一种乙烯裂解炉的多工况监控方法  

A novel method for monitoring ethylene cracking furnace with multiple operation modes

作　　者：王晓阳[1] 王昕[2] 王振雷[1] 田亮[1,3] 

机构地区：[1]华东理工大学,化工过程先进控制和优化技术教育部重点实验室,上海200237 [2]上海交通大学,电工与电子技术中心,上海200240 [3]中国石化上海石油化工股份有限公司,上海200237

出　　处：《计算机与应用化学》2012年第9期1119-1122,共4页Computers and Applied Chemistry

基　　金：国家自然科学重点基金(61134007);国家自然科学基金项目(21276078,61174118);优秀青年基金项目(61222303);国家高技术研究发展计划(863)资助项目(2012AA040307);上海市重点学科建设项目(B504);流程工业综合自动化国家重点实验

摘　　要：随着进料负荷、产品组分等过程参数的改变,生产过程的工况也随之改变,而传统的基于多元统计过程监控方法都是假设过程处于单一工况下,因此,将传统方法应用到多工况过程时,往往不能获得很好的监控效果。在对复杂系统进行监控时,由于过程的非线性,传统的基于线性模型的监控方法由于忽略了系统的非线性特征,监控性能也大打折扣。本文针对工业过程中的多工况和非线性监控问题,提出了一种基于即时学习(Lazy Learning)和Greedy-SVDD的多工况过程监控方法。首先使用Lazy Learning对过程进行多模型局部建模,获得局部模型输出和过程真实输出的残差,通过对残差进行监控从而去除多工况的影响。然后用能够很好地处理非线性问题的支持向量数据描述(SVDD)方法对残差建立过程监控模型。为了解决SVDD方法用大样本建模时计算复杂度非常高的问题,本文用Greedy方法提取建模数据集的特征样本用于SVDD建模。最后将上述方法应用在TE模型和乙烯裂解炉的监控上,仿真结果证明了该方法的有效性。Traditional methods for process monitoring usually works under the assumption that the process only has one steady state mode and the relation between variables is linear, but when applied to the process with multiple operation modes and strong nonlinearity these methods don＇t have a good monitoring result. In order to solve these problems, a process monitoring method is proposed based on Lazy Learning and Greedy-SVDD. First, the multiple operation modes of process are eliminated by building multiple local models using Lazy Learning, and the residual error between real process output and model predicted output is got for monitoring. Then, the method called support vector data description （SVDD） is used to build nonlinear monitoring model. In order to solve the problem that the computation burden for SVDD modeling is very heavy due to the tremendous size of the modeling data set, Greedy method is used to extract the feature samples for SVDD modeling. In the end, the proposed method has been applied to the TE （Tennessee Eastman） process and the ethylene cracking furnace to show its efficiency.

关 键 词：多工况 即时学习 多模型局部建模 Greedy特征样本提取 支持向量数据描述 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]