程的安全平稳高效运行日益受到重视. 因此, 对工业过程进行及时有效的监测与故障诊断具有重要意义. 一般而言, 工业过程采集的数据具有较强的动态性, 有效提取数据中的动态信息并进行分析极其重要. 本文基于动态内部主元分析(DiPCA)进行动态性分析并结合隐马尔科夫模型(HMM), 提出了一种新的故障诊断框架, 实现了动态过程故障检测与故障分类. 首先, 利用DiPCA算法提取正常工况下数据的动态特征; 然后, 利用HMM能够有效处理时序数据的特点, 对所提取的动态特征进行建模, 构建了动态过程的故障检测框架; 并利用HMM强大的模式分类能力, 对故障数据进行建模, 实现故障的分类; 最后, 将提出的方法用于田纳西-伊斯曼过程, 验证了该方法的有效性与优越性. |
| 英文摘要 |
| With the expansion of industrial production process, it is meaningful to ensure the safety, stability and efficiency of the production process. Generally speaking, data collected from industrial processes often contain strong dynamicity. Thus it is extremely important to extract and analyze the dynamic information of process data. In this paper, a new fault detection and classification framework for dynamic process is proposed. Firstly, the dynamic inner principal component analysis is used to extract the dynamic relationship among the normal data; then, dynamic features are modeled to build the fault detection framework for dynamic process using hidden markov model; and the fault data are also modeled by hidden markov model to build the fault classification model; finally, the Tennessee Eastman process is used to test the effectiveness and superiority of the proposed framework. |
|
|
|
|
|