| 多模态过程的全自动离线模态识别方法 |
| |
| 引用本文： | 张淑美,王福利,谭帅,王姝.多模态过程的全自动离线模态识别方法[J].自动化学报,2016,42(1):60-80. |
| |
| 作者姓名： | 张淑美 王福利 谭帅 王姝 |
| |
| 作者单位： | 1.东北大学信息科学与工程学院 沈阳 110819 |
| |
| 基金项目： | 国家自然科学基金(61533007,61374146,61403072),流程工业综合自动化国家重点实验室基础科研业务费(2013ZCX02-04),中央高校基本科研专项资金(N140404020),华东理工大学探索研究专项基金(22A201514050)资助 |
| |
| 摘 要： | 多模态是复杂工业生产过程的普遍特性.不同模态具有不同的过程特性,需要建立不同的模型,因此离线建模数据的模态划分与识别是整个多模态过程建模的关键问题之一.目前,常用的聚类算法需要对其结果进行人工分析和后续处理,无法真正实现多模态过程的全自动模态识别.因此,本文提出一种全自动的多模态过程离线模态识别方法.首先通过宽度为H的大切割窗口对数据进行切割,利用改进的K-means聚类算法对窗口单元进行聚类;根据聚类结果,对稳定模态淹没现象进行处理,得到模态的初步划分结果;最终,利用小滑动窗口L,对稳定模态及过渡模态交接区域进行细划分,准确定位稳定模态与过渡模态的分割点.算法实现了多模态过程的全自动离线识别,并给出合理有效的识别结果.仿真分析表明此方法能够实现模态的自动识别,且识别结果准确.
|
| 关 键 词： | 模态识别 多模态过程 过渡模态 稳定模态 全自动 |
| 收稿时间： | 2015-03-04 |
A Fully Automatic Offline Mode Identification Method for Multi-mode Processes |
| |
| ZHANG Shu-Mei,WANG Fu-Li,TAN Shuai,WANG Shu.A Fully Automatic Offline Mode Identification Method for Multi-mode Processes[J].Acta Automatica Sinica,2016,42(1):60-80. |
| |
| Authors: | ZHANG Shu-Mei WANG Fu-Li TAN Shuai WANG Shu |
| |
| Affiliation: | 1.College of Information Science and Engineering, Northeastern University, Shenyang 1108192.State Key Laboratory of Synthetical Automation for Process Industries(Northeastern University), Shenyang 1108193.Key Laboratory of Advanced Control and Optimization for Chemical Processes of Ministry of Education(East China University of Science and Technology), Shanghai 200237 |
| |
| Abstract: | Multimode is a general characteristic of complex industrial processes. Different modes have different process characteristics, and different models should be established. Therefore, offline mode identification is one of the critical problems for multimode processes modelling. Presently, the commonly used clustering methods cannot realize offline mode identification of multimode processes automatically because human analysis and further processing are needed to gain the final identification result. A fully automatic offline mode identification method is proposed in the paper. First, the data is divided into a series of data segments by a cutting window with the designated width H. The improved K-means clustering method is used to assign the segments into different clusters. According to the clustering result, the missing stable modes are dealt with to obtain the preliminary mode identification results. Finally, the regions between the stable modes and transitional modes are further analyzed by a small moving window L to determine the accurate boundaries between different modes. The mode identification of multimode processes can be realized automatically by the method for a reasonable and effective identification result. Feasibility and practical value of the method are evaluated by case study. |
| |
| Keywords: | Mode identification multi-mode process transition mode stable mode fully automatic |
|
| 点击此处可从《自动化学报》浏览原始摘要信息 |
|
点击此处可从《自动化学报》下载全文 |
|
