连续搅拌反应釜的自适应模糊辨识与预测控制  

Adaptive Fuzzy Identification and Predictive Control of Continuous Stirred Tank Reactor

作　　者：许娣 高钰凯 佃松宜[1] XU Di;GAO Yu-kai;DIAN Song-yi(School of Electrical Engineering,Sichuan University,Chengdu 610065,China)

机构地区：[1]四川大学电气工程学院,成都610065

出　　处：《科学技术与工程》2020年第20期8268-8275,共8页Science Technology and Engineering

摘　　要：化工领域为保证生产安全,对温度、压强、浓度等工艺指标有严格的要求。连续搅拌反应釜属于典型的化工设备,存在较强的非线性和时滞性,传统的建模与控制方法无法满足其精度要求。针对连续搅拌反应釜系统提出一种自适应模糊辨识与预测控制的方法。首先根据模糊划分C均值聚类算法得到模糊隶属度和初始聚类中心,在此基础上采用分层遗传算法进一步优化连续搅拌反应釜T-S模糊模型的参数。其次,采用自适应机制遗忘因子递推最小二乘法来估计T-S模糊模型的后件参数。最后,基于得到的T-S模糊模型,对连续搅拌反应釜进行自适应模糊广义预测控制,仿真结果验证了该算法的有效性。To ensure the safety of production in chemical industry, requirements on process indexes such as temperature, pressure, and concentration are strict. A continuous stirred tank reactor(CSTR) is a typical chemical equipment, and it has strong nonlinearity and time delay. However, the traditional modeling and control methods could not meet the accuracy requirements. Therefore, an adaptive fuzzy identification and predictive control method for CSTR was proposed. First, the fuzzy membership degree and initial clustering center were calculated based on the fuzzy partition C-means clustering algorithm, and then the antecedent parameters of T-S fuzzy model of CSTR were further optimized by combining hierarchical genetic algorithm. Secondly, the forgetting factor recursive least square method with self-adaptive mechanism was used to estimate the post parameters of T-S fuzzy model. Finally, based on the T-S fuzzy model, the adaptive fuzzy generalized predictive control was applied to the CSTR, and the effectiveness of the algorithm was verified against the simulation results.

关 键 词：连续搅拌反应釜(CSTR) 模糊辨识 T-S模糊模型 自适应模糊广义预测控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]