| 引用本文: | 雷海,曹健,李擎.改进弗雷歇距离的AROMF及在流程工业故障诊断应用[J].控制理论与应用,2026,43(4):832~842.[点击复制] |
| LEI Hai,CAO Jian,LI Qing.Improved Fréchet distance based AROMF and its application in process industry fault diagnosis[J].Control Theory & Applications,2026,43(4):832~842.[点击复制] |
|
| 改进弗雷歇距离的AROMF及在流程工业故障诊断应用 |
| Improved Fréchet distance based AROMF and its application in process industry fault diagnosis |
| 摘要点击 118 全文点击 24 投稿时间：2024-05-22 修订日期：2025-05-21 |
| 查看全文 查看/发表评论 下载PDF阅读器 HTML |
| DOI编号 10.7641/CTA.2024.40291 |
| 2026,43(4):832-842 |
| 中文关键词 故障诊断 弗雷歇距离 局部保留投影 田纳西–伊士曼过程 |
| 英文关键词 fault diagnosis Fréchet distance locality preserving projections tennessee eastman process |
| 基金项目 北京信息科技大学校科研基金项目(2023XJJ10)资助. |
|
| 中文摘要 |
| 针对复杂的动态流程工业过程中故障模式匹配因曲线长度或参数化差异导致的匹配难题, 本文提出了一
种新颖的基于弗雷歇距离的自适应秩序形态滤波器集成局部保留投影的故障诊断方法. 首先, 在本文所提出的基
于弗雷歇距离的自适应秩序形态滤波器集成局部保留投影(LPP-FAROMF)方法中, 保留数据局部流形信息的局部
保留投影(LPP)算法, 用来提取出与故障相关的特征信息, 解决了数据的高维和非线性问题, 进而得到所需的模板信
号和测试信号; 其次, 将弗雷歇距离引入自适应秩序形态滤波器(AROMF)的距离误差中, 用来对测试信号进行基于
弗雷歇距离的自适应顺序形态滤波器(FAROMF)变换得到输出趋势特征, 该方法降低了对曲线长度和参数化的一
致性要求, 而且可以综合考虑空间相似性和动态特征, 因此计算输出趋势与模板趋势之间的误差更加准确且合理;
最后, 田纳西–伊士曼过程被用来验证所提出方法的性能. 结果表明, (LPP-FAROMF)在故障诊断中表现出更高的准
确度、鲁棒性和速度. |
| 英文摘要 |
| Addressing the challenge of fault pattern matching in complex dynamic process industries due to variations
in curve length or parameterization, this paper presents a novel Fréchet distance-based adaptive rank-order morphological
filter (FAROMF) integrated with locality preserving projections (LPP) for fault diagnosis. Firstly, in the proposed method,
the LPP algorithm preserves the local streaming information of the data is used to extract the fault-related features, the
high dimensionality and nonlinearity of the data can be overcome to obtain the required template signals and test signals.
Secondly, the Fréchet distance which is introduced into the distance error of AROMF is used to obtain the output trend
characteristics from the FAROMF transform of the test signal, which reduces the consistency requirements of curve length
and parameterization, considers the spatial similarity and dynamic characteristics, so the error between the output trend
and the template trend is calculated more accurately and reasonably. Finally, Tennessee Eastman process (TEP) is used to
validate the performance of the proposed method. The results show that LPP-FAROMF exhibits higher accuracy, robustness
and speed in fault diagnosis. |
|
|
|
|
|