化工生产的数据调谐与稳态辨识

发布时间：2018-02-27 00:06

本文关键词： 数据校正稳态辨识鲁棒性 F-检验法　出处：《北京化工大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着社会的不断发展和进步,化工产品在人们的日常生活中扮演着越来越重要的作用。对于逐渐成为“世界工厂”的我国而言,化工行业的地位也越发变的无可比拟,因此,如何对化工生产过程实施操作优化以在不增加能耗和减少污染的前提下提高产率,是十分值得探究的课题。高精确性、高可信度的稳态数据对于实施操作优化有着决定性的影响。如果实施操作优化所依据的数据不具有精确性和可信度,那么操作优化可能会失败,甚至对整个化工生产过程造成不可逆转的伤害。数据校正技术与稳态检验技术的意义就在于能够为操作优化提供高精确性、高可信度的稳态数据。论文首先讨论了数据校正技术的基本理论,重点对数据校正技术中的重点和难点过失误差处理部分做了较为详致的研究。基于鲁棒估计的数据校正技术由于能够同步完成对两类误差的处理,论文中也做了重点的回顾,并根据鲁棒估计理论的基本原理,提出了两种具有鲁棒性的目标函数,通过实例,论证了两种目标函数在数据校正中的优效性。论文对目前较常用的两种基于滤波思想的稳态检验技术做了细致的回顾和研究,在对经典F-检验法的学习过程中,发现并指出了其所存在的两个基本问题,并以此为切入点提出了一种改进的F-检验法,给出了评估稳态检验效率的三种指标,通过仿真实例论证了改进的F-检验法的效率。CSTR由于其自身具有诸多优点,被广泛的应用在化工行业中,本篇论文中也以此为对象模型对研究成果进行了检验,通过将数据校正技术与稳态检验技术相结合,提取了具有高精确性、高可信度的稳态数据。这就为后续的系统建模、过程优化、参数估计等工作打下了扎实的基础。
[Abstract]:With the continuous development and progress of society, chemical products are playing an increasingly important role in people's daily life. For our country, which has gradually become the "world factory", the status of chemical industry has become more and more unparalleled, so, How to optimize the operation of chemical production process in order to increase production rate without increasing energy consumption and reducing pollution is a topic worth exploring. Reliable steady-state data have a decisive effect on the implementation of operational optimization. If the data on which the optimization is based is not accurate and reliable, the optimization may fail. The significance of data correction technology and steady-state inspection technology is that it can provide high accuracy for operation optimization. The paper first discusses the basic theory of data correction technology. The key and difficult fault error processing part of the data correction technology is studied in detail. Because the data correction technology based on robust estimation can process the two kinds of errors synchronously, the paper also makes a review of the key points. According to the basic principle of robust estimation theory, two objective functions with robustness are proposed. In this paper, two kinds of steady-state test techniques based on filtering thought are reviewed and studied in detail. In the course of studying the classical F- test method, the two objective functions are proved to be superior and effective in data correction. Two basic problems are found and pointed out, and an improved F- test method is proposed, and three indexes to evaluate the efficiency of steady-state test are given. The efficiency of the improved F- test method. CSTR is widely used in the chemical industry because of its many advantages. This paper also takes this as the object model to test the research results. The steady-state data with high accuracy and reliability are extracted by combining the data correction technique with the steady-state test technology, which lays a solid foundation for the subsequent work of system modeling, process optimization and parameter estimation.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TQ015.9

【参考文献】