基于知识管理的精馏塔智能化控制模型  

Intelligentized Control Model of Distillation Process Based on Knowledge Management

作　　者：吉旭[1] 朱立嘉[1] 范云峰[1] 赵丰[1] 唐建华[1] 

机构地区：[1]四川大学化工学院,四川成都610065

出　　处：《高校化学工程学报》2006年第4期628-633,共6页Journal of Chemical Engineering of Chinese Universities

基　　金：国家自然科学基金(20306017)。

摘　　要：为解决多变操作环境下精馏塔在线控制调优问题,分析了精馏塔传统的控制策略与缺陷,提出精馏塔基于知识管理的智能化控制框架。框架引入知识管理体系结构,解决了模型构建、表述、引用等更适应于工程化应用的信息化处理技术,采用单神经元自适应控制策略,确定不同条件下动态优化的馏出物组成,进而确定各控制点的控制参数。并研究精馏塔知识管理的模型表达方式,在面向对象建模方法基础上,采用面向智能体的建模方法,建立了客体(对象)类、主动实体(智能体)类和最优化模型类,通过知识推理将三个类关联形成精馏塔管控一体化的知识化智能平台,其中知识推理采用了基于人工神经网络的数据挖掘技术。实际应用证明该体系结构是可行的。In order to research the optimized controlling method for distillation process under the uncertain circumstance, general control strategies of distillation process were analyzed, and the intelligentized control frame based on knowledge management was proposed and illustrated. Knowledge management was introduced to figure out cyber-techniques of model abstraction, expression and application, which is suitable to the engineering circumstance. Odd nerve cell self-adapting strategy was used to determine on-linely the optimized controlling parameters considering material cost, operating cost and market condition. As an important technique to realize the strategy, the object class, agent class and optimized model class were brought forward by using agent oriented modeling method. Furthermore, an illation method was established to link the three kinds of class to construct the intelligent plot of distillation tower controlling. The plot was aimed to realize integration of management and process controllin~ and to ensure the distillation process to operate at its best profit. The knowledge illation was characterized by data mining based on neural networks. The frame was proved to be feasible.

关 键 词：知识管理 神经网络控制 数据挖掘 面向智能体 

分 类 号：TQ021.8[化学工程] TP13[自动化与计算机技术—控制理论与控制工程]