考虑设备性能退变的热虹吸式再沸器软测量研究
发布时间：2018-10-07 17:24
【摘要】：热虹吸式再沸器是精馏塔的重要附属设备,为了给再沸器的热量控制提供一个较好的初值,本文采用神经网络和移动窗两者相结合的方法,建立热虹吸式再沸器换热量软测量模型,并对各软测量模型的影响因素进行如下研究:(1)对再沸器进行变量分析,在机理模型的基础上,确定易测可控的关键变量,从中选择五个变量作为数据模型的输入输出变量;(2)利用支持向量机建立了再沸器换热量与各影响因素之间的数据模型,考察不同影响因素对换热量软测量结果的影响;(3)利用BP神经网络对非线性模型的拟合优势,采用神经网络方法建立再沸器换热量软测量模型,考察不同影响因素对换热量软测量结果的影响;(4)针对生产过程的时变特性,以及软测量技术实施过程中对模型预测可信度的要求,结合移动窗方法改进了神经网络模型,使模型尽可能贴合再沸器运行状态。在保证预测精度的前提下,降低了模型更新频率,减少了计算;计算结果表明,采用神经网络加移动窗方法建立热虹吸式再沸器换热量软测量模型,可以为时变生产过程的在线检测提供经验与技术支持。
[Abstract]:Thermosyphon reboiler is an important auxiliary equipment of distillation column. In order to provide a better initial value for heat control of reboiler, the neural network and moving window are combined in this paper. The soft sensing model of heat transfer of thermosyphon reboiler is established, and the influencing factors of each soft-sensing model are studied as follows: (1) the variables of reboiler are analyzed, and the key variables which are easy to measure and controllable are determined on the basis of the mechanism model. Five variables are selected as input and output variables of the data model. (2) the data model between the heat transfer of reboiler and various factors is established by using support vector machine. The effects of different factors on the results of soft measurement of heat transfer are investigated. (3) using the advantage of BP neural network to fit the nonlinear model, the reboiler soft sensor model of heat exchange is established by using the neural network method. The effects of different factors on the results of soft measurement of heat exchange are investigated. (4) according to the time-varying characteristics of the production process and the requirement of the reliability of the model prediction in the implementation of soft sensing technology, the neural network model is improved with the moving window method. Make the model fit the reboiler running state as well as possible. On the premise of ensuring the prediction accuracy, the updating frequency of the model is reduced and the calculation is reduced. The calculation results show that the thermal siphon reboiler heat transfer soft sensing model is established by using the neural network and moving window method. It can provide experience and technical support for on-line detection of time-varying production process.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ051.65
本文编号：2255035
[Abstract]:Thermosyphon reboiler is an important auxiliary equipment of distillation column. In order to provide a better initial value for heat control of reboiler, the neural network and moving window are combined in this paper. The soft sensing model of heat transfer of thermosyphon reboiler is established, and the influencing factors of each soft-sensing model are studied as follows: (1) the variables of reboiler are analyzed, and the key variables which are easy to measure and controllable are determined on the basis of the mechanism model. Five variables are selected as input and output variables of the data model. (2) the data model between the heat transfer of reboiler and various factors is established by using support vector machine. The effects of different factors on the results of soft measurement of heat transfer are investigated. (3) using the advantage of BP neural network to fit the nonlinear model, the reboiler soft sensor model of heat exchange is established by using the neural network method. The effects of different factors on the results of soft measurement of heat exchange are investigated. (4) according to the time-varying characteristics of the production process and the requirement of the reliability of the model prediction in the implementation of soft sensing technology, the neural network model is improved with the moving window method. Make the model fit the reboiler running state as well as possible. On the premise of ensuring the prediction accuracy, the updating frequency of the model is reduced and the calculation is reduced. The calculation results show that the thermal siphon reboiler heat transfer soft sensing model is established by using the neural network and moving window method. It can provide experience and technical support for on-line detection of time-varying production process.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ051.65
【参考文献】
相关期刊论文 前10条
1 夏陆岳;朱群娣;王海宁;潘海天;;动态选择性集成神经网络软测量建模[J];计算机与应用化学;2016年02期
2 张朝环;;热虹吸式再沸器的压力平衡设计[J];化工进展;2015年S1期
3 高凌云;齐鸣斋;;分壁式精馏塔模拟及节能研究[J];现代化工;2015年07期
4 薛凌燕;贾保先;;化工行业智能监测系统的研究[J];自动化仪表;2014年12期
5 李晓东;张信;;再生塔热虹吸再沸器优化选型[J];化工设计;2014年03期
6 王春鹏;于佐军;孟凡强;;折息移动窗递推PLS算法及其在聚丙烯生产过程中的应用[J];化工学报;2013年12期
7 姜文革;;在线监测与故障诊断系统在煤化工设备管理中的应用[J];中国设备工程;2013年07期
8 朱群雄;甄玉山;;用于化工软测量的基于移动窗的过程神经网络[J];清华大学学报(自然科学版);2012年09期
9 廉磊;高前进;江燕斌;;丙酮-水精馏过程模拟和优化[J];化学工程;2012年08期
10 曹鹏彬;胡加强;;化工装备运行状态的无线传感器网络监测系统[J];武汉工程大学学报;2012年07期
相关硕士学位论文 前2条
1 冯可新;换热器参数化建模技术研究[D];哈尔滨工程大学;2013年
2 甄玉山;基于过程神经网络的化工过程软测量研究[D];北京化工大学;2012年
，本文编号：2255035
本文链接：https://www.wllwen.com/kejilunwen/huagong/2255035.html
